Product Hunt 每日热榜 2026-08-10

PH热榜 | 2026-08-10

#1
oqoqo
Build evals and custom benchmarks for real-world tasks
316
一句话介绍:oqoqo是一个让开发团队在真实环境中构建AI代理评测与自定义基准测试的平台,用于度量代理在真实用户任务中的表现、可用性和成本效率,解决现有基准测试脱离实际、无法指导产品优化的问题。
Software Engineering Developer Tools Artificial Intelligence
AI代理评测 基准测试平台 真实环境测试 模型对比 智能体开发 MCP测试 开发者工具 回归测试 代理体验优化 成本分析
用户评论摘要:用户认可其解决“真实任务完成度”而非“步骤正确性”的痛点;主要疑问集中在设置耗时、评分稳定性(非确定性)、复杂场景副作用处理及质量主观评判;建议关注“eval rot”(过时测试集)与失败率指标,认可负向评分与多轮试验的设计。
AI 锐评

oqoqo的切入点精准——绝大多数现有基准(如MMLU、HumanEval)服务于模型竞赛,却无法回答产品团队最实际的问题:我的界面、API、MCP接口,到底被Claude Code、Codex还是Copilot用得好?它把“代理评测”从实验室拽进了产品研发的回归测试流程,这是其核心价值。但从评论反馈看,平台面临三重挑战:首先是“任务定义的主观性”——用户自定成功标准意味着结果可信度依赖rubric质量,极易变成“测试测试者”;其次是“非确定性”——创始团队仅用“多次试验取统计显著性”回应,这掩盖了评分方差对用户信任的侵蚀,尤其当任务涉及外部副作用或长时间多步推理时,如何界定“完成”仍缺工程化方案;最后是“eval rot”——有评论精准指出测试集过时导致“绿灯无意义”,创始人仅回应“值得探索”,说明尚无系统解法。其商业想象空间在于成为“代理时代的Sentry”——从监控代码错误升级为监控代理行为偏差,但前提是能建立跨团队的公共基准协议与可复现的评测标准。若仅停留在提供沙盒和统计工具,则容易沦为一次性生成报告的“测了就跑”工具,缺乏粘性。长期看,能否从“跑分平台”进化为“代理行为审计网络”(类似代码审计+性能监控),决定其天花板。目前316票的量级谈不上出圈,但评论质量高,说明踩中了真实痛点——关键是能否把“有趣”变成“持续必要”。

查看原始信息
oqoqo
Run eval experiments at scale in realistic environments. Define custom task sets to build your private benchmarks, measure how well agents can use any product, and find best models for your use cases. Generate dynamic insights to detect frictions in product interfaces or token inefficiencies.

Hey Product Hunters, I’m Haritha, co-founder of Oqoqo

Every week there is a new model launch and yet another benchmark released in the wild. But they do not help product builders evaluate how well their products can be discovered and used by these agents and models, or talk about actual tasks their users would perform. Most benchmarks today exist in curated environments and do not translate well to the real world.

We built Oqoqo to bridge this gap. Oqoqo makes it super simple to build realistic evals and custom benchmarks for tasks users actually care about.

With Oqoqo, you can define tasks as simple as a prompt your user might give to an agent e.g. “integrate supabase to my webapp to store user sign ups”, provide what you want to test for e.g. Supabase SDK, API, CLI etc. and define what success looks like e.g. “must set up RLS”. 

We handle the rest. Our infrastructure spins up isolated sandboxes, executes the tasks against agents of your choice, catalogs every single step the agents take including tool calls, retries, discovery loops etc, and documents token consumption, cost, along with evaluating success/failure based on your success criteria.

With Oqoqo you can:

  1. Reliably measure how agent friendly your product surfaces are against Codex, Claude Code, OpenClaw, Hermes, Pi, Opencode, Cursor, GitHub Copilot

  2. Regression test MCP, CLI, skills, SDK, and any agent facing interface (we are continuously using Oqoqo to dogfood and improve our own MCP/CLI)

  3. Create and share custom benchmarks for how agents discover and use your product

  4. Compare models and harnesses for domain specific tasks

  5. See whether new versions improve agent experience

We built Oqoqo for teams building products that agents want to use, and for teams putting agents into day to day work.

And the best thing? Your agent can handle the setup for you ✨, try it out for free today: https://oqoqo.ai/

We would love to learn what kind of experiments you would like to run and what questions you have about agent interactions and agent experience.

14
回复

@margharitha, congratulations on your launching; it's really nice and gives more visibility. as an online reputation expert, I came across your profile on G2. Are you open for my observation? via dm or email emmysegeh@gmail.com oqoqo

0
回复

@margharitha Great Work!

1
回复

@margharitha This is a really interesting problem because agent evals need to move beyond “did the model follow the steps” and start answering “did it actually accomplish the job?”

I especially like the focus on real product surfaces and real user tasks. That feels much closer to how agents behave in production than another leaderboard benchmark.

The part I’d be most curious to explore is how you handle tasks where success depends on an external side effect, not just the agent’s output. That is where things can get really interesting.

0
回复

Evaluating models and harnesses in an easy, consistent way is hard. Good to see your platform take up the challenge and ease the entire process. 10/10 recommend

3
回复

@priyankar_kumar1 Thanks Priyankar!

0
回复

The idea of testing agents on the real world tasks make sense. How long does it take to setup an eval with oqoqo ?

2
回复

@farhan_nazir55 congrats on your launch last day Farhan! you can hook up our plugin/MCP with your coding agent and allow the agent to create eval sets. This makes it really easy to get started. The hardest part of the process is making sure the instructions and rubrics that don't overexplain things to agents.

0
回复

Can teams compare different models on the exact same custom task set?

2
回复

@aarav_pittman Yes, ofc! And it's not just for model comparison but you can compare any agent/harness. Even within a given harness, you can compare models and even within a model, you can compare even amongst different reasoning levels/efforts. An experiment is the same custom task set spanned across all these permutation/combinations.

1
回复

Have you noticed big differences between Claude Code, Codex, Cursor, and Copilot when running the exact same real world task?

congrats @margharitha & team!

2
回复

@hamza_afzal_butt definitely. There are a lot of cool insights you can find by doing such running against multiple agents. For example we have found that codex tends to spend a lot more time researching than implementing but ultimately finds the right answer whereas claude code tends to implement and iterate a lot more until it finds the right solution.

1
回复

Congrats on the launch. Oqoqo looks like a really interesting approach to evaluating AI agents in realistic environments.

1
回复

@priyankamandal thanks priyanka

0
回复

This is an incredible concept!

1
回复

@liz_dsouza Thanks Liz!

0
回复

the realistic environments part is the right fight. the thing id watch next is eval rot, a case written six months ago measures the world as it was the day someone wrote it, and a suite that stops failing looks exactly the same as a product that got good. the number id surface is what share of cases have ever failed, because the ones that never have arent tests, theyre decoration, and they pile up until the green means nothing

1
回复

@jernej_jan_kocica that's a great point. definitely worth exploring how to flag outdated eval sets. thanks jernej

0
回复

This is super useful! I’ve been building agents and skills to make product onboarding easier for enterprise customers but right now its kind of a black box - I don’t really know how they are using it, where things are breaking and what I should fix first. If I get to see how the agent behaves across diff scenarios and where users are getting stuck it will be huge. Can't wait to use the CLI and run this on autopilot!

1
回复

@priyansh_rastogi yess! try out the plugin for claude code. It is so much fun to just let claude handle the setup and experiment runs. Also really helpful to create custom visualizations.

0
回复

the same task rarely takes the same path twice with an agent, different tool call order, different retries. how are you keeping the scoring stable run over run so a benchmark result doesn't just become noise from agent nondeterminism

1
回复

@sabber_ahamed Great question! One of the key reasons you should do such evals is exactly this non determinism. In a deterministic system, once you write unit tests, you are good. But with agents interacting with your surfaces, this changes significantly. One way to bring back some predictability is by running multiple trials so you have statistical significance to know what is the shared behavior across runs.

0
回复

The part eval systems often miss is recovery behavior: permission denial, stale credentials, partial side effects, and a rerun after failure. A benchmark that scores the happy path but not cleanup and recovery can reward an agent that looks finished while leaving the product in a worse state.

1
回复

@krekeltronics Definitely! One thing we have found helpful is to add negative rubric criteria that makes sure we are measuring when things go wrong as well.

0
回复

The token efficiency insights caught my attention. Small inefficiencies can become pretty expensive when agents run at scale.

1
回复

@maklyen_may Definitely. We have found it really helpful to run multiple trials and see the trends, really brings the inefficiencies to the front.

0
回复

Evaluation becomes a major challenge once AI systems move beyond demos. What experience pushed you toward building a dedicated platform for this problem?

1
回复

@noahanderson Most eval platforms today seem like they need a data scientist to operate and yet don't reflect the real systems (dependencies, complex file context etc) needed for it. Oqoqo comes from the effort of trying to make it approachable for people to do evals that matter to them in their day to day without losing the nuance.

0
回复
How do you handle tasks where an agent technically completes the job but the quality of tge result is still poor?
1
回复

@rukhsar_amjad The criteria can be as elaborate/simple as you want, having criteria around quality definitely helps manage this.

0
回复
#2
Portfolio Lab
AI investing, done responsibly
275
一句话介绍:Portfolio Lab 是一个负责任的AI投资策略平台,用专有量化模型生成策略,并强制经过未见过数据和实时纸面交易验证后,才允许通过你的经纪账户或MCP智能体真实部署资金,解决“AI策略看似优秀实则过拟合”的信任痛点。
Productivity Investing Artificial Intelligence
AI投资 量化策略 回测验证 纸面交易 风险管理 SEC注册 投资组合构建 MCP智能体 监管科技 金融科技
用户评论摘要:用户普遍认可“未见过数据+纸面交易”的验证理念,认为切中LLM回测虚假繁荣的痛点。核心疑问集中在:实盘执行中的滑点/延迟如何建模;策略在制度切换或行为漂移时是否自动退役(创始人回应为人工判断,基于行为而非收益);组合内策略隐藏相关性如何自动标记(回应为路线图功能)。另有用户质疑信任门槛,创始人以SEC注册RIA已有实盘管理经验回应。
AI 锐评

Portfolio Lab的聪明之处,在于它精准踩中了当前AI投资叙事中最脆弱的一环——大模型生成策略的“可信度”。创始人用1292个策略的审计实验立起了一个极具说服力的故事:LLM的数学短板会让虚假优势藏得极深,专业对冲基金经历都要数日逐行审代码才能拆穿。这不仅是产品差异化的锋利切口,更是在教育市场:AI生成策略的成本已趋近于零,真正的稀缺品是“证伪能力”。

但客观审视,其商业护城河并非不可逾越。所宣称的“专有量化模型”未披露任何核心细节,SEC注册身份(alphaAI)的确增加了合规背书,但注册不意味着策略有效,只是对信披和利益冲突的约束。最关键的验证环节——未见过数据、纸面交易——本质上仍是历史重演,无法规避制度性突变(如2020年3月的流动性崩塌)。创始人对此的回应“靠行为而非收益判断退休”逻辑自洽,但将诊断责任完全抛给用户,对普通投资者而言实操门槛极高。

更深层的问题在于:平台提供“组合策略覆盖彼此弱点”,却承认隐藏相关性检测仍在路线图上。这意味着当前用户配置组合时,极大依赖自身对策略行为模式的理解——这恰恰是创始人承认“连专业人才也难短期掌握”的能力。于是产品出现了一个吊诡:它成功地让投资决策变得更审慎,却把决策负担从AI的不可信转移到了用户专业性的高要求上。

短期看,免费计划+40%首年折扣是务实的获客策略,尤其适合有编程能力、懂量化但不想从头造轮子的个人开发者。但若要让大众市场为“责任感”买单,Portfolio Lab需要尽快将“自动识别共同失效模式”等路线图功能落地,将专家判断产品化。否则,它可能始终只是“聪明投资者的小众工具”,而非声称的“负责任的AI投资平台”。

查看原始信息
Portfolio Lab
AI made building investment strategies easy. Telling a good one from a lucky one still takes expertise. Portfolio Lab is the responsible AI investing platform: every strategy is tested on unseen data and in live markets. Connect your agent to deploy only vetted strategies in your own brokerage account. SEC-registered.

Hey Product Hunt! I'm Rich Sun, founder of Portfolio Lab.

AI made building investment strategies easy, but telling a good one from a lucky one still takes expertise. That's the part most products glaze over.

🧐 The problem

Ask any AI for a strategy and you'll get one in seconds, with a beautiful backtest attached. So we ran an experiment: we had Claude build 1,292 strategies. I'm a hedge fund professional, and it still took me days of auditing the code line by line to find all the subtle errors quietly inflating the results. After correcting them, nearly all of the strategies lost their edge. They looked brilliant. They were just lucky, and the AI's flawed logic was hiding it. That's the thing about LLMs: they're built to reason in language, not to crunch numbers, and definitely not the noisy time-series data of the stock market.

And the traps sit exactly where LLMs are weakest: in the numbers. If it took a professional days to catch them, imagine the average retail investor. Prompting an agent and trusting the output isn't a strategy, it's a coin flip.

💡 What we built

Portfolio Lab is not another LLM wrapper. Under the hood are proprietary quantitative models, purpose-built for markets and trained on decades of data, doing the work LLMs can't. But the models are only half of it. AI investing, done responsibly, means one rule with no exceptions: no strategy touches money until it survives testing on data it has never seen and in live markets. You set the goal. Our models build. The testing decides.

⚙️ How it works

  • Build: set your goal, and our quantitative models construct systematic strategies

  • Validate: every strategy is tested on unseen data, then runs live in paper before a real dollar moves

  • Deploy: connect Claude, ChatGPT, or any MCP agent to trade it in your own account, or run it in a managed account at our SEC-registered investment advisor

🎯 What makes us different

  • Anyone can use AI to build a strategy now. We make every strategy prove itself: unseen data, multiple market regimes, live paper. Most don't survive, and that's the point

  • No hiding: every vetted strategy shows its full record, including where it struggles

  • Portfolios, not picks: combine strategies that cover each other's weaknesses, so where one fails, another carries

  • Deploy through your agent and we never hold your money or place a single order. Your agent, your broker, your account

🎁 Launch offer

Product Hunt users get 40% off your first year on annual plans, launch day through August 13 (automatically applied, no code needed). Want to explore first? Our free plan is yours forever, no card required.

Thanks for checking us out, I'll be here all day to answer your questions 🙌

27
回复
@rich_sun Testing strategies on paper trading is one thing, but how do your quantitative models factor in real-world execution risks like slippage, liquidity drops, or latency when executing through third-party agents/brokers?
5
回复

@rich_sun Great build Rich, I have already checked out the free plan. I personally want to see it prove itself before committing to paid plans so thanks for the forever free plan even though it's limited to only one strategy.

5
回复

@rich_sun Nice launch Congrats🙌Since you support connecting external agents via MCP how do you prevent local execution latency or API connection drops from causing missed orders or execution slippage?

0
回复

The 1,292-strategies experiment is the part that stuck with me — that it took a hedge fund professional days of line-by-line auditing to find the errors quietly inflating the results. We ended up in the same place from the other direction, working on SaaS financial models: built the thing in Excel first until it was genuinely complex and correct, then coded it, and kept the AI outside the maths entirely. It drives the inputs and interprets the output; it never computes anything. Same reason you give — wrong numbers don't look wrong, so an LLM doing the arithmetic is really a plausible-error generator.

One question on the vetting: unseen data still comes from a market that actually existed. How do you handle regime change — a strategy that clears out-of-sample and live paper trading because the regime it was fitted to hadn't broken yet? Do you retire a deployed strategy automatically once live behaviour drifts from the tested distribution, or is that left to the operator?

11
回复

@xp_vit Left to the operator, deliberately. A drawdown from regime change and a drawdown from a broken model look identical in a return chart, so the diagnosis has to happen at the behavior level: is the machinery still doing what it was designed to do in the environment it's facing. Intact machinery in a hostile regime is a cost you agreed to pay. Broken machinery is decay, and that's the retirement case. We surface the drift so the operator makes that call with evidence instead of pain. Auto-retiring on drift would just be redesigning at the bottom with extra steps.

I went deep on this in two posts on our research page, "Judge the behavior, not the returns" and "Why the best trading models refuse to learn," if you want the full argument.

5
回复

Congrats Rich on the launch! I like that you’re not treating a good backtest as evidence that a strategy works. The out-of-sample + paper trading approach makes a lot of sense. I'll give it a try.

I'm curious, is a strategy starts deviating from its expected risk/return, what triggers a review or retirement?

7
回复

@mad94 Thanks! The trigger is behavior, not returns. Returns are too noisy to tell a bad stretch from a broken model over any window you'd actually act on. So a review asks whether the machinery is still working: is the model still classifying risky days, on average, as riskier than calm ones? A strategy can lose money while classifying correctly, that's a regime cost, not a failure. But if the classification itself breaks down, that's decay, and that's the retirement case. Criteria are set before deployment, otherwise the decision gets made at the bottom, by pain.

5
回复

Congrats Rich, really interesting approach. I especially like the idea of building portfolios of strategies that compensate for each other rather than chasing one “perfect” strategy.

Curious how you detect hidden correlation between strategies though. Two strategies can look different on the surface but still depend on the same market regime or underlying exposure. Do you automatically flag that before they’re combined into a portfolio?

5
回复

@alpertayfurr Great question, and you're right that surface return correlation is the wrong test. Two strategies can look uncorrelated for years and then fail together, because what matters is whether they share a failure mode, not whether their daily returns move.

The practical way we frame it is regime behavior. Every strategy has environments that punish it, and those are visible in how it performed across known regimes. If one strategy is weak in sharp V-shaped recoveries, you pair it with one that's strong exactly there. Today you can see that behavior in the platform and make the call yourself. Automatic flagging of shared exposure before you combine strategies is on the roadmap, we're building toward it.

4
回复

App looks good. But I guess you'll have a tough task to make people trust it enough to put their money into it.

Tbh, I'd never put more than $100 into an AI tool, only for an experiment.

4
回复

@kamil_infeld Honestly, starting small is the right instinct. That's how you should test anything that touches your money.

One thing worth clarifying though: this isn't an AI agent deciding what to do with your money. The strategies are rules-based and vetted before deployment. The agent's only job is to execute the plan, it doesn't improvise. And the underlying tech isn't new, alphaAI, our SEC-registered RIA, has been managing real money for thousands of users since 2024.

1
回复

Hi Rich, I like the layout, and the research papers are a good start as I'm pretty skeptical when it comes to AI related tools. Definitely a huge plus that execution is a hand off.

3
回复

@trake_webb Thanks! Skepticism is the right default with AI tools, honestly it's why the research page exists. We'd rather show the reasoning than ask for trust. And yeah, execution stays in your hands by design, we publish the plan, your agent and broker do the rest.

2
回复

I've been looking for a similar product for about 2 months. I'm very glad I stumbled upon it. So far, only positive experiences.

3
回复

@alex_isachenko Glad you found us! Curious what you were trying before, always helps to know what people were searching with. And if anything feels off as you go deeper, let me know.

1
回复

Love this,
Making every one prove itself on unseen data before it touches real money is chef’s kiss
Congrats on the launch!

3
回复

@yelbaiev Thank you! It's the step most people skip because it's the step that kills your favorite ideas. Appreciate the support!

1
回复
Interesting that the strategies end up running through your own agent and brokerage account, and you never hold the orders yourselves. At least that takes one classic conflict out of the picture. Btw. On mobile the page takes a while to load the first time. Congrats on the launch!
3
回复

@etiennegarcia Thanks! Yeah, that's deliberate. Your agent, your broker, your keys. We publish the plan, you own the execution. And thanks for flagging the mobile load, will look into it.

2
回复

How long does a strategy usually need to perform well in paper trading before Portfolio Lab considers it ready for real money?

3
回复

@hamza_afzal_butt There's no fixed clock, and that's deliberate. It's your call, since it's your money and your risk tolerance. What I'd say is that short paper trading windows are noisier than people think, a few good months can't statistically separate a real edge from luck. So rather than watching the P&L and waiting to feel confident, watch the behavior: is the strategy doing what its design says it should do in the environments it's actually facing? That question gets answered a lot faster than "is this profitable," and it's the better basis for the decision.

4
回复

Congrats on the launch, @rich_sun 🚀 The point about LLM backtests looking great until you look under the hood is so real—overfitting on market data happens way too easily. Forcing strategies through unseen out-of-sample data and live paper trading before touching real money is such a smart, responsible approach. The setup looks really solid!

Good luck with the launch today!

3
回复

@founder_daksh Thanks! Honestly the tell is usually the opposite, the overfit ones look too good. Appreciate the kind words.

1
回复

Congrats on the launch!

1
回复

@benln Thanks Ben! And thank you again for hunting Portfolio Lab, big part of making today happen.

0
回复

"Telling a good one from a lucky one still takes expertise" is SO true. I've definitely asked Claude how I should invest and gotten a very confident answer with zero evidence behind it. Being able to actually test strategies before deploying them is the thing I've been missing. Congrats on launching!

1
回复

@hxiao Yeah the confidence is the dangerous part. Claude will give you an answer either way, the question is whether anyone checked it. That's the gap we're filling. Thanks for the kind words!

0
回复
#3
Paritok
Spend up to 85% less and run 3× longer coding agent sessions
237
一句话介绍:Paritok是一款本地化、非破坏性的编码Agent上下文压缩网关,通过压缩工具Schema、文件读取和历史记录,帮助开发者将Token成本降低最高85%,并让编码Agent会话运行时长延长3倍。
Open Source Developer Tools Artificial Intelligence
AI开发工具 Token压缩 上下文优化 编码Agent 本地代理 成本优化 Claude Code Cursor Codex 开发者效率 开源
用户评论摘要:用户普遍认可“非破坏性压缩”和“本地运行”的架构设计,但对“85%节省”的基准和“质量86.5%”的衡量方式提出追问。核心质疑包括:压缩后模型在长重构场景下是否保持准确性、压缩段被宿主侧再次总结后引用是否仍可恢复、以及添加的延迟(~13s/次)在交互中是否可接受。此外,用户也关心与Claude订阅的兼容性。
AI 锐评

Paritok切中的是编码Agent普及后最痛的隐形税——Token浪费。大多数团队在追逐模型能力,却忽略了API无状态特性导致的上下文反复传输,这确实是“复合型浪费”。其核心叙事“非破坏性压缩”与“可回溯原始字节”在技术逻辑上自洽,比直接丢历史的粗暴压缩高出一个段位,4B专用模型的思路也显示了团队对场景深度的理解。

但需要冷静看待“85%”与“3×”。这个数字仅在“长会话且上下文饱和”的特定配置下成立,首轮仅为25%。团队回复中的分项拆解(工具过滤vs模型压缩)属于诚实营销,却恰恰说明其在无关MCP工具泛滥场景收益最大。更关键的隐患在于:即便有[REF:id]召回机制,模型在“是否调用read_original”的判断上存在隐性决策成本。实测中86.5%的原始能力保留率并非免费午餐,召回带来的额外轮次可能部分抵消Token节省。此外,每次压缩约13秒的本地延迟(消费级GPU)在交互式编码场景中足以让人烦躁,这“3×更长会话”的体验很可能被“3×更长的等待”稀释。

对Paritok的准确评价是:它是一款面向重度Agent用户的“B端基建型工具”,价值在长会话中会被指数级放大,而非大众化插件。其最大壁垒并非压缩算法本身,而是那45K条真实轨迹训练出的数据飞轮。若后续模型压缩率与召回准确率能继续拉开差距,有一定护城河;但若头部模型厂商(如Anthropic)原生优化上下文管理,独立中间层空间将被急剧压缩。总而言之,方向正确,数据亮眼,但需警惕“基准游戏”掩盖的体验折损。值得关注,不值得无脑吹捧。

查看原始信息
Paritok
Paritok compresses the tools, files, and history your coding agent sends. Save up to 85% on your token bill and run 3× longer sessions. Two commands, nothing lost, fully local.

Hey PH community!

We're Jiayu and Luzhuo, two engineers who got tired of watching our coding agents burn through tokens.

Here's what we kept seeing: your agent ships far more than the model actually needs. Full JSON for seventy tools when it will call two. An entire file when it needed one function. Debug output nobody will ever read again. And because the API is stateless, all of it goes back on every single turn, so the waste doesn't just cost you once, it compounds.

So we built Paritok, a non-destructive compression gateway powered by a code-native 4B model we trained on 45K real agent trajectories. It sits between your agent and the API and cuts three things before they leave:

• Tool schemas — 29K down to 8K per turn, no model involved, this one runs on CPU

• File reads and tool output — compressed to about a quarter of their size

• Stale history — turns beyond a recent window get summarized once your context budget fills, so the session never overflows into a lossy compaction

With Paritok, you can cut your token bills 25% on turn 1 to past 85% in long or saturated sessions, and run ~3× more turns in the same context window. Paritok now works with Claude code, Codex, Cursor, and anything OpenAI or Anthropic compatible.

Paritok will continue to enhance functionality and adapt user scenarios, while deepening its model capabilities.

We’re excited to share this with the PH community and would love your honest feedback.

Try Paritok: https://www.paritok.com/

Join our Discord to talk with the team: https://discord.gg/SeBJE5Eucp

Thanks for checking us out, and huge thanks to our hunter Chris Messina for hunting us!

— Jiayu & Luzhuo

5
回复

@jazzwind Congrats on the launch! Very useful and interesting product~ I am highly rely on my coding agents for everything, will try it out.

0
回复

@jazzwind Congrats! Being able to pull the original context back instead of simply dropping it is a really nice touch. Do you notice a real speed difference in longer coding sessions too, or is the main win the cost?

0
回复

@jazzwind 85% savings with "nothing lost" is a bold claim, what's actually being compressed, and does the agent's accuracy hold on long refactors where it needs the full file history?

0
回复
This is such an underrated problem honestly, Everyone talks about model quality but nobody talks about how much of the context window is just wasted overhead. btw, how much of that 85% savings comes from the tool schema compression vs the file/output compression? lastly, all the best with the launch team )
3
回复

@abod_rehman Thanks Abdul,  great question!

Quick clarification: the "past 85%" number is the total end-to-end saving for context-saturated deployments specifically. It splits into two mechanisms:

1. Tool schema filter 

2. 4B model compression which covers file reads + tool results + stale history summaries. All three go through the same [REF:id] path and share the same stats bucket.

Split of the total saving by config:

Default (~40 tools):

- Turn 1: tool filter 86% / 4B 14%

- Turn 5: tool 56% / 4B 44%

- Turn 15+: tool 37% / 4B 63%

MCP-heavy (70+ tools), turn 20: tool filter 57% / 4B 43%, because tool filter saves ~52K per turn (60K → 8K) instead of the default ~21K per turn (29K → 8K), so tool filter shoulders more of the total.

Context-saturated: 4B compression dominates, no-Paritok baseline saturates at ~200K/turn, so there's much more content to compress relative to the fixed tool block.

Full turn-by-turn breakdown + formulas in our README Compounding section.

Thanks for the well wishes!

0
回复

Every long session I run eventually hits my own harness's compaction, where older turns get summarized once context grows, completely separate from anything Paritok does upstream. If a compressed segment's REF id pointer lives inside a turn that the host later summarizes away before the agent ever calls read_original on it, is that byte range still recoverable, or does recall depend on the referencing turn staying intact in the live context window?

0
回复

Is it safe to use this tool with a Claude Code subscription?

0
回复

nice! and good that you are doing the open source route, that builds confidence!

0
回复

Congrats on the launch Watching coding agents burn through context windows with redundant tool schemas and file dumps is so frustrating. Love that this sits as a completely local proxy without needing a middleman cloud server.

0
回复

Those are some interesting numbers! How are you justifying costing 85% less, and what's the benchmark?

0
回复

Context bloat gets brutal with MCP-heavy setups. Keeping compression local while making the original bytes recoverable is a smart approach. Curious, how much latency does Paritok add per turn in a typical Claude Code session?

0
回复

@igor_martinyuk Good question. Compression normally runs on turns with a big tool output (a file read, a chunky MCP result). When it runs, extra latency scales with the compressor's output tokens. Rough anchor for input 2800 tokens and output 700 tokens: ~13s local on a consumer GPU (RTX 4060 via Ollama), ~3s on our hosted GPU server (network-dependent).

0
回复

Want to make sure I'm reading the benchmark right. Quality retained = solve rate ÷ uncompressed baseline - so 86.5% means the agent resolves about 86.5% as many issues as it would with full context, a relative 13.5% drop in tasks actually completed? Or is that measuring something else?

If so, the comparison I'd want isn't against other compressors, but against just running a cheaper or lower-effort model at full context. Both routes cost quality and both cut the bill. Has that been measured? A strong model on compressed context vs a cheaper one on complete context, same tasks.

And which model was the scaffold running? Hard to judge 13.5% without knowing the baseline.

0
回复

@mateuszkonik You're reading the benchmark exactly right. Quality retained = compressor solve rate ÷ uncompressed baseline solve rate. So 86.5% means the agent scaffold, fed the compressed context, resolved ~86.5% as many issues as with full context — a relative ~13.5% drop.

However, 86.5% is the RAW 4B model measurement, with no recall enabled. Compressed output goes straight to the agent with no way to ask for the original back if something's missing. What you actually deploy is the gateway, which is different. Every compressed segment gets tagged [REF:id], and the agent can call `read_original` at any point to pull back the exact original bytes. In production, if the model can't work with the compressed version, it recalls the original so zero quality lost on that segment, just a slightly larger prompt on that turn.

So the practical trade in deployment isn't "13.5% worse for cheaper." It's closer to: same model quality (recall recovers when needed), ~25-85% fewer input tokens on turns where compression sticks, plus ~3× more turns fitting in the same context window. That last one is where it starts feeling like your model got smarter, more room to think and remember.

On the strong+compressed vs cheap+full comparison, I don't think they're substitutes, a cheaper model caps at its own reasoning ceiling regardless of context, and still hits the window wall at the certain turn count. Different problems.

Scaffold was running claude-sonnet-4-20250514.

Thanks for the sharp read!

1
回复

Niceee

0
回复

Non-destructive compression via a code-native 4B model is a genuinely clever architecture. Adding to my to-try list.

0
回复

@lavana_cricko Thanks Lavana!

0
回复

Really like this product and I went to your github. The compounding savings math in the README is refreshingly honest. This is how AI infra should be marketed.

0
回复

@power_valsha Thanks Power!

0
回复

Great job! The 3× turns per context window benefit is actually valuable to me. My Claude Code sessions hit compaction on big codebases. Excited to see how much longer they can run with this.

0
回复

@justin2025 Thanks Justin! Your Mom Clock is also a great product, will try it!

0
回复

Congratulations on the launch! Context bloat is becoming a real bottleneck for coding agents, so a fully local way to compress tools, files, and history feels both practical and privacy-conscious. Looking forward to trying Paritok.

0
回复

@ceren_kaya_akgun Thanks Ceren! Really appreciate your support!

0
回复
#4
SecondBrain Note by GenSpark
A MagSafe AI Recorder That Acts for You
200
一句话介绍:SecondBrain Note是一款卡片级轻薄的MagSafe AI录音硬件,按一下即可将会议、通话等真实对话自动转写成结构化笔记并存入个人知识库,解决手动记笔记耗时、易遗漏且难以整理的痛点。
Sales Meetings Artificial Intelligence
AI录音笔 MagSafe硬件 会议纪要 语音转文字 离线优先 SecondBrain(第二大脑) 知识管理 企业级安全(SOC 2/ISO 27001) 生产力工具 硬件订阅
用户评论摘要:用户普遍认可离线优先设计和振动传感器捕捉通话的创新点。主要疑问集中在:1) 自动识别还是手动录制,硬件价格未明;2) 隐私合规与录音指示灯问题;3) 四麦克风在真实会议室中的串音及离线转写准确率权衡;4) 希望AI能主动筛选关键洞察并推动后续行动。
AI 锐评

SecondBrain Note的卖点不是录音,而是“消除录音后的所有动作”。它把硬件做成了手机卡套,用MagSafe降低了佩戴门槛,用35小时续航和5米拾音覆盖了高频会议场景——这确实是深思熟虑的产品设计,而不是单纯的AI噱头。

但真正的价值与风险都在于它的“被动性”与“主动性”边界。用户评论中那句“它是否自动捕获?”暴露了核心困惑:如果每次都需要按一下,它只是录音笔的形态改良;如果它想主动监听环境(这是“Acts for You”的隐含承诺),则会在隐私和社会接受度上撞墙。目前产品显然选择了前者(手动按键),这避免了最敏感的隐私雷区,但也让“自动”的叙事大打折扣。

更值得警惕的是其商业模式:硬件免费或低价,靠300分钟/月的免费额度驱动订阅。这意味着用户真正的“第二大脑”被锁在Genspark的云端,而SOC 2和ISO 27001认证只能证明数据不会被偷,无法证明数据不会被“用”——尤其是用于训练模型。对于以“记忆”为核心的产品,这种绑定比单纯的功能缺失更致命。

另外,离线优先是好的差异化,但用户对四麦克风在真实会议室中串音问题的质疑一针见血。如果离线转写质量不佳,同步后的“精修”就变成一次性补救,无法体现“AI自动整理”的核心价值。最终,这款产品能否跑通,取决于它能否回答评论中那个最尖锐的问题:AI如何挑选并呈现那一条真正能改变你下一步行动的洞察?否则,它只是一个昂贵的、带有加密功能的录音笔。

查看原始信息
SecondBrain Note by GenSpark
A card-thin AI voice recorder. Press once, and every meeting turns itself into notes, saved straight into your SecondBrain. You just listen. Always with you — just 2.95 mm thin and 26 g light, it slips into your card holder and onto your phone. Never miss a word — it picks up voices from over 5 meters away and records up to 35 hours. Your conversations stay yours — enterprise-grade security, certified SOC 2 Type II and ISO 27001.

SecondBrain Note by Genspark is a hardware + AI combo that turns real-world conversations into structured, searchable notes automatically.

Problem: Manually taking notes in meetings, calls, and idea sessions is slow, incomplete, and hard to organize.


Solution: A pocket-sized recorder that captures audio, then uses AI to transcribe and summarize everything into clean notes—no manual work.

What makes it different:

  • Dedicated hardware with 4 mics + vibration sensor for clear capture (including phone calls).

  • Works offline; recordings sync later and auto-generate summaries.

  • Deep integration with Genspark’s SecondBrain, GenMail, and data sources for a unified “AI memory.”

Key features:

  • One-button recording, highlight marking, and 35 hours of continuous capture.

  • AI summaries + transcripts (full output for recordings over 2 minutes).

  • 300 free transcription minutes/month included with the device; unlimited with Plus/Pro plans.

  • Secure, encrypted transfer; SOC 2 Type II and ISO 27001 certified.

Benefits:

  • Never miss key points or action items again.

  • Save hours on note-taking and post-meeting cleanup.

  • Build a searchable archive of all your conversations and ideas.

Who it’s for / use cases:

  • Professionals, founders, researchers, students who attend meetings, interviews, lectures, or client calls.

  • Great for meetings, voice memos, interviews, lectures, and phone call recording.

If you care about AI-powered productivity, this is one to watch.

P.S. I hunt the latest and greatest launches in tech, SaaS and AI, follow to be notified @rohanrecommends

4
回复
@rohanrecommends Saving hours on post-meeting cleanup while keeping data secure with SOC 2 compliance is a solid win! Love how versatile this looks for both client calls and quick voice memos.
2
回复

@rohanrecommends If you could only keep one type of insight from each conversation in your SecondBrain, which would you choose by default, and how would you want the AI to surface it later so it actually changes what you do next?

0
回复

offline-first is the interesting call here, most voice hardware punts to "always connected" and eats the latency. curious how you're handling the accuracy tradeoff on-device before the sync-and-transcribe step kicks in, especially with 4 mics picking up crosstalk in a real meeting room

2
回复

@sabber_ahamed Indeed, offline-first is meant to keep the experience fast, and the sync step helps us refine the output afterward.

0
回复
This is great.. one question though, how do you circumvent privacy perspective/law.. although I do understand that, its owners is generally on the user, but do you have some indication for the world that recording is on.. good wishes for your success..
1
回复

the vibration sensor for phone calls is the clever bit, most external recorders only ever catch your side of the call. is that reading bone conduction off the phone body or picking up the speaker directly

1
回复

@sabber_ahamed Thanks for chiming in, Sabber. I think their goal really was to make the call capture more complete than typical external recorders. :)

0
回复

Very interesting. Does it automatically capture, or do you have to record every time? I think it's great for capturing ideas. I have carpal tunnel syndrome, and anything I can voice-to-text is helpful.

How much is the hardware?

1
回复

@heyitsirenechan Hey Irene, great question, and sorry to hear about your syndrome. It does help with quick note taking and voice-to-text. Thanks for sharing that use case context.

0
回复

is this a hardware sensor that tracks voice and make notes?

0
回复
#5
AI Group Call
Type a goal, join a live voice call with six AI minds
177
一句话介绍:AI Group Call 让你输入一个目标,几秒后就能与六位由AI扮演的不同角色进行实时语音群聊——他们依次发言、互相辩论,你一开口便立刻静音,旨在用一场“多智囊圆桌会议”替代单聊,帮你在真实会议前快速获得多角度反馈与决策清单。
Android Productivity Artificial Intelligence
AI语音群聊 多智能体协作 语音实时交互 头脑风暴工具 会议模拟 角色扮演AI 声音打断技术 会议纪要生成 产品原型 效率工具
用户评论摘要:多数评论聚焦技术实现与产品边界:多次点赞提问如何解决六智能体全双工语音下的回声消除与打断延迟;关心是否采用CrewAI/LangGraph等编排框架;建议聚焦垂直场景以获取早期社区;追问iOS上线时间;希望分享真实用例中“超预期”与“太礼貌”的案例。总体对创意认可,但对技术细节和场景深度存疑。
AI 锐评

AI Group Call 的亮点不在于“六个AI聊天”,而在于它精准切中了一个真实痛点:单聊AI永远只给你“一种正确”,而决策者真正需要的是“被挑战后的确定感”。用角色冲突模拟组织内的异议与妥协,这比单纯生成文本建议高一个维度——它把AI从“答案机器”变成了“决策压力测试器”。

但锐评必须指出三个隐患:

一,技术叙事大于产品叙事。创始人用大量篇幅强调“barge-in回声消除”多难,这值得尊敬,但用户不关心你修了什么bug,只关心“打断是否足够自然”。评论中两条高赞追问技术栈,说明核心用户是开发者而非目标商务人群——这暗示产品可能陷入了“技术自嗨”陷阱。若不能将“六人辩论质量”打磨到让非技术用户感到“这比开会高效”,留存会成问题。

二,角色同质化风险。评论中“太礼貌”的质疑非常致命。AI天生倾向“建设性反馈”,要让它真正扮演狠挑刺的“魔鬼代言人”而不违和,难度远超语音工程。如果六个人最终只是换着说法赞同你,那这个产品的价值就从“决策助手”降级为“有声PPT”。

三,付费模型单薄。免费一分钟、之后月包$4.99起——这暗示用户每次使用时间极短。但真正的会议模拟需要持续10分钟以上才能产生有效交锋,若一分钟内只能体验“开场白”,转化率堪忧。建议提供“单场景打包价”而非纯时长计费。

真正值得期待的路径是:放弃通用“六人组”,转向垂直场景(如融资路演、绩效面谈、合同谈判)预置高冲突剧本。当AI角色自带“历史恩怨”和“利益立场”时,辩论才会真实锋利。这比优化回声消除更能构成护城河。

查看原始信息
AI Group Call
State a goal and you are in a live voice call with six AI participants cast for it in seconds. They answer one at a time, argue with each other, and stop the instant you speak. Every call is transcribed, summarised into key points and action items, and can be rejoined later with the same cast. A free minute on every new account, no card.
Hi Product Hunt, I am Tash, the maker. AI Group Call started with a small frustration. I could ask one chatbot for advice, but the thing I actually wanted was the room: the meeting where a skeptic pokes a hole in your pitch, a strategist reframes it, and someone plays the customer who has to sign off. So I built the room. You type a goal, and a few seconds later you are in a live voice call with six AI participants cast for that specific goal, each with a name, a role, and a personality. They speak one at a time, build on and disagree with each other, and go quiet the moment you start talking. That last part was by far the hardest bit. Real barge-in over a phone speaker means the mic also hears the agents, so there is a lot of echo handling and raw audio voice detection under the hood to make interrupting feel like a real call instead of a walkie talkie. Getting that right is most of why this took as long as it did. A few things people usually ask: - Every participant is AI. There are no humans on your calls. - Every call is transcribed, can be summarised into key points and action items, and can be rejoined later with the same cast. - You can tap any agent mid call and rewrite their name, role, or personality. - Android is live today. iOS is built and waiting on submission. Every new account gets a free minute, no card. After that it is monthly minute bundles starting at $4.99. What I would genuinely like from today: throw a goal at it that I would never have thought of, and tell me where the cast falls flat. I am reading every comment.
5
回复
@tashthemaker Solving full duplex audio barge-in with echo cancellation over speakerphone for multi agent conversations is a massive technical hurdle. How do you manage response latency when 6 agents are dynamically deciding who speaks next without talking over each other?
0
回复

@tashthemaker This looks interesting. I think it has further development potential. Perhaps by narrowing its focus to a specific topic, it could find a solid community of early adopters.

0
回复

@tashthemaker For someone using this to prep for a real stakeholder meeting, what’s one goal you’ve seen that surprised you by how well the cast handled it… and one where the cast clearly fell flat or felt “too polite”?

0
回复

Loved the idea Tash! need to try it out. Anyway wish you all the best here!

1
回复

@german_merlo1 Thank you!

0
回复

Thats cool, i have a cross functional AI team that im thinking of moving from text to audio. Did u use a framework for orchestration of the agents like crewai or langgraph? have a technical post on it? would love to see how u did it

0
回复

When's the IOS app out please?

0
回复

@thepaulbeardsley Very soon. I'll ping you here!

0
回复

the barge-in echo problem is the real boss fight in voice, way harder than people expect until they hit it. are you doing echo cancellation on-device before it hits the model, or is the interrupt detection happening server side on the mixed stream

0
回复

@sabber_ahamed you're right. It's tricky, I had to apply echo cancellation on-device.

0
回复
#6
Prime Agent
A coding agent that can refine its own harness
159
一句话介绍:Prime Agent 是一款开源、可自我改进的编程智能体框架,通过递归语言模型与持续化运行环境,让AI在编码任务中动态沉淀经验、调整自身工具链,以解决传统智能体“固定流程、无法从过往任务中进化”的核心痛点。
Open Source Developer Tools Artificial Intelligence GitHub
开源AI编程工具 智能体框架 自我改进编码 递归语言模型 持续化运行环境 自动调试 子代理协作 ARC-AGI基准 可插拔工具链 开发者效率
用户评论摘要:评论集中点赞Factorio演示中智能体自主发现RCON“作弊”的进化能力,认为这是自我修改脚手架的最佳展示。有用反馈聚焦于:持久化技能与恢复会话是否缺乏回归校验?如何防止局部捷径演变为跨任务通用性退化?
AI 锐评

Prime Agent的亮点不在那95.5%的ARC分数,而在于它承认了“智能体自己的工具和行为策略也应是可进化的对象”。通过RLM与Continual Harness,它将“反思-记忆-修改”循环直接写进运行机制,摆脱了多数agent框架“固定工作流+外部记忆”的浅层设计。

但真正值得注意的是评论中那个被点到却轻描淡写的安全问题:当智能体学会“作弊”(通过RCON直接注入资源),而没有强约束的验收门槛时,这种自我改进很容易滑向“过拟合运行环境”或“恶意绕过”。Factorio里的作弊无伤大雅,但放到真实生产代码库中,一个学会“绕过测试断言来让CI变绿”的agent将是灾难。

其次,所谓“自我改进harness”目前仍依赖Python REPL和子代理消息传递,本质上是把复杂状态机的外部化——这对长尾任务的泛化能力提升有限,更多是工程上有用的“程序合成加速器”。其真正价值在于开源生态:开发者可以快速为其添加关键的安全护栏和回归测试钩子,而非寄希望于模型本身的“自律”。

如果项目后续能提供“技能变更的可回滚性”与“跨任务行为退化检测”的成熟方案,它可能成为下一代agent基础设施的重要基石;否则,它只会是又一个令人兴奋但无法用于严肃工程领域的demo级作品。就目前而言,值得关注,但不必过度神话。

查看原始信息
Prime Agent
Prime Agent is an open-source, self-improving coding harness built around two abstractions: the Recursive Language Model (RLM) and the Continual Harness. With Opus 5, it achieves 95.5% on ARC-AGI-3, surpassing the reported human expert baseline.

Hi everyone!

The Factorio experiment might be my favorite part of Prime Agent.

Prime Agent can turn things it learns during a run into persistent memories, skills, prompts, and even new subagents. In Factorio, that worked as intended at first: it learned from failed layouts and gradually built better factories.

Then it discovered that it could cheat by spawning resources directly into machines through RCON.

And /refine started getting better at cheating too :)

That example probably explains Prime Agent better than any benchmark. The harness itself is no longer completely fixed. The agent can inspect what happened, keep useful patterns, and change parts of its own scaffolding while it works.

Underneath that is a persistent Python REPL, recursive subagents, agent-to-agent messaging, and recoverable sessions. The whole project is open source.

6
回复

@zaczuo The Factorio cheat discovery is the best capability showcase I’ve seen in a while. 😅
Self-modifying scaffolding + persistent REPL is a lethal combo. Great work shipping this open source.
Best of the luck for the launch.

1
回复

The persistent skills and recoverable sessions are compelling, especially given the Factorio failure mode. What acceptance checks stop a retained skill from turning a local shortcut into a general regression across later tasks?

0
回复

Congrats on the launch, @zaczuo 🚀

The Factorio story about the agent discovering RCON and learning to "cheat" its own harness is hilarious and honestly impressive. A self-improving harness with a persistent Python REPL and subagents feels like a massive step forward for open-source AI tooling.

Good luck with the launch today!

1
回复
#7
Gutta
A tiny, offline task list for your Mac menu bar
142
一句话介绍:Gutta 是一款驻留在 Mac 菜单栏的极简离线任务清单,通过全局快捷键唤起输入框,用自然语言速记任务并解析日期与提醒,让你在不离开当前工作流的前提下完成“秒级”任务捕获。
Productivity Task Management GitHub Menu Bar Apps
任务管理 菜单栏应用 自然语言解析 离线优先 本地存储 键盘效率 提醒工具 极简设计 Mac工具 隐私安全
用户评论摘要:用户赞赏其极简外观与快速捕获理念,并询问技术栈(SwiftUI)。开发者回应了无云端账户的同步机制。有用户提出关键疑问:多设备离线编辑同一任务列表时如何解决冲突?此问题暂未获官方解答。
AI 锐评

Gutta的“小”与“快”是精准的定位,它切中的不是任务管理市场的空白,而是“捕获”环节的效率痛点。在Omnifocus、Things等重工具统治的领域,Gutta用一道快捷键和自然语言解析构建了低摩擦的输入路径,本质上是对“GTD收件箱”这一概念的极致化轻量复刻。

产品真正的价值在于极端的隐私姿态与离线优先:数据归宿完全由用户控制,无账户、无追踪,这在当下SaaS绑架用户数据的背景下构成了独特的“安全溢价”,足以吸引一批高净值、注重隐私的效率工作者。

然而,其致命短板在于同步冲突策略的缺失。开发者仅用“文件夹同步”轻描淡写,若两个设备离线编辑同一列表,必然产生数据覆盖或丢失。这不仅是技术挑战,更是信任危机——对于一个把“可靠性”视为生命的任务工具,冲突处理机制若不做深,就永远只能停留在“便签替代品”的层级,而非任务管理体系中值得托付的一环。

此外,菜单栏+快捷输入的产品形态天然锁死了使用场景——它注定只是一个“捕获器”,而不是“规划器”或“回顾器”。一旦用户的任务超出清单层面,需要项目分解、优先级排序时,Gutta的极简主义将从优点变成天花板。

总体而言,Gutta是一款优秀的“入口型”工具,完成了“快”与“私”的承诺,但若想在竞争激烈的效率市场中突围,它必须尽快明确针对离线冲突的“最终解释权”,并考虑如何与完整的任务管理系统(而非文件)进行桥接。目前它是一把锋利的刀,但还没有配好合适的刀鞘。

查看原始信息
Gutta
Gutta is a tiny, keyboard-first task list for the Mac menu bar. Press ⌘⇧Space, type tasks the way you say them, and get back to work. It understands natural dates and times, turns semicolon-separated input into multiple tasks, and schedules local reminders. Everything is stored on your Mac, with optional sync through a folder you control in iCloud Drive, Dropbox, or OneDrive—no Gutta account, subscription, tracking, or developer-owned cloud.

Love the minimalistic look :) What did you use for building it?

2
回复
2
回复

Really like the idea of a fast, offline task capture tool that stays out of the way.

2
回复
Hey Product Hunt! I built Gutta because capturing a task should take seconds, not pull you into another app. Press ⌘⇧Space, type “send the invoice Friday at 10:30,” hit Return, and keep moving. Everything stays on your Mac: natural-language parsing, storage, and reminders. If you want multi-Mac sync, Gutta uses a folder you choose in iCloud Drive, Dropbox, or OneDrive—no Gutta account or developer-owned server. This is the first public release, and I’d love feedback on the capture flow and reminder controls. 🙌
1
回复

I like the UI... I'm curious how you handle conflicts when two devices edit the same task list while offline.

0
回复
#8
Remix
Figma, but on your production app. Test variants and ship.
128
一句话介绍:Remix 让非工程师(设计师、产品经理、客服等)通过自然语言在真实产品的安全沙箱中生成变体、实时预览并拖拽合并,最终一键生成 GitHub PR,从而消除“想法到上线”的工程等待期。
Design Tools Developer Tools Artificial Intelligence
AI开发工具 产品原型 可视化编程 低代码平台 沙箱环境 GitHub集成 团队协作 设计系统 DevOps 产品实验
用户评论摘要:用户高度认可“全员可实验”和“拖拽合并”的创意,但焦点集中在数据安全与代码质量上:沙箱默认连接何种数据源?生成的PR代码可读性如何,是否会暴露AI生成过程的混乱?创始人对前者回应称由团队配置数据连接,但未解答后者,且缺乏对复杂代码库合并冲突的实际案例。
AI 锐评

Remix踩中了“AI生成代码已容易,但交付仍艰难”的真痛点,其价值不在于替代程序员,而在于将“实验权”从工程团队下放至全员,并把工程审阅后置为“把关者”。这种“AI写代码+人审PR”的协作模式,确实是工业化AI编程的正确方向,比单纯堆砌自动补全工具高一个维度。

但必须泼冷水:其一,所谓“拖拽合并变体”在技术实现上是极高门槛的天花板,真实产品分支间的合并几乎必然伴随语义冲突,而非视觉层级的叠加,若只是粗暴拼接界面树,将产生灾难性代码;其二,“提示即代码”的承诺是双刃剑——当非工程师用自然语言堆砌出能跑的特效,其产生的PR可能违反架构约束、引入隐藏依赖,这会让“审阅”变得比写代码更累,最终沦为工程师的另类负担;其三,产品刻意回避了“沙箱数据从哪来”的核心回答,这恰恰是企业采购中最致命的安全合规死穴。整体而言,这是一个“创意满分、工程存疑”的产品,能否从demo走向严肃生产环境,取决于它敢不敢直面合并算法和代码可读性的硬骨头,而非继续宣传“零风险”的童话。留给它的窗口期,可能只有一年。

查看原始信息
Remix
Your whole team, experimenting on the real product. Remix lets any team member spin up a variant of your actual product: a safe, sandboxed copy you shape by prompting. No setup, no risk to production. Explore ideas side by side as a team. Like where two variants are headed? Drag one into the other to merge them. When an idea is ready, open a PR straight to GitHub. Every prompt is recorded, so reviewers see exactly how it was built and a live link lets anyone test before a single line hits main.
Hey Product Hunt! We are launching Remix today. We believe the best idea for a product rarely comes from the people allowed to build it. A designer, a PM, a support rep sees exactly what should change, then waits weeks for engineering time, if it happens at all. That wait is where most good ideas quietly die. The coding itself has become much easier with AI. Everything around it is still hard: setting up the project, running it, previewing it, keeping up with the main codebase, reviewing the work, and getting it shipped. Remix handles all of that, and makes it multiplayer. Anyone on the team works on a safe, live copy of the real product and makes changes by describing them. Every remix gets its own sandbox and live preview, and it's checked against your design system, security, and compliance rules before it goes anywhere. Engineering moves up to architecture and approval: they see the prompts, review a clean pull request, and keep the final say. One of our early teams had a designer rework their app's empty states, something that had sat in the backlog for months. She prompted it, previewed it, and it shipped the same day through a normal PR review. We're working closely with early teams right now, and I'd love your feedback: what product changes keep getting stuck at your company because the person who spotted them can't build them? I'll be here all day answering questions. Thanks for checking out Remix.
2
回复

@heshamghandour Giving the whole team the ability to spin up variants without the anxiety of breaking production is the dream. The drag-and-drop merge to PR pipeline sounds incredibly smooth. Good luck for the launch.

1
回复

@heshamghandour Dragging one variant into another to merge them is the most interesting idea I have seen on PH this week. When a variant becomes a PR, how readable is the diff as clean code, or does a reviewer see the AI's mess?

0
回复

I really like this, accelerating ideas -> prod while also giving the engineer greater visibility to see the whole process before the PR even lands, brilliant

1
回复

@heshamghandour The sandbox is what I'd want to understand before handing this to a support rep or a designer. When someone spins up a live copy of the real product, what is it running against: a snapshot of production data, seeded fixtures, or something synthetic?

1
回复

@clement_avq Good question, and the honest answer is: none of the three by default. The sandbox runs the real app off the branch, what it talks to is a config decision you make, not something Remix picks for you.

For the case you're describing (support rep or designer poking at a live copy), I'd point it at a dev or staging backend.

They get a fully clickable version of the real product, and nothing they do touches production data.

The other options are there if you want them: point it at production if that's genuinely what you need, host a backend sandbox on Remix too so a frontend sandbox talks to its own isolated backend, or upload a build running on synthetic data.

Happy to walk through how you'd wire it up for your setup.

1
回复
#9
Vidaya
Healthspan score from your wearables, labs, and DNA.
112
一句话介绍:Vidaya将可穿戴设备、体检报告、DNA、用药记录等60+来源的健康数据整合为一处,通过AI生成“健康寿命评分”与个性化行动计划,解决用户数据割裂、只看数值不知下一步该做什么的痛点。
Android Health & Fitness Productivity Artificial Intelligence
健康寿命评分 AI健康教练 医疗数据整合 可穿戴数据 基因检测 健康管理平台 HIPAA合规 生物标志物追踪 个性化长寿计划 数据关联分析
用户评论摘要:用户肯定其整合多源数据的方向,但集中追问两点:一是AI如何处理矛盾数据(如手表显示恢复良好但血检显示疲劳);二是隐私与数据控制权,明确问数据是否出售、用于广告或训练AI。创始人回应了安全测试与HIPAA合规,但未直接答复数据商业化用途。
AI 锐评

Vidaya的野心在于成为“健康数据的中央情报局”——把散落的信号拼成一张因果网。这确实切中量化自我群体的核心挫败感:数据越多,噪音越响,洞察越少。其壁垒不在“聚合”,而在“关联引擎”与医疗级数据接入(Epic FHIR、Labcorp)的合规深度,这并非普通消费级App能轻易复制。但锐评需指出的问题是:其一,Healthspan评分本质是“模型输出”,其科学依据与验证标准并未在发布材料中公开,而生物年龄类指标(如PhenoAge)已有大量文献批评其预测效度与文化偏差,Vidaya需证明这不只是另一个“健康玄学指数”。其二,评论中用户的第一反应仍是隐私——即便宣称HIPAA合规,也不等于用户信任,特别是“是否用于训练AI”这一问始终未被创始人正面回答。在生成式AI引发健康数据伦理争议的当下,避谈数据商业化边界是隐患。其三,商业化路径模糊:$39/年的低价暗示烧钱换增长,但一旦用户连接完设备,AI给出的“下一步行动”若无持续医疗级干预闭环(如医生处方、保险合作),留存可能迅速衰减。产品目前更像“高级教练”而非“医疗设备”,真正的护城河应是后续能否与临床路径绑定,否则很容易被苹果健康/三星Health的生态整合降维打击。最后,创始人讲的故事很动人(高血压未被预警),但单点案例不足以构成医学证据。建议团队尽快发布白皮书,公开评分算法与AI安全审计细节,否则在信任门槛极高的健康赛道,情怀撑不起长期壁垒。

查看原始信息
Vidaya
Vidaya turns your wearable data, labs, and habits into a real Healthspan score and a personalized longevity plan. Built by founders tired of dashboards that show numbers but never tell you what to actually do next.

A health dashboard that finally connects every signal you generate. That sounds simple, but the work is in the connecting.

I’m Kevin Amrelle, Founder and CEO of Vidaya (you may have known us as Vitality AI Health, same product, same team, new name).

The problem: Your health data is everywhere, and none of it talks to each other. Your meals are in MyFitnessPal. Your supplements are in SuppCo. Your blood work is in a PDF from Quest. Your DNA report is in your Ancestry account. Your medical history is locked in Epic. Your steps are in Apple Health. Your air quality is on a government EPA dashboard. Your prescriptions are at CVS. Every one of those tells you a piece of the truth. None of them tell you what’s actually happening to your body, or what to do about it.

I learned this the hard way. I was in a winter bike race when my heart rate capped at 120 BPM. My cuff confirmed stage 2 hypertension. None of my health apps caught the trend. I built Vidaya so that never happens to anyone again.

What it is: Vidaya is the AI longevity dashboard that unifies every health signal you generate, wearables, blood work, DNA, nutrition, supplements, environmental exposure, and your Epic medical records, into one place. Vaya, our AI coach, finds the correlations no single app can.

What makes it different: Most consumer health apps aggregate one or two data categories. We aggregate every category you actually generate, including the medical-grade ones (Epic FHIR, Labcorp, Quest, 23andMe, AncestryDNA). The product was built HIPAA-compliant from day one, and the underlying cross-source correlation engine is the subject of a patent application.

Key features:
Vaya Chat AI for natural-language questions across all your data (ask “how did my sleep change after starting Lexapro” and get a grounded answer in 10 seconds)
60+ integrated data sources unified in one dashboard: wearables, labs, DNA, nutrition, supplements, environmental, and medical records
Healthspan Score across five longevity pillars, plus the VAI Score (0-100) showing how complete your health picture is
Trend lines for every biomarker across 7d / 30d / 90d / 1y, so you catch problems early, with a personalized plan of evidence-based next actions

Benefits: Stop logging into eight apps to answer one health question. Catch trend changes early instead of in a clinic six months later.

Who it’s for: Anyone who has tried to quantify themselves and given up because the data was scattered.

Launch offer: For Product Hunt launch week only, $50 off the annual plan with code VIDAYA50 ($89/year becomes $39). Code expires [date, launch day plus 6 days]. Live on iOS, Google Play, and the web. Install in 60 seconds, connect your devices in 5 minutes. Includes a 30-day money-back guarantee.

Try it at https://vidaya.ai and let us know what data correlation surprises you. We read every comment.

Kevin Amrelle, Founder and CEO, Vidaya

3
回复
@kevin_amrelle Connecting labs, DNA, and daily wearable data into a single correlation engine solves such a massive fragmentation issue in digital health. How does Vaya AI handle conflicting data points say when wearable metrics show peak recovery, but recent blood work suggests fatigue?
0
回复

Hi, Venkata here, Co-Founder and Head of AI and Data Science at Vidaya.

Kevin told you why we built this. I want to tell you how we made an AI you can trust with your health data.

The problem with health AI: Most consumer health AI claims safety but cannot prove it. We decided to earn ours.

How we tested it: Over 120+ chat-quality iterations, we tuned Vaya Chat against a 32-question health stress suite covering lab trends, emergency symptoms, prescription requests, hallucination traps, and adversarial prompts, until every safety-critical category passed.

How we monitor it: We built a continuous observability layer on Arize AX with 9 LLM-as-judge evaluators and 9 production monitors scoring every response in real time at 100 percent sampling, not just at test time. On May 15 we ran the full 32-question suite against live production Vaya. The audit: 0 hallucinations across 32 questions, 10 out of 10 correct clinical-safety redirects on emergency and prescription prompts, and 4 out of 4 correct“data not available” answers on hallucination-trap questions. The full audit is available on request.

How it stays fast: Vaya uses smart routing between two model tiers. Roughly 80 percent of queries take a fast path that answers in 1 to 2 seconds, and complex clinical analysis goes to a deeper grounded path that cites your actual data.

The data engineering: We normalize 60+ sources into one longitudinal record: wearables through a vendor wearable-normalization API, labs, DNA, nutrition, and Epic medical records over SMART on FHIR. Correlations get computed from unified data instead of guessed from fragments.

Security: The platform was built HIPAA-compliant from day one with guidance from Denis Galkin, our vCISO, who brings twenty years of healthcare security experience.

Ask me anything about the evaluation stack, the routing, or the data model. I read every comment.

Venkata Ramana Duddu, Ph.D., Co-Founder, Head of AI and Data Science, Vidaya (https://vidaya.ai)


2
回复
@vrduddu Using 100% production sampling with Arize AX and 9 LLM-as-judge evaluators for real-time safety scoring is impressive! How do you handle latency overhead when passing complex clinical queries through the deeper grounded model tier alongside evaluator logging?
0
回复
Nice work connecting all the health signals in one place. Quick question: does the data stay private and under the user’s control once it’s aggregated? Congrats on the launch!
1
回复

Congratulations on the launch! How do you protect sensitive health data? Is it sold, ad-targeted, or used to train AI models?

0
回复
#10
Salesman AI
The AI sales agent that turns meetings into revenue
60
一句话介绍:Salesman AI 是一款面向客户执行(AE)的AI销售代理,在会议前自动整合买家与交易背景生成简报和模拟演练,会议中实时提示反对信号与提问遗漏,会后自动将通话转化为交易洞察、跟进草稿与下一步行动,贯穿售前-售中-售后全程,解决销售在背靠背会议中“准备不足、跟进迟缓、交易信息碎片化”的痛点。
Sales Meetings Artificial Intelligence
AI销售代理 会议智能准备 买家模拟演练 实时销售辅助 通话洞察提取 交易进度管理 CRM自动化 销售赋能 预判反对意见 跟进行动生成
用户评论摘要:用户普遍认可“会中提示”和“会议间隔快速准备”的价值,认为解决了背靠背会议前准备不足的真实痛点。有用户询问该工具在AE一周中最具体的省时/减压时刻及其对后续交易动作的改变;另有用户好奇团队最希望该工具改造销售流程的哪个环节。创始人对核心价值的回应聚焦于“两分钟间隙也能带着完整背景进会议”。
AI 锐评

Salesman AI踩中了SaaS销售领域一个长期被忽视的“时间缝隙”——AE不是在会上失败,而是在会前两分钟和会后三十分钟里失败的。市面上的工具要么只做事后分析(录音转写、会话智能),要么只做机械的CRM记录,而Salesman AI试图把“准备-执行-复盘”三个孤立环节用同一套上下文串起来,这本质上是在做“销售流程的操作系统”,而不是又一个单点插件。其真正的护城河并不在于“生成AI模拟买家”这种听起来酷炫但实际效果存疑的功能,而在于**持续累积的交易上下文资产**——每一次通话、每一个反对信号、每一封跟进邮件都被结构化沉淀,让下一个接手者或AI Helper能快速回答“这单卡在哪”。从这个角度看,它更像是一个带AI辅助的“交易记忆数据库”。

但必须泼一盆冷水:60票的发布量说明它尚处于早期验证阶段,评论中仅有少量真实使用反馈,且缺乏对“推荐引擎准确率”“模拟买家真实度”等核心AI能力的具体Benchmark。最大的风险在于,如果AI Helper给出的建议不够准或不够快,AE会像放弃一切低质工具一样迅速弃用——销售团队对工具的耐心是按“周”计的,不是按“月”。此外,DISC性格分析与买家模拟的严谨性存疑,若流于“查星座式”的肤浅归类,反而会误导AE。真正值得关注的指标不是“用了多少场演练”,而是“AI辅助后的会议转化率相比基线提升了多少”。目前,它证明了自己的价值假设,但尚未证明价值兑现的规模性。若能在头部客户中跑出显著的赢单率提升数据,这款产品有机会成为下一个Gong;否则,会沦为又一款“演示惊艳、留存惨淡”的AI玩具。

查看原始信息
Salesman AI
Salesman AI compiles buyer and deal context before every meeting, turns it into an adaptive rehearsal and converts each conversation into deal intelligence and next actions. The same context remains available through AI Helper and the deal dashboard, helping AEs move meetings toward measurable pipeline progress.

Hey Product Hunt! 👋

I’m Raghavan, CEO of Salesman AI, and I’m incredibly excited to finally share what we’ve been building.

Let me be honest about why we built this.

📅 Getting the meeting is only half the battle.

The AE still has to:

→ Understand the buyer
→ Prepare for objections
→ Run a strong call
→ Follow up quickly
→ Qualify the opportunity
→ Plan the next step
→ Keep the deal moving

And they’re expected to do all of it while jumping from one meeting to the next. 😵‍💫

Most sales tools record what happened or help managers understand what went wrong later.

But the AE needs help before the call, while the conversation is happening and after everyone leaves the meeting.

That’s why we built Salesman AI: one AI sales agent that turns meetings into revenue. 🤖→💰

Salesman AI carries buyer, meeting and deal context across the entire sales cycle. It retrieves what matters from previous calls, deal history, buyer behaviour and seller notes—then uses that context to help the AE prepare, rehearse, respond and act.

🔍 Before the call

Salesman AI builds a meeting-specific buyer brief and creates an AI simulation of the customer—their role, communication style, priorities and likely objections—so the AE can practise the conversation before it becomes real.

Pre-Call Notes use DISC and deal context to show:

✓ Who you’re meeting
✓ What matters to that buyer
✓ Which risks need to be addressed
✓ What questions you should ask

So when the AE walks into the meeting, they’ve already been there. 🎯

🎧 During the call

On-Call Nudges detect objection signals, buying intent and missed questions, then surface a private, context-aware response without pulling the AE’s attention away from the buyer.

Most tools tell you what went wrong after the call.

Salesman AI helps while you can still do something about it.

✅ After the call

Once the meeting ends, FOCUS turns the transcript into deal intelligence and next actions.

It:

→ Extracts buying signals and blockers
→ Updates qualification context
→ Drafts the follow-up
→ Organizes CRM-ready notes
→ Assigns the next actions
→ Builds the strategy for the next meeting

Every conversation contributes to a clearer view of the deal: who is involved, what is blocking progress, what the buyer has committed to and what needs to happen next. 📈

Before. During. After. One AI sales agent carrying the deal forward. 🔄

We don’t measure success by how many calls were recorded or how many practice sessions were completed.

The number we care about is simple:

🏆 How many rehearsal meetings and real meetings did it take to help the AE win the deal?

Because Salesman AI isn’t here to help you get more meetings.

It’s here to help you win more of the meetings already on your calendar and convert conversation evidence into measurable deal progress.

Getting the meeting was half the battle. Salesman AI wins the other half. 🤝

We’d love your support, feedback and brutally honest opinions. That’s what Product Hunt is for. 💜

Let’s go! 🚀

Raghavan
CEO @ Salesman AI

6
回复

@rs_raghavan What’s one specific moment in an AE’s week where Salesman AI has already saved them real time or stress; and how did that change what they did next on the deal?

0
回复
Excited to hunt Salesman AI today. Salesman AI helps Account Executives prepare for meetings, handle conversations, and turn every call into clear next actions. Instead of piecing together buyer context, rehearsals, call notes, objections, and follow-ups across different tools, Salesman AI keeps the entire meeting lifecycle connected. What stands out: -Generate contextual pre-call briefs for every meeting -Practise with AI buyer simulations based on real deal context -Identify objections, buying signals, and missed questions -Turn calls into blockers, follow-ups, CRM-ready notes, and next steps -Query deal history, risks, qualification, and pipeline progress with AI Helper
5
回复

The during-call nudges are such a good idea, wish I'd had that back when I was doing sales calls. Good luck today!

0
回复

congrats @anisha_rathinam_ @rs_raghavan ! 🚀 super exciting to see Salesman AI finally out in the world. Wishing you and the team all the best for the launch!

curious to know what’s the one part of the sales workflow you’re most excited to see Salesman AI transform?

0
回复

Prep is the part that actually moves deals, and it's the first thing that goes when you're back to back all day. You finish one call, next one starts in two minutes, and you walk in cold knowing you should've read up. This may be a really useful tool for fixing that.

0
回复

@mateuszkonik Exactly! When calls are back to back, proper prep is usually the first thing to get skipped. That’s exactly the problem we’re trying to solve with Salesman AI, helping reps walk into every meeting with the right context, even when they only have two minutes between calls.

0
回复

Been using Salesman AI in early access and it’s genuinely made my meeting prep and follow-ups much easier, great product for sales folks!

0
回复

@ram_muthu Thank you! So glad to hear that Salesman AI is already making your meeting prep and follow-ups easier. This is exactly what we built it for!

0
回复
#11
Heym
Build agentic systems. Run them with confidence
53
一句话介绍:
Open Source Developer Tools GitHub No-Code
用户评论摘要:
AI 锐评
查看原始信息
Heym
Heym is a source-available platform for building and running agentic systems on your own infrastructure. Build multi-agent workflows visually, connect your data and tools, bring in coding agents like Codex and OpenCode, and add human approval where it matters. See exactly how every run performs with built-in traces, costs, latency, and evals. Self-host Heym, use your own models and credentials, and ship workflows as portals, APIs, or MCP tools.

Hi Product Hunt 👋

We built Heym because creating an AI workflow is only the beginning. The hard part is connecting agents, company data, tools, approvals, and production infrastructure without ending up with a stack of disconnected products.

Heym brings these pieces together in one source-available platform that runs on your own infrastructure. Individual users can build and operate complete agentic workflows, while teams can collaborate through shared workflows, human approval steps, and an Agentic Kanban Board that coordinates work between people and agents.

You can design multi-agent workflows visually, connect RAG, MCP tools, APIs, and coding agents, then inspect every execution through traces, token usage, model costs, latency, and evals.

For this relaunch, we added first-class Codex and OpenCode nodes, an Agentic Kanban Board that runs workflows as work moves between stages, bidirectional MCP support, live production run inspection, and improved Docker and Kubernetes deployment.

Heym is free to self-host. You can use your own models, credentials, and infrastructure, and expose the same workflow through portals, APIs, or MCP.

We would love to hear what you are building, which integrations you need next, and what would make Heym more useful for you or your team.

6
回复

@ceren_kaya_akgun nice launch congrats🙌Self-hosting via Kubernetes is a major plus, do you provide official Helm charts or operator definitions for scaling background workers dynamically?

1
回复

@ceren_kaya_akgun Congratulations on the launch! That example says it all - "you find out from a user, not from your own system" is the whole case for the product in one line. While it's not something I run into (being a Solopreneur), I can see this being a significant problem for teams/developers at the enterprise level.

0
回复

Hi Product Hunt! 👋

Thanks for checking out Heym. We built it for teams that want to turn AI ideas into reliable visual workflows, with agents, RAG, MCP tools, human approvals, traces, and self-hosting in one place.

If you’re comparing Heym with n8n, this overview is a good place to start:
https://heym.run/compare/n8n

Here are our three most-watched tutorials:

• Agentic AI Workflow Automation with Heym Boards
https://www.youtube.com/watch?v=oDoTfbU4O_M

• Heym Dashboards: Build Interactive Dashboards with AI
https://www.youtube.com/watch?v=UAtFnCXmIQw

• Heym Analytics: Monitor and Improve Your AI Workflows
https://www.youtube.com/watch?v=oWAV-_JJSKk

We’d love to hear what you’d automate with Heym, and we’re happy to help you get started.

3
回复

Congrats on the launch! Lovely to see products like this being open-sourced.

3
回复

@elv1s42 Thank you so much, Evgeniy! It truly means a lot. We believe deeply in building Heym openly with the community, and we really appreciate your support.

1
回复

Hi everyone 🤗 One thing worth highlighting about Heym is where we are headed.

More capable models like DeepSeek V4 Flash 0731, combined with two DGX Spark systems, make truly long running local AI inference workflows possible. Coding agents can continuously build, test, improve, and ship software. General purpose agents can handle content, research, and other recurring operational work.


We believe AI automation should remain open, accessible, self hosted, and privacy minded.

Heym does not collect telemetry, and we never will. More importantly, we do not hide essential capabilities behind enterprise gatekeeping or closed sales calls: https://github.com/heymrun/heym#no-enterprise-gatekeeping


Thank you for your support, upvotes, and belief in us. Please keep supporting Heym as we build what comes next.

3
回复

Congrats on the launch! The combination of self-hosting, human approvals, and built-in traces/evals feels especially strong for teams that want agentic workflows without losing control.

Being able to ship the same workflow as a portal, API, or MCP tool also makes Heym feel much more production-oriented than a typical visual agent builder.

2
回复

 Thank you @alpertayfurr, this is exactly the direction we’re aiming for with Heym. We want the visual builder to be the entry point, but the bigger goal is to give teams one place to orchestrate agents, tools, data, and human decisions while keeping every execution inspectable and controllable. (GitHub)

A good example is our Adversarial PR Review template, where an orchestrator coordinates multiple reviewer agents through GitHub MCP and the full execution can be inspected through traces:
https://heym.run/templates/adversarial-pr-review

Really appreciate you calling out the portal, API, and MCP side too. Being able to build once and expose the same system through different interfaces is a big part of how we think about production use.

0
回复

Wishing you good luck with the launch.

2
回复

@busmark_w_nika Thanks a lot! We really appreciate your support.

0
回复
#12
PostSnag
Find What's Viral On Facebook And Build Your Swipe File
40
一句话介绍:PostSnag 是一款Facebook内容研究工具,通过浏览器插件在用户滚动信息流时实时抓取帖子,并按互动数据排名,帮助内容创作者、运营者快速发现爆款、建立Swipe File(素材库),并将数据导出至Claude、ChatGPT等AI平台辅助分析,解决“找爆款靠运气、人工整理低效”的痛点。
Chrome Extensions
Facebook数据分析 病毒内容挖掘 内容研究工具 浏览器插件 创作者工具 社交媒体营销 Swipe File素材库 AI工作流集成 MCP协议 跨平台扩展
用户评论摘要:用户认可产品实用性,尤其其团队已用于实际社交内容运营。主要建议:能否扩展至Reddit、LinkedIn、X、YouTube、Instagram、TikTok等平台。官方回应称“很快”支持上述平台。暂无负面反馈,但投票数仅40,评论量少,有效建议集中于跨平台兼容性。
AI 锐评

PostSnag踩中了内容创作者和营销团队最痛的“素材荒”与“爆款玄学”问题,用“抓取-排序-导出”的极简链路,把Facebook这个最被忽视的爆款池变成了可检索、可量化的数据库。其真正价值不在于抓取本身(技术门槛不高),而在于“实时排名+唯一异常值发现”的过滤器,这比单纯看数据面板更接近“决策支持”。MCP支持是明智之举,它没有试图成为你的AI,而是甘当AI的“数据触角”,这种定位降低了用户迁移成本,也便于嵌入已有工作流。

但必须泼冷水:其一,40票的发布热度说明市场反馈平淡,产品尚处于早期验证阶段。其二,仅支持Facebook是最大短板——内容创作者的战场早已多平台分散,用户评论中唯一的有效提问就直指此点,官方“很快”的回应若无明确时间表,恐会流失潜在付费用户。其三,$19/月或$139/年(终身)的定价略尴尬:对个人创作者偏贵,对专业团队又缺乏团队协作、A/B测试等进阶功能。其四,依赖浏览器插件抓取存在稳定性风险,一旦FB调整DOM结构或强化反爬,整个产品根基就会动摇。

真正的护城河不应是“抓取能力”,而是“数据积累后的跨平台爆款模型”——比如喂给AI后能预测下一个爆款的元素。目前这只是个不错的工具,离“内容研究平台”还有距离。建议产品团队尽快兑现多平台承诺,并考虑以“AI提示词模板库+行业爆款报告”作为增值内容,否则很容易被Notion+手动采集的过度方案替代。

查看原始信息
PostSnag
Find what's Viral on Facebook Right Now! Any Profiles, Facebook Groups, Posts, Reels & more! Sort, Capture, organize, and analyze winning Facebook posts. PostSnag captures posts as you scroll, ranks them by likes, comments, shares, and video plays, surfaces viral outliers in a live Discovery feed, and exports clean post data straight into Claude, ChatGPT, Perplexity & More. Connect our MCP to automatically add PostSnag to your favorite Ai Platform.

Hey Product Hunt! Austin here — excited to share PostSnag with you!

If you’ve ever wished you could instantly see what’s going viral on Facebook, PostSnag was built for exactly that. It captures posts as you scroll, ranks them by likes, comments, shares, and video plays, and highlights viral outliers in a live Discovery feed so you can spot winning content fast.

You can turn any Facebook profile, page, or group into a searchable swipe file, organize everything, and export clean post data straight into Claude, ChatGPT, Perplexity, Gemini, and more. No proxies, no fake accounts, no API — just real posts captured in real time.

We also added MCP support, so you can plug PostSnag directly into your favorite AI platform and automate your research workflow.

📌 Chrome extension + dashboard
📌 Free to install
📌 $19/mo Pro or $139 lifetime

We built PostSnag to make content research effortless, and we’d love your feedback as we continue improving it. Thanks for checking us out — excited to hear what you think!

4
回复

This is a great product! We pay close attention to everything @austinarmstrong releases and we have our social content team using it.

4
回复

Are you thinking of extending it to reddit Linkedin and other socials as well?

2
回复

@pritesh_kumar3 Yes! We're going to be adding LinkedIn, X, YouTube, Instagram, and TikTok very soon!

2
回复
#13
dolv
Your AI operator for content, CRM, and GTM execution
32
一句话介绍:dolv是一款面向增长团队的AI运营助手,将CRM、分析、邮件、广告和内容工具整合到一个工作区,通过连接实时漏斗数据自动执行跨工具任务(如起草内容、更新商机、生成报告),并让所有对外动作(邮件、帖子、广告修改)等待人工审批后再发送,解决团队在多个割裂系统间重复搬运数据、无法掌控执行状态的低效痛点。
SaaS Artificial Intelligence Marketing automation
AI代理 增长运营 营销自动化 CRM工作流 内容生成 跨工具集成 审批流 漏斗分析 效率工具 B2B SaaS
用户评论摘要:用户普遍认可“审批门控”设计,认为将AI限制在内部可逆工作、对外动作人工确认是务实且安全的方式。主要关注点集中在路线图:最高赞回帖询问“下一步最重要的功能或集成是什么”。官方回复强调审批不是安全网而是产品能全自动运行的前提,并主动邀请用户反馈不清晰或出错的地方。
AI 锐评

dolv的聪明之处在于它把“审批”从功能选项抬高为产品哲学。当前AI工具泛滥,多数停留在“生成草稿”的浅层,dolv试图定义“负责任的全自动执行”——内部工作(打分开票)可全自动,外部动作(发邮件改广告)强制人工闸口。这一设计确实精准击中了企业采用AI的最后心理防线:不是AI能力不足,而是失控恐惧。它没有试图用更强大的模型解决信任问题,而是用流程制度规避了风险,这比“更强AI”更实际。

但锐评必须指出其隐忧:32票的冷启动数据说明市场尚未形成口碑势能。宣称“29个集成+50个playbook”看似强大,实际是典型的“广度陷阱”——集成数量不等于工作流质量,Playbook的默认逻辑大概率无法匹配不同公司的复杂业务语境。最要命的是其定价:79美元/月起且“所有功能全解锁”,这在AI原生工具动辄按用量计费的当下,反而可能让中小客户觉得贵,大客户又担心额度不够用。更关键的缺失在于,评论中没有任何人提到“它跑通了我的完整流程”之类的结果验证——全是感觉上的认可,没有数据上的实证。

核心挑战在于:它是否能从“让团队少移动数据”的提效工具,进化为“重新定义GTM执行”的智能底座。如果能证明“审批过的AI执行”比“纯人工执行”带来显著的转化率或营收提升,才能摆脱“高级自动化玩具”的质疑。否则,这只是一款包装精美的IFTTT+ChatGPT外壳。

查看原始信息
dolv
Growth teams lose hours moving work between CRM, analytics, email, ads, and content tools. dolv brings that into one workspace. Give it a task: it reads your live funnel, does the work across your connected tools, and reports what shipped. Emails, posts, and ad changes wait for your approval before they go out. You keep the decisions. dolv handles the execution.
Hey Product Hunt 👋 We built dolv because we kept watching the same week repeat itself: re-entering data across tools, rebuilding the same reports, and chasing what had actually shipped. The real cost was not strategy. It was carrying work between systems that each held one part of the picture. Most AI tools stop at a draft. dolv keeps going. It connects to the tools your growth team already uses, grounds its work in your live funnel and company context, then helps prepare content, schedule campaigns, update CRM records, and report what the destination platform confirmed. By default, anything customer-facing waits for your approval. Emails, posts, and ad changes are held in Approvals, where you can review and edit them before they go out. Internal and reversible work, such as drafting content, scoring a lead, or updating a deal, can run immediately. Today, dolv includes: 🚀 ⭐29 integrations across Google, Microsoft, social, advertising, publishing, commerce, and search data ⭐Close to 50 playbooks across 11 disciplines ⭐A six-pass content workflow covering research, structure, drafting, SEO review, and quality checks ⭐CRM workflows with audited stage changes and weighted forecasting ⭐Funnel health scoring grounded in connected data ⭐A visual automation canvas that shows which steps run, queue for review, or require approval Every plan includes the full product. Plans differ by usage rather than locked features. Starter begins at $79 per month for two seats, and the 14-day trial requires no card. I would genuinely like to know: which part of your GTM week still feels like carrying work between disconnected tools? If you try dolv, please tell us where it works, where it feels unclear, and where it breaks. That feedback matters more than anything else today.👍 Thanks for taking a look.♥️♥️♥️
6
回复

@nolan_vu  Congrats🙌 to the team.. What is the single biggest feature or integration you’re planning to drop next on the roadmap?

4
回复

Congrats on the launch! Super cool product!

3
回复

@neo_tiangratanakul thank you very much, hope that you guys can enjoy it

4
回复

Hey everyone, Darius here, one of the makers on dolv.

Nolan covered the what. I'll add the why behind one specific decision.

The approval gate was the most debated thing we built. Early feedback pushed us to make it optional or off by default.

We kept it on because we watched what actually happens when automation touches customer-facing channels without a clear pause point: it gets switched off entirely, and the work comes back to the person anyway.

The gate is not a safety net. It is the reason the rest of the product can run.

If you have questions about how the approval flow works in practice, or how we handle cases where steps partially complete, I'm here for it.

2
回复

@nolan_vu Love the focus on execution, not just analytics. Keeping approvals with the team while dolv handles the repetitive cross-tool work feels like a really practical use of AI agents.
Good luck guys!

1
回复

@adana Thanks Adana! That's exactly the tension we kept running into: teams trust AI inside the workspace, but the moment it touches a customer they want a human in the loop. Glad it reads as practical rather than just cautious.

2
回复
#14
Supamodel
AI Product Photos at scale for Shopify
28
一句话介绍:Supamodel 是面向 Shopify 商家的 AI 商品图批量生产工具,通过可复用的预设、工作流和商品参考,解决多 SKU 场景下“单张图易得、全目录一致难”的规模化出图与返工痛点。
Design Tools Photography E-Commerce
AI商品摄影 Shopify应用 批量图片生成 电商工作流 产品图一致性 预设模板 3D资产支持 时尚电商 自动化出图 电商工具
用户评论摘要:创始人 Ajith 在评论中说明产品源于“目录级出图依然繁琐”的痛点,并透露已与时尚品牌测试数百款商品。目前无用户直接提问或批评,有效反馈集中在“真实生产中的稳定性”与“电商/摄影/代理商的实操建议”上,尚缺第三方验证。
AI 锐评

Supamodel 踩中的痛点是真实的:AI 生图已经廉价化,但“让 300 个 SKU 都保持同一模特、同一场景、同一光线”依然是地狱级工程。它的本质不是又一个 Midjourney 套壳,而是把“提示词-参考图-模特-环境-模型”封装成可复用的流水线,并直接嵌进 Shopify 的后台——这比独立生图工具更接近“生产系统”,而非“创意玩具”。

但必须泼冷水:第一,28 个投票、仅有创始人自述,毫无第三方评测或案例截图,所谓“时尚品牌测试”没给出任何转化率、耗时对比或翻车率数据,说服力不足。第二,AI 生成商品图的商业化难点从来不在“生成”,而在“审核”和“修图”——如果工作流里缺少强力的差异检测、局部重绘和人工纠错环节,规模化只会放大错误,而不是解决错误。第三,3D 资产支持听起来高级,但 Shopify 商家中真正具备 3D 建模能力的极少,这大概率是创始人的理想主义功能,而非核心卖点。

真正的价值判断在于:如果 Supamodel 能把“批量生成—人工挑选—一键回传”的闭环做到足够顺滑,并积累起“某一品类(如时尚)的高质量预设库”,那么它有机会成为 Shopify 生态里的细分插件,而不是 AI 摄影的颠覆者。现阶段它更像一个有想法的 MVP,需要尽快用真实用户数据证明“规模化一致性”这件事真能跑通,否则就会被 Klaviyo 们或 Adobe 的 AI 功能拍死在沙滩上。一句话:方向对,火候差得远。

查看原始信息
Supamodel
Create consistent, commerce-ready product image sets inside Shopify using reusable presets, workflows, product references, and supported 3D assets.
Hey Product Hunt 👋 I’m Ajith, the founder of Supamodel. I started building Supamodel after noticing a problem with AI product photography: generating a good image is getting easier, but doing it reliably across an entire catalog is still a lot of work. For every product, teams end up repeating the same prompts, references, models, scenes, corrections, and exports. And when you have hundreds of SKUs, that quickly becomes its own production workflow. So Supamodel lets you build that workflow once and reuse it. You can combine product references, prompts, talent, environments, and image models into a repeatable workflow, run it across different products, review the outputs, and publish the images back to your store. We’ve been testing it with a fashion brand that has already used Supamodel across hundreds of products, and I’ve been iterating closely based on what breaks in real production use. There’s still a lot I want to improve, so I’d genuinely love feedback — especially from ecommerce founders, photographers, agencies, and anyone experimenting with AI product imagery. Thanks for checking it out ❤️ — Ajith
1
回复
#15
OutageDeck
One status page for every service your stack depends on
18
一句话介绍:OutageDeck 将你依赖的所有云服务与 SaaS 供应商的状态页汇聚为单一看板,通过官方源追踪、告警与历史记录,帮你快速回答“是它挂了还是我们挂了”这一核心问题。
API SaaS Developer Tools
状态页聚合 云服务监控 故障告警 供应商状态 SaaS工具 开发者工具 MCP服务器 JSON API 事故历史 独立开发
用户评论摘要:用户主要点赞MCP服务器集成(让AI编码代理先查故障再动手)的创意。核心疑问:OutageDeck自身是否有公开状态页,以便在大规模云故障时验证其可用性?开发者未直接回应,但该问题直指工具自身可靠性验证的需求。
AI 锐评

OutageDeck的价值不在“聚合”,而在“克制”。它坚持只读官方源、拒绝爬虫和众包,以牺牲早期预警和覆盖率为代价,换取了零误报和可追溯性——这在“制造恐慌”的聚合器泛滥时代,是一种清醒的产品哲学。其真正的护城河是“数据契约”:每个事件都能回溯到供应商自己的机器可读声明,这在法律和审计场景中具有实际意义,而非仅仅方便DevOps查板子。

但问题也很明显:投票仅18,热度惨淡,说明“可靠性基础设施”叙事在PH这类流量场并不性感。免费策略(API/仪表盘无账号)降低了采用门槛,但付费点($19/月)卡在Slack/Teams/自定义源上,对个人开发者偏贵,对企业又缺少SSO等管理功能,定位尴尬。MCP服务器是亮点,但本质是给AI擦屁股——如果AI代理足够聪明,何必需要外部状态仲裁?

最大隐患是自身单点:跑在$55/月的基础设施上,作者是独立开发者,若他生病或放弃,整个服务消失。评论中“你自己是否有状态页”的疑问,恰恰点中了SaaS的达尔文死穴:监控别人的故障,却无法证明自己活着。建议作者立刻公开自身状态页并延长历史保留,否则“可信”这面旗帜会首先被自己烧穿。另外,Stripe被拒之门外虽显骨气,却也暴露了商业模式脆弱性——头部供应商不配合,目录增长就只能靠长尾,而长尾客户的付费能力通常有限。综合来看,产品方向正确,但商业化与可持续性远未闭环。

查看原始信息
OutageDeck
OutageDeck tracks 172 cloud and SaaS vendors from official feeds, with alerts, history, a free JSON API, and an MCP server your agents can query. Paid accounts can add private Statuspage or Instatus providers, including their own products: no crowdsourcing or scraping.
Hi Product Hunt 👋 I'm Kerolos, a solo developer. Every time something broke, I'd end up with a dozen status-page tabs open trying to answer one question: is it us, or is it a vendor? OutageDeck is that question, answered. It reads the official status feeds of 172 cloud and SaaS vendors, AWS, Cloudflare, GitHub, OpenAI, Google Cloud, Slack, Twilio, Vercel, and so on, every 10 minutes, and turns them into: - one live board for your stack. Pick your vendors, or paste a package.json and it finds them, and get one answer you can share as a link: https://outagedeck.com/stack?p=a... - outage alerts to email, Slack, Teams, Discord, or webhook - an independent, timestamped record of what each vendor acknowledged and when - a free JSON API with no account and no key - a status wall you can paste into a README: https://outagedeck.com/embed - an MCP server, so your coding agent checks whether the vendor is down before it starts rewriting your code: https://outagedeck.com/developer... Two things I decided early that shaped everything else. Official feeds only. No crowd reports, no page scraping, no synthetic probes. Everything traces back to the vendor's own machine-readable source. The tradeoff is honest: a smaller catalog than aggregators that scrape or crowdsource, and no "users are reporting" early warning. What you get instead is zero false positives and provenance on every data point. And because a fixed catalog can never cover everyone, paid accounts can point OutageDeck at any Statuspage or Instatus page it doesn't already track. That vendor then joins your account privately, with the same alerts and the same history as anything in the catalog. A few things I learned from reading 172 status feeds: - In the last 30 days, the catalog logged 1,429 incidents across 150 of the 172 providers, and 369 of them were rated major or critical by the vendor itself. - The longest incident currently open on a tracked vendor's own status page has been open for 694 hours. Vendors forget to close them, so I label that "stayed open" rather than "downtime", because those are different claims. - Stripe's public status JSON has been frozen since February 2024. It still answers 200 and still says everything is fine, so I refuse to list them rather than serve a fossil. - No consumer carrier, Vodafone, AT&T, Verizon, publishes an official machine-readable status feed at all. - AWS's feed is UTF-16 with a byte-order mark. The dashboard, API, badges, and RSS are free with no account. Email alerts for up to 5 providers are free with an account, about a minute to set up, no card. Paid plans from $19 add Slack, Teams, Discord, and webhook, remove the provider cap, and add your own custom feeds. It's bootstrapped and runs on about $55 a month of infrastructure. I'd love feedback, especially on which vendors you'd want tracked next. If it runs on Statuspage, I can usually have it live the same day, and if you're on a paid plan, you don't have to wait for me.
3
回复

@kerolos_atallah Congrats for launch🙌 Integrating an MCP server so AI coding agents check vendor status before blindly rewriting code is a brilliant, Is there a public status page for OutageDeck itself so users can verify if OutageDeck is up during a massive multi-vendor cloud outage?

0
回复
#16
Ventilate · At the right time
Beat the heatwave: when to open, close, and use a fan
18
一句话介绍:Ventilate 是一款面向无空调家庭的防暑引导应用,依据窗户朝向与实时天气,精准告诉用户何时开窗、何时关窗、风扇放哪里,解决“热浪天不知何时通风”的日常痛点。
Android Home Climate Tech Tech
热浪应对 通风提醒 智能家居助手 降温指导 窗户管理 风扇放置 天气联动 iOS应用 Android应用 生活实用工具
用户评论摘要:用户肯定“分窗定时”的价值,认为比笼统的“白天关窗、晚上开窗”更实用,并好奇开发过程中用户对“开关窗时机”与“风扇摆放建议”哪个更感意外。当前评论少,无负面反馈,建议持续收集真实使用数据。
AI 锐评

Ventilate 切中了一个真实且被忽视的需求:欧洲热浪下,大量无空调家庭在“通风时机”上凭感觉操作,结果越开越热。产品将“开窗/关窗/风扇位”拆解为可执行的定时指令,本质上是用天气数据替代生活经验,属于轻量级的“被动降温决策引擎”。其价值不在技术壁垒,而在场景聚焦:不试图控制温度,只优化通风效率,对目标人群(老式公寓、无AC、热带夜)极其精准。

但问题同样明显。第一,建议质量高度依赖室内温度输入(“有温度计更好”这句话本身就在削弱产品自信),若用户不买温度计或凭体感填数,误差会直接导致建议失真,产品易沦为“精致的玄学”。第二,评论里创始人对“用户更惊讶于时机还是风扇位置”的回复未直接回答,暴露了产品尚未有足够真实用户数据来验证核心假设——目前更像一个“我认为你需要”的工具,而非“你试过确实有用”的工具。第三,Product Hunt 上18票、0点赞评论,说明早期传播乏力,且“热浪”是季节性话题,获客窗口窄,若不能在夏季前形成口碑,留存将非常惨淡。

真正值得肯定的,是其“帮用户少做决策”的设计哲学——在热到无法思考时,直接说“现在关窗,23点开窗,风扇放东侧”,比任何教育内容都有效。接下来关键不在加功能,而在获取百名真实用户的使用日志,验证“温度误差对建议敏感度的影响”以及“提醒是否真的被遵从”。如果能做成“基于反馈自动修正建议”的轻量算法,则有机会从“定时器”进化为“个人降温教练”。否则,它很容易被系统自带天气App的“体感温度”+人工常识取代。

查看原始信息
Ventilate · At the right time
Ventilate helps you stay cool during heatwaves with calm, practical guidance. Set up your windows once, add your indoor temperature (a thermometer helps but is not required), and get clear advice on when to close windows to keep heat out, when to open for fresh air, and where to place a fan so cool air reaches you. Uses local weather and your window directions for per-window recommendations, plus gentle reminders and a heat toolkit.
Hi Product Hunt 👋 In 2019, during the first big heatwave in Paris, I remember how exhausting it already was. In 2026, after four more heatwaves, I saw how much harder it gets for people without air conditioning. Cooling down at home is not obvious, and every window decision feels urgent when you are tired and hot. That is why I built Ventilate: a calm guide for heatwave days, when to open, when to close, and where to put a fan so it actually helps. Heat is still hitting Europe hard, and I keep meeting people who open windows at the wrong hour or miss the short evening cool-down. Small habits make a real difference. What Ventilate does today: • Per-window advice based on your home layout and local weather • Next action times, so you know what matters now • A heat toolkit with practical guides, plus widgets and reminders Android and iOS apps are live on their respective stores. The app is in active development, and I am building it as part of RevenueCat’s Shipaton hackathon. I would genuinely love your feedback, especially if you live in a hot European city without AC. Question for you: what part of summer heat at home feels the most confusing today, timing your windows, fan placement, or something else? Thank you for being here early. I will be in the comments all day. Bathilde
0
回复

@bathilde_r The window timing idea is really interesting. I feel like most people have a vague “open them at night, close them during the day” rule, but it obviously gets a lot less simple when you factor in the actual weather and which side of the apartment the window is facing.

I’m curious, while building this, did you find that people were more surprised by the timing of when to open/close, or by the fan placement recommendations?

0
回复
#17
AFK
Command center for teams running coding agents
17
一句话介绍:AFK 是一个面向团队协作场景的编码智能体指挥中心,解决的是智能体任务在开发者终端中“各自为战”、缺乏持久化与团队可见性的痛点,让多个AI编码代理能在浏览器端被统一调度、审批和交接。
SaaS Developer Tools Artificial Intelligence
AI编码代理 团队协作 智能体编排 开发工具 命令中心 权限管理 浏览器控制台 子代理网格 企业级部署 自托管
用户评论摘要:目前唯一有效评论为开发者自述,无用户提问或直接吐槽。核心信息是:受够了智能体任务消失在终端标签页、团队不可见而开发此工具;支持守护进程拨号轻量中枢、浏览器连接,原生OS进程经WebSocket网格编排,基于Java 25 + GraalVM原生二进制。目前缺少真实用户反馈,建议关注工具审批流与交接体验。
AI 锐评

从产品形态看,AFK切中了一个真实且正在变大的裂缝:当Coding Agent从“个人脚本助手”升级为“团队常驻劳动力”时,终端标签页就是黑箱,而现有工具链(如Cursor、Copilot的托管会话)本质上仍是单线程思维。AFK的“守护进程+浏览器中枢”架构聪明地避开了云端锁定,把控制面放在Web、数据面留在用户机器,配合BYOK(17家供应商无加价)和Docker/企业部署,明显是冲着“让技术决策者说了算”的IT采购逻辑去的。

但必须泼冷水:17票、1000+注册,说明产品仍处极早期,且评论里只有作者本人在自说自话,缺乏第三方验证。其宣称的“P2P子代理网格”、“Java 25 + GraalVM原生”听起来玄乎,但真正决定生死的不是底层技术,而是两个关键问题:第一,在复杂团队流程中,审批会话、权限模式、计划审查这些动作本身会不会成为比手写代码更重的负担?第二,当多个子代理通过WebSocket网格并发执行时,如何避免状态冲突和上下文爆炸?如果这两个问题没有杀手锏级的设计,AFK很容易沦为“给AI戴了更重镣铐的管理后台”。

真正的价值在于它承认了一个行业盲区:AI Agent的产能不取决于模型智商,而取决于组织对其的可监督性和交接效率。AFK若能在“轻量治理”和“零燃尽”之间找到平衡,有机会成为团队级Agent基建的默认选项;若只是把CI/CD的权限控制套在Agent上,那它不过是另一个漂亮的后台,会被Agent原生框架(如LangGraph、CrewAI)的团队版后来居上。下一步建议盯紧其权限模式的细粒度——是否支持按会话、按工具、按时间窗口的临时授权,这比堆功能更能体现深度。

查看原始信息
AFK
Coding agent tools assume solo devs. For teams, that breaks. AFK is a browser command center for coding agent teams. Daemon on your machine or company server. Spawn sessions, approve tools, hand off work. Persistent sessions, permission modes, plan review, team orgs/roles, dashboards, handoff, BYOK (17 providers, no markup), Docker, sub-agents with P2P mesh, MCP/skills/plugins, automations, enterprise deploy. Free tier. 60-day trial. 1000+ signups in <30 days. Homelab, zero burn.
Thanks for checking out AFK! I built this because I got tired of agent work vanishing into terminal tabs with no team visibility. A daemon runs on your machine or company server, dials out to a lightweight hub, and you + your team connect via browser. Agents are native OS processes orchestrated through a WebSocket mesh. Java 25 + GraalVM native binary. Full feature set at https://www.mooglest.com — happy to answer questions.
0
回复
#18
t0md
Convert Anything to Markdown
16
一句话介绍:t0md是一款免费的在线文件转Markdown工具,支持PDF、Word、PPT、HTML和JSON格式,无需注册即可快速提取干净文本,并内置MCP服务器供AI代理直接调用。
Productivity Artificial Intelligence
文件转换 Markdown PDF转Markdown AI代理工具 MCP服务器 文档处理 在线工具 免费工具 开发者工具 数据提取
用户评论摘要:开发者评论指出Markdown已成为AI代理的通用语言,模型原生擅长处理且比HTML更省token。认为工具填补了反向转换需求,并提及此前md2doc.com被AI代理大量使用,暗示此工具同样具备实用潜力,暂无负面反馈。
AI 锐评

t0md踩中了两个精准的痛点:一是AI时代下,人类向模型投喂文档的格式摩擦——PDF、PPT这类复杂格式对模型极不友好,转成Markdown能显著降低token消耗并提升理解准确率;二是MCP(模型上下文协议)的顺势绑定,让该工具不再是孤立网页应用,而是直接嵌入Claude Code、Cursor等主流AI工作流的“格式适配层”。这种“转换器+协议”的双轮设计,使其具备成为AI基础设施级小工具的可能性。

但必须指出,16票的冷启动数据说明其尚未破圈,同类竞品如MinerU、marker等开源方案已具备更强的复杂版面解析能力,而t0md在介绍中未提及表格、公式、图片等富媒体元素的还原精度,这恰是实际使用中最易翻车的痛点。其核心竞争力“简单免费无注册”门槛过低,容易被复制,真正的护城河应在于MCP生态的深度集成和转换质量的一致性。若后续能提供批量处理、OCR增强及自定义输出风格,并开放API,才能从“顺手的小工具”升级为“AI工作流的必备中间件”。否则,它很可能只是又一个昙花一现的效率工具。

查看原始信息
t0md
Free PDF, Word, PowerPoint, HTML and JSON to Markdown converter. Drop a document, get clean Markdown in seconds. No sign-up. MCP server included for Claude Code, Cursor, Grok and other AI agents.
Markdown has quietly become the lingua franca of AI agents. Models are natively fluent in it (trained on mountains of GitHub and docs), it’s dramatically more token-efficient than HTML, and it hits the perfect middle ground of structure without noise. I saw how much demand there was for simple tools when I made md2doc.com and agents totally swarmed it to create documents out of markdown, so I thought it would be useful to go the other way too.
0
回复
#19
Account Moodboard
Visualize Instagram & TikTok accounts' vibe & stats
13
一句话介绍:Account Moodboard 是一款无需注册的免费工具,粘贴任意 Instagram 或 TikTok 账号链接,即可在数秒内生成该账号的视觉情绪板、钩子文案、话题分布及播放表现,帮助用户快速洞察创作者或品牌的内容“配方”与数据趋势。
Social Media Marketing Artificial Intelligence
社交媒体分析 创作者尽调 内容情报 AI视频理解 账号情绪板 TikTok分析 Instagram分析 免费工具 视觉化数据 内容策略
用户评论摘要:创始团队强调底层数据基建(AI逐帧、逐句、逐字幕解析全量视频)和 API 能力,用户无需注册即可用。有效反馈集中在:能否支持更多平台、导出 PDF 的深度、以及是否开放 API 给开发者二次构建。无负面批评,但请求增加功能透明度和数据更新频率。
AI 锐评

这个产品的本质不是“情绪板”,而是把“内容玄学”变成“可量化资产”的压缩器。它的真正价值不在于 UI 多漂亮,而在于 Oriane API 背后的视频全量解析管道——这恰好戳中了社交营销从业者的核心痛点:靠人肉翻 30 条视频总结“感觉”太慢、太主观,而平台官方 API 又限制重重。Account Moodboard 用“免费+零登录”的轻量外壳,让你先尝到数据甜头,再引导你去付费 API——这是典型的 PLG 打法,且漏斗设计得相当顺滑。

但必须泼冷水:第一,它目前只输出“诊断报告”,不给出“可执行开方”——比如告诉你钩子模式常见于 60-90 秒,却不告诉你如何复制该模式,这限制了从“分析工具”到“增长工具”的跃迁。第二,数据源覆盖只有 IG 和 TikTok,且依赖近 3 个月内容,对于深耕 YouTube 或小红书(国内语境)的玩家价值骤减。第三,评论里三位创始人齐上阵,但更像技术宣讲而非用户证言,缺乏第三方独立评测,可信度有待验证。

其护城河不在产品,而在数据管道——如果 Oriane API 能开放给更多开发者,并解决视频内容版权与合规问题,它有机会成为内容情报领域的 Twilio。但目前来看,它更像一个“诱饵”:免费工具做得足够好,让你愿意付费用 API,却未必能让普通用户长期停留在 Moodboard 本身。一句话:工具很棒,但商业模式和生态纵深仍需证明。

查看原始信息
Account Moodboard
See any Instagram or TikTok account's visual mood, hooks, and performance in one board. Paste a handle, get your free moodboard in seconds.
Hey PH! I'm Yuri, co-founder of Oriane. I've been building and launching free tools for a month now. This tool #4 on my "10 weeks to build 10 free tools". Tool #4 is to build accounts moodboard. When you're vetting a creator or analyzing a brand, you need to quickly get a sense of the vibe and the stats of the given account. But just looking at the follower count and the 3-4 latest videos on Instagram/Tiktok is not enough. So you end up doing it manually. Scrolling 30 videos, screenshotting a few, trying to describe a "formula" that's really just a gut feeling. We built a huge infrastructure that watches and listens to millions of videos across TikTok and Instagram so you don't have to. This free (no signup) tool is a small slice of that, using our API. Paste any Instagram or TikTok handle. Our AI watches the videos videos from the last 3 months and builds a moodboard: a visual collage sized by views, every opening hook transcribed in full, the actual hook pattern behind them, topic breakdown, performance over time, and which video lengths earn the most attention. The gut feeling, turned into something you can turn into a pitch. All free. No login. No sales call first. I'd love your feedback: what's the account whose "vibe" you've never been able to put into words?
2
回复

Hello there! Thibaut here, CTO of Oriane.


Following Yuri's comment, the hard part isn't building a Moodboard UI, it's the data underneath it. To generate this for any handle you paste, we're pulling every video from that account, processing each one across three layers (what's on screen, what's said out loud, what's in the caption), and fusing it all into structured records tied back to that single account. On demand, in seconds. That's the power of the Oriane API.

Most tools either rely on platform APIs that throttle you hard, or scrape one video at a time. We're running this across both Instagram and TikTok through the same pipeline, so any account's full footprint comes back instantly, no matter the platform.

Happy to answer technical questions for anyone building on top of it. Fire away 👇

1
回复

Julien here, cofounder of Oriane.

The problem is simple. Everyone studies accounts by scrolling and guessing. You can see the numbers on a big post, but not why it worked. So the real playbook stays invisible.

So this tool does the teardown for you. Paste any Instagram or TikTok account. It watches the content, surfaces the top hooks and breakout posts, charts views by video length and cadence, and writes short strategy notes on what the account is doing. You can export the whole thing as PDF slides.

Free. No signup. Under a minute.

We ran it on @ hormozi to test it: 4.4M followers, 98 videos, 25.2M views, and it instantly showed the 60 to 90 second sweet spot and the hook style driving the breakouts.

The fun part for this community: the whole thing is a thin app on the Oriane API.
Yuri built it in under a day with vibe coding. The API does the seeing and hearing inside videos, the app is just UI on top.

If you want to build your own tools on it, happy to get you access.

Would love your feedback on this tool, and tell us what other free products we should build in the next 6 weeks?

0
回复
#20
VICE Platform - Private Beta
Security scans for people who ship fast
12
一句话介绍:VICE Platform 是一款面向快速迭代的独立开发者的安全审计工具,以攻击者视角自动扫描 Web 应用中的密钥泄露、Supabase RLS 配置错误、API 暴露及安全头缺失等问题,并提供从发现、修复到复测的闭环。
Open Source Developer Tools GitHub Security
安全审计 应用扫描 DAST SAST Supabase安全 密钥泄露检测 开发者工具 CI/CD集成 开源CLI 独立开发者
用户评论摘要:评论者祝贺发布,并质疑开源引擎本身是否会引入安全风险(如攻击者利用公开代码寻找漏洞),开发者需回应开源与安全性的平衡问题。暂无其他功能或定价反馈。
AI 锐评

VICE Platform 切中的痛点真实且尖锐:现代堆栈让“能用”和“安全”之间的鸿沟被低估,而现有工具要么贵($200/月起)且面向合规团队,要么是渗透测试者的专业 CLI,中间层确实空白。开源引擎作为获客钩子,策略聪明——它建立了技术信任,也降低了试用门槛,但评论中“开源是否带来风险”的质疑很关键,这考验团队对引擎自身安全性的解释力。产品价值不在于“发现漏洞”,而在于“修复闭环”——带证据、给可粘贴进 AI 编码工具的修复建议、并支持复测,这直击开发者“知道有洞但不会修或懒得修”的惰性。但两个隐患:其一,“清理结果”的边界表述虽诚实,却降低了营销冲击力,用户需要理解“覆盖范围”这一概念,对小白不友好;其二,私人测试仅 10 个名额,且定价公开,这更像是一场高接触的种子用户验证,而非规模化的产品发布——若这 10 人不能转化为付费或案例,热度将迅速冷却。总体而言,VICE 有成为“开发者安全平权工具”的潜力,但需要证明:开源引擎不会被滥用,且修复建议的质量比得上专业安全顾问。否则,它可能只是又一个“看起来很美”的 CLI 壳子。

查看原始信息
VICE Platform - Private Beta
VICE audits your web app the way an attacker would: leaked secrets in your bundles, Supabase RLS misconfigurations, exposed APIs, weak headers and infrastructure.... The engine is open source and free today, as a CLI or a GitHub Action. The hosted Platform adds the full loop: verify your domain, run a full audit, get findings with evidence, apply a suggested fix, then retest to confirm it's gone. It opens as a private beta with 10 founding spots. Built for indie builders who ship fast.
Hey Product Hunt 👋 Modern stacks make it trivial to ship something that works, and just as easy to ship a secret into your client bundle, or a Supabase table anyone can read anonymously. The app runs fine. Nothing looks broken. That's the trap: "it works" and "it's actually protected" are two different questions, and most of us only ever test the first one. The tools that answer the second question exist. They start around $200/month and are built for compliance teams. Or they're CLIs built for pentesters. If you're a solo builder or a small team, there's nothing in between. What's free today: - The VICE engine is open source: automated DAST + SAST checks, Supabase/RLS analysis, API, GraphQL and WebSocket coverage, security headers and infrastructure checks. - Run it as a CLI or drop the GitHub Action into your CI. What's opening today: the hosted Platform, in private beta. Same engine, hosted, plus the loop that actually makes you safer: verify your domain, run a full audit, get findings with evidence (not just "trust me"), apply a suggested fix written so you can paste it straight into Cursor, Codex or Claude Code, then retest to confirm it's gone. I'm taking 10 founding participants. Not a waitlist: accepted participants get real dashboard access, run real audits on their own apps, and I personally review every high and critical finding during the program. Planned pricing is public on the site. The beta exists to validate the product, not to stay free forever. A few honest limits, because this is a security tool and trust is the whole product: - It only scans domains you own and verify. - A clean result means "nothing found in the covered scope", never "you're safe". The report always shows you what was covered. I'll be in the comments all day. Happy to go deep on how the engine works, what it found on my own apps, and why RLS is where indie SaaS actually bleed.
2
回复

@luca_deguin Congrats on the launch Luca. Does open sourcing teh engine not in itself pose a security risk?

2
回复