PH热榜 | 2026-08-17

Hey Product Hunt! 👋

Three months ago, I quit my job to build Meridian. My co-founder and I kept running into the same problem many developers face: we would ask claude to write an update, then spend ten minutes reconstructing the context it had missed. It knew what we told it, but not the four unexpected problems we had solved along the way.

That unplanned work is real, and bigger than most people think. I posted about it on Reddit: "We software developers massively underestimate how much unplanned work we do every single sprint" and it went viral, which told us we weren't the only ones feeling this.

Meridian does the remembering for us. It runs quietly in the background, groups the day into a timeline, matches the work to open Jira and GitHub tickets, and drafts the update - so by the end of the day there's a work log, a standup note, and a daily summary ready to paste, not write.

A few things I care about, as a developer myself:
- We care deeply about privacy - nothing leaves your device until you hit approve.
- Open source - you can audit exactly what it's doing.
- Free for individual developers. No seat, no credit card.
- Bring your own AI - runs on the Claude Code, Cursor, or Codex CLI you already have, or Groq for free if you don't.

Try it

Free download — Meridian for macOS (Apple Silicon, macOS 14+) or Meridian for Windows (Win 10/11, 64-bit). Everything else is at meridiona.com, and if you like what we're building, star us on GitHub.

We’re live today! I’d love your honest feedback, especially on the onboarding, activity timeline, and drafted work updates. Every comment helps us improve Meridian.

@akarsh_hegde This is a really interesting approach! I especially like the idea of taking the “remembering and reporting” part out of the workflow so people can focus on the actual work. And running entirely on-device with no account is a big plus for privacy. Great concept - wishing you a successful launch!

This is a fascinating product, congratulations! If a lot of my work happens on a remote VM that I SSH into from a local computer, does this support that, or plans to support that?

Meridian seems to absolutely need an account and can't be used without it. What is the account for, and if it is to get product updates, am curious why it's gating usage. The app also requires quite a bit of access such as audio and screen recording, as well as accessibility. Am not sure the claim "privacy first" matches these required permissions especially with llm egress into the mix?
Also curious why the app isn't on the app store where there is a third party attestation you're not doing anything weird?
I understand the open source claim, but as long as the user doesn't compile the app themselves there is less visibility into what it does, and 2 of your claims from the launch description sadly don't seem to match the actual app :( .

(Disclosure: I'm building in an adjacent space and have built developer tools at scale, which is why I care about how privacy claims get made in these categories)

Great idea! End-of-day updates feel less like a task and more like something that just happens in the background. Definitely something I’d find useful in my daily workflow.

Congrats on shipping. Any plans to launch a team plan?

From the positioning standpoint, this can someday also replace those time-tracking apps. Well done!

Does it not auto-log the tasks based on what it sees on the browser? I tried it but it didn't log what I am doing inside Chrome.

This is cool! I’ll be able to focus on the actual work instead of getting stuck in operations.

Such a smart and needed tool for career growth. Congrats on bringing this to life!

Finally an AI that translates "Fixed some stuff in the backend" into human words so i can actually write a changelog😅

Can non-devs use this product? I would love to!

Is it voice-led right?

I have been using the product for more than 2 months now and I am one of the beta users, I love the product and how it's shaping up, great going!

You quit your job three months ago to build this, and it's MIT-licensed and free for individual developers. That's a lot to give away on day one. Congrats.

Three months ago you quit your job to build this. It's MIT-licensed and free for individual developers, a lot to give away on day one. Congrats on shipping it.

Huge congrats to you and the team launch, Akarsh! I really like the problem you’re tackling, the invisible, unplanned work that happens between tickets and commits is often where the real story of a project lives.

The privacy-first and open-source approach makes the product even more compelling, especially for something that’s essentially building a memory layer around how people work.

Wishing you and the team all the best.

Congrats. Looks very interesting. 😊 What's on your roadmap?

Wait so my devs will actually update the Jira without me asking 14 times a day? Take my non-existent money:D

Congrats on the launch!

This is really cool. How does it map the work with existing tickets.. lets say in Jira?

nobody is going to believe that I actually got something done while just being at my macbook all day long. :)

I was always reluctant to post, but wanted to do Day 1, Day 2 kind of post for #buildinpublic,
With Meridian, finally I can do it , glad to discover the product.

Congrats Adithya and Akarsh! The local-first approach and automatic Jira/GitHub updates solve a real developer pain point. Looking forward to seeing Meridian grow. 🚀

This comment was fully written by an actual human. Yes really

Hi hunters,

Ran here, co-founder of xpander.ai.

If you ever found yourself troubleshooting OpenClaw or Hermes at 2am, staring at Cowork or Claude code waste your tokens on failed jobs, or got stuck in a 100-prompt loop trying to achieve something with your "AI Assistant" - then this product launch is for you.

We built Omni so you can start using AI, instead of tinkering with it.

Omni is your agentic teammate

You ask, and Omni executes like an extremely useful generalist agent. Then, it can actually build your custom agents that are mission-specific for any business process you have - it wires the connectors and skills, sets up the system prompt and schedules, then tests and ships it.

After that, it does the part that used to eat up your time as you tried to stand up something that really brings value: reading the logs, troubleshooting failed runs, optimizing performance, and benchmarking models for cost and quality to pick the right one for the task.

The babysitting is Omni's job now, not yours.

Real asks from early users:

• "Benchmark this agent across latest Sonnet, GPT, and Qwen models. Tell me which is cheapest at the same quality."

• "Watch our cloud bill. When something spikes, find the cause and post it in Slack."

• "Every morning, pull yesterday's signups, enrich them, and drop the interesting ones in our channel."

• "Last night's run failed. Figure out why and fix it."

Omni works great if you're a solo-user, but is designed for Multiplayer AI work - Agent builders can share their agent with other team members, so their hard work doesn't stay confined to only making themselves great at their job, but the entire team or organization.

It's built on xpander enterprise-grade agent infrastructure

Omni is an omni-present AI Agent built on top of the xpander platform - a battle tested agent infrastructure product that already drives Agentic AI in Fortune 500 companies across various industries, and different sizes small-medium enterprises. Omni is just the latest addition to that platform, that makes driving it automatic, easy, and fun - instead of manually standing up AI infrastructure and building agents, you can start just getting ROI from agents.

Some of these capabilities that are relevant if you're a part of a company with agent infrastructure requirements:

• Self-deploys to any cloud, your VPC, on-prem, or fully air-gapped.

• Sandboxed on isolated compute

• 2,000+ spec-enriched tools and MCP connectors, plus create a connector to any API

• Runtime environment optimized for security - for example, secrets are injected from a vault into downstream tools, with the model never seeing a secret.

• Choose any model from any provider, instead of getting married to the major AI labs.

• Permissioned per user and audited on every action. This platform has passed significant enterprise security reviews.

Omni puts the power of agentic AI in your hands without all the required tinkering.

We really hope you'll love the product and the experience - we're waiting for you to let us know here in the comments, or in the built-in feedback screen in the product.

Free to start now at chat.xpander.ai.

Real question, how much of the 'fixes itself' actually works vs needs me to step in.

Not being cynical, I've just been let down by this claim before.

Congrats on the launch!
BTW, when a target site changes its layout, does the Bot auto-detect the break and self-heal, or does it silently start returning bad data until you notice?

what happens if a scheduled run hits something that needs a human decision. does it wait, ping someone, or just die?

The mock data testing before putting an agent in the cloud is a really useful touch. It could save a lot of debugging time later.

Huge congrats on the launch! "Stop babysitting your AI agents" hits the exact pain point so many builders and teams are facing right now as agentic workflows scale. Love seeing tools that focus on execution reliability and autonomy rather than just hype. Excited to see how this evolves and saves teams endless hours!

Congrats on the launch! sounds like an interesting tech.

Can you please share more details about the runtime environment? are you using some off-the-shelf harness or did you build your own?

the "stopping when the lid closes" line hit home - I was literally in a forum thread yesterday about caffeinate vs actual lid-close sleep on macOS for this exact reason, so this feels like the real fix past that hack. curious about the "compares models to pick the right one for the task" part specifically - once an agent is running fine on a given model, what triggers a re-comparison? is it every scheduled run, only on failure, or only when I ask, and if it decides to switch mid-flight does that need my approval or does it just swap silently

Who pays for the tokens on scheduled runs? that's usually where these get expensive fast.

Sounds cool. I liked the part about the daily signup summary and catching the bug in the push to the PR.

Benchmarking models for cost at the same quality sounds so useful. going to check it out soon.

I am wondering how the agent does not touch the source code but touches the database and possibly security design? Does the product owner get to know what has been added on by a user and who then is responsible when something does not work well?

Let us know if you have any questions, we'll be here answering everything!

Absolutely love Vendo!

Awesome team. Congratulations!!!

Congratulations on the launch! Curious to hear, is there anything that users built with Vendo that surprised you?

The whole customization layer is open source, repo and all, and it installs with a single command. You could have kept that closed. Congrats, easy to root for.

Always been thinking this is possible with AI. Here it is! Congrats.

Very interesting guys. I really like the idea, but I am curious to howthis really work in practice. Does the user request really buid a feature in realtime for just that particular customer? Have seen some cool demos of the new gimini for this lately. Definetly some potential here. Congrats on the launch!

Very Interesting! Can this pull in data from tools outside our product, or is it only our own data?

I found navigating many SaaS UIs (e.g., Rippling) frustrating. They're built to serve every possible customer, so the UI is crowded with features and buttons that have nothing to do with what I came to do. As an end user, being able to personalize the UX will be super useful!

interesting - can def see adv of this as a way to add val to customers. hows it sandboxed tho ?

This makes so much sense, I’d love to have customizability in my software. Would also makes sense to identify what people are building and make it generally available to everyone. Is this something you help with?

The future is personalized software! Super cool to see. What kind of analytics are there to see what people are building?

Good to see the second launch, and this one is open source with a one-command install. It runs on the customer's own API and never touches their source code, which is a hard line to hold.

This is the future of SaaS.

Congrats guys. How do you handle a PM's fear that customers will customize themselves into pure mess and it will be hard to take it back? and also generate support issues they aren't responsible for?

Amazing product! One question though, can we control which parts of the product are open to this and which aren’t?

👋 Hi Product Hunt, I’m Tzahi, co-founder and CEO of Clears.ai.

Before starting Clears, I spent more than 10 years leading R&D teams. Over the last two years, my Co-founder and I have spoken with hundreds of CTOs and VP R&Ds about how AI is changing software development.

We kept seeing the same thing: developers were writing code faster with tools like Claude Code and Cursor, but their organizations weren’t shipping significantly faster.

The bottleneck had shifted from coding to execution. Tasks still require context, clarification, coordination, validation, and constant follow-up before they become production-ready code.

We built Clears to automate that process.

Clears is an agentic execution platform for autonomous software delivery, helping R&D organizations move beyond individual AI tools to agentic execution across the entire SDLC. 

What makes Clears different:

🔹 Starts with the right context
Clears creates and continuously maintains a deep organizational context layer that enables agents to execute complete workflows, from requirements and implementation through validation.  

🔹 Delegates the right work
Clears scores tickets for risk, complexity, and confidence before determining what is ready for autonomous execution.

🔹 Runs the whole backlog
Every ticket gets its own live agent session, so teams can execute work in parallel and step in whenever needed.

🔹 Keep your existing workflow
Two-way sync with your existing tools means there’s no new system for your team to adopt.

Teams can connect Clears in about 30 minutes and start running real tasks the same day.

We’re excited to launch on Product Hunt and hear what builders and engineering leaders think. We’ll be here throughout the day answering questions and listening to your feedback.

You can try Clears with your own team through our free trial and see how it fits into your workflow: [FREE TRIAL]

Thank you for checking out Clears 🙏

Tzahi
Co-founder & CEO, Clears.ai

A lot of teams have AI tools already, but connecting them into one workflow is where things seems to get messy. Intersting direction.

Congrats to your team on the launch!

What caught my attention is the focus on the whole software delivery process, not just code generation. That feels like the harder problem to slove.

Really cool! Delivering a software end to end by doing whole things using workflows. Congrats on launch.

Super cool concept moving from coding to full agentic software delivery. Congrats on the launch!

Congrats on the launch! The risk/confidence scoring before autonomous execution is the part I want to understand better, is that scoring done statically from the ticket description, or does Clears actually inspect the codebase to calibrate complexity before spinning up the agent session?

Hi Product Hunt! I’m Vinitra, co-founder of Scholé AI! :)

We built Scholé because most learning tells you what to know, without giving you enough chances to practice what you’ll actually need to do. We're really, really passionate about getting learning right.

Today, we’re launching ✨scenarios✨: real-world situations built directly into adaptive lessons.

You can create a scenario to practice anything you want to get better at: e.g. pitching to an investor, handling a difficult retail customer, explaining a new AI concept to a teammate, or infinite more.

The key difference is that these aren’t standalone role-plays. Scenarios dynamically appear throughout your learning journey, integrated into your lessons. Scholé tracks what you understand and adapts what you learn and practice next.

We do this through a system of pedagogical agents working together to construct the right lesson for you: at the right difficulty, with the right modalities (we have over 20 now!), on the right topic, with the right examples. We won #1 on Product Hunt with our first launch in May! And we have a lot more coming.

Scenarios brings together three ideas we care a lot about: learning by doing, learning by teaching, and scenario-based learning. We co-designed this tool with UNESCO and Decathlon, and are very grateful for their support.

We’d love for you to try for free (https://schole.ai/scenarios) and tell us what you think! No Scholé account needed. 🎉

Congrats on number two. You co-designed this one with UNESCO and Decathlon, which takes longer than building alone. Free with no account needed is generous.

Congrats on the launch! I like that the scenarios are embedded into the learning journey instead of being separate role-play exercises. Adapting what someone practices next based on how they handled a real scenario feels much closer to actual skill development than just generating personalized lessons.

Congrats on the launch 🚀 This looks great. Curious how progress shows up over time. Is there a view that tells a team lead where their people got stronger?

Congrats on the launch!

Congrats on the launch! Can I create scenarios and share it with other people? I mean - as a manager, I might want to use Schole for coaching my team - is that possible?

Hey Product Hunt!

Today we're launching OpenTrade, an open-source macOS app that helps manage Claude Code / Codex agents that trade via Robinhood.

When Robinhood released their MCP in May, that marked the first time most people could point AI agents at their portfolios. Overnight, everyone had the potential to have a financial advisor in their pocket keeping an eye on their investments 24/7.

At least that was my expectation! What I found instead was just the foundations for that vision:

• Agents couldn't proactively react to market events

• Mistakes made by agents risk the entire portfolio

• Difficulty managing multiple agents on the same account

That spurred the development of OpenTrade, which provides:

• Monitors that can notify agents periodically or based on custom scripts agents write themselves

• Headless sessions so agents can work in the background

• Guardrails and order approvals specific to each agent

• An IDE-like environment to easily manage multiple agents

As avid Robinhood users ourselves, we recognize the hesitance to trust third-party software in this context, which is why OpenTrade is open-source – anyone can inspect and modify it. We also firmly believe that a product to manage your finances should be completely under your control. OpenTrade spawns your own configured Claude Code or Codex agents on your machine – never on a server.

We'd love to hear from the Product Hunt community, particularly from the active investors in the crowd:

1. If you're using agents with your portfolio today, what works and what doesn't?

2. Are there other brokers that you'd be interested in trying with OpenTrade?

Thanks for checking us out! OpenTrade is built by Exla (YC W25). @viraat_das and I will be around all day.

The idea is compelling, but for me trust will come down to the guardrails and audit trail. If I can clearly see what the agent tried to do and why, I’d be much more willing to test it.

Managing multiple agents from one place sounds useful, especially if each agent has a different role or set of permissions.

Very cool. What is your team using in terms of inputs to influence your agents' trading decisions?

I am a not a active trader, would love a tool like this , what creds do i need to get started with it , do es this work in India?

Love that you kept this open-source and local-first. I don't trust giving any bot access to my broker on a server. Are there any guardrails/templates for beginners so we don't accidentally have an agent blow up the portfolio?

Very cool! Would you consider building it in native swiftUI etc? And any thoughts on mobile apps? iOS / iPad etc? Different UX for sure but would be very interesting. Like a remote side kick for the desktop mothership.

Congrats on the launch! Running the agents locally while keeping per-agent order approvals feels like the right approach for something as sensitive as trading.

Curious how you handle conflicts when multiple agents want to act on the same position at the same time?

Hey Product Hunt 👋 Jason here

We started Treg as our internal tool, out of a simple frustration: the data & system an agent needs for real work - keyword volume, backlinks, ad libraries, enrichment - either sits inside SaaS bundles priced for humans. $139/mo, and you don't even know what's in the box OR requires days of OAuth app setup & verification.

Instead we think future should be task-based instead of vendor-based - agent just ask the task, it knows all endpoints with request/response/price, decide the best API for the task - and to pay for the result.

So:
• 2,600 agent-friendly tools across ~40 providers: SEO/GEO, social, leads, ads, scraping
• Search by task ("backlinks for a domain"), see price / request / response, call it
• Pay per call, no subscriptions - 0% markup. Same unit price as the subscription; BYOK if you already pay and those calls are never metered
• It also handles the painful OAuth setups (Google/Meta ads, social posting, Business Profile) - connect once, every teammate's agent can act
• The proxy relays the real upstream request and injects credentials server-side - your agent never holds a secret, and we never model the upstream API, so we survive provider changes

Open source (AGPL), self-hostable.

Tell me what your agent does with it - and what's missing from the catalog. I'm here all day.

This is super cool, just spent a few mins to setup and try, spent less than $1 and got some really useful analytics for tiktok, ig, plus a solid best practices breakdown

I've been thinking about this as it relates to product discovery. Many products, especially APIs and data providers, will primarily serve agents which changes the discovery process.

@jasonzhou1993 Treg seems like something the foundational models and harnesses should offer. Wdyt?

interesting stuff. how do you handle GDPR and EU AI Act compliance?

the proxy-injects-credentials-server-side part is the piece I keep coming back to. it solves the OAuth pain for sure, but it also means every request and response body for stuff like leads/ads/scraping data is passing through your infrastructure, not just the auth token. for someone routing real customer or audience data through this, what's the logging/retention policy on that traffic - is it truly zero-log passthrough, or is there some window where request/response payloads are kept for debugging billing disputes

Hey Product Hunt,

Marcos here from @Replay

Last time we launched Replay QA, the most common feedback was the same: "How can my team use this?"

Most production code is AI-generated now, and most tests are written by the same AI. CI goes green. The app looks done. Then someone opens it in a real browser.

That's why we built Replay QA. Give it a URL, and it explores your app the way a new QA hire would, no spec, no one to show them around. It clicks through your app in a real browser, then reports what broke, where, and with root cause detail and a suggested fix. Every session is recorded with our time-travel debugger so you can replay the exact moment something went wrong.

This launch is about teams:

• PR checks - Replay runs on every PR, catching issues before merged. It posts debugging context right into the comments. No CI config.

• Team projects - invite teammates, share findings, @mention in bug reports.

• Localhost support - check before you push. You drop in a localhost URL, and we walk you through the reverse proxy setup - everything configured with you.

Drop in your local or live project URL, then connect GitHub to auto-retest on deployments or PRs.

If your app is posted publicly, it's free to test. Try it at: qa.replay.io

We'll be here all day answering questions about PR checks, how QA fits alongside testing tools, and - if you have an app you'd love us to test.

Thanks to everyone who gave us that feedback and to the beta testers who told us when we were wrong.

🚨 Looks like GitHub is experiencing issues again today... if you're running Replay QA connected to a repo (and if you're pushing bug reports to GH Issues), you might encounter some weirdness. https://www.githubstatus.com/ (we're keeping an eye on things)

this is th kind of startup program I appreciate because it helps with a real technical cost rather than just offering another collection of perks I may never use.

Let's go!! Congratulations with the launch @Skriptr !!

The Devil’s Advocate feature is the part that stands out to me. Having AI challenge a student’s argument with evidence from their own sources feels far more useful for learning than simply polishing the final answer. That’s a much healthier direction for AI in education.

Been following these guys for a while. Beta was great, now its FANTASTIC.
True supercharging of the student, no cutting corners just helping you do better, faster.

@rolf_bekkelund

This is one of the very (VERY) few products that I hope rules the educational space because this is a pushback against AL-SLOP, Go write it for me and then I'll hack it shortcuts, and epidemic hallucinated sources flooding the academic landscape. If students can get nudged (or pushed/dragged) back to critical thinking and original work based on that thought, AI might, just might, land in the right place within those halls of learning.

Good luck hardly covers it. Maybe prayer? :-)

Well done.

This is actually quite interesting! Could you share any recent successes you've had?

I started my career as a PM. Wild how much the role has changed today. It's always been highly leveraged but now 10x with tools like this.

Looks amazing!

So excited to try this out at @Equals !

There's something really appealing about opening your work and already feeling a few steps ahead

it shifts the whole mood of getting started, and that matters way more than it usually gets credit for

congrats Sahil ;)

Hey Product Hunt, Sahil here, co-founder of Samepage.

We are back with our 2nd Product Hunt launch, and today we’re launching Samepage Artifacts.

We built Artifacts because PM writing usually starts with a blank page, then a bunch of digging through Slack, Linear, customer calls, analytics, and old docs just to gather enough context to write something useful. Even today, starting with any LLM often loses the thread, misses important context, or doesn't exhibit real product expertise.

Artifacts is the PM’s writing surface. It helps turn the context already scattered across your tools into polished docs like PRDs, feature briefs, release notes, and status updates. Any Artifact can then be sent directly into systems like Jira, Linear, Confluence, Slack, etc. to be picked up and worked on.

Instead of manually stitching everything together before you can write, you start with the context and get to a strong draft much faster. In fact, Artifacts will have drafts pre-loaded and waiting for you given Signals (our first PH product launch) is always actively monitoring what progress is being made or what release notes should be written.

Signals does a great job of surfacing important information like feature ideas from last week's sales calls, and then Artifacts lets you package that up into a feature brief or PRD that you can send over to Linear as an issue or assign directly to Linear's coding agent to build.

For any PMs out there, hope you'll give it a shot and let us know what you think! We have plenty more coming but excited for this one.

Paul here, building product at Samepage.

Like many PMs, I've been doing more mockups and UI work directly in the codebase. I tend to use Cursor because, unlike Claude and ChatGPT, I can keep the codebase open in front of me and occasionally edit line by line. After all, sometimes it's faster to type a few small corrections than to chat with the copilot and wait for the round trip.

But Cursor is tuned for engineers, not PMs. I wanted a writing surface for writing product documents instead of writing code. That's why we built Artifacts.

With Artifacts, Samepage AI will write your draft, and then you can hop in to iterate with the copilot or manually edit the document in place. When your document is done, send it directly into your system of choice for your engineers or marketers to pick up.

Try it for free and let me know what you think!

I know that you said that the work can then be sent into Jira, Confluence, and Slack, but does the context go the other way around as you continue to build out a feature set to keep the context up to date with Samepage on Artifacts (or Signals)?

As someone who's in Slack and Linear all day I'm very excited to give this a try!

Hey Product Hunt 👋

I'm Kemone a former Googler on the original GTM team for Google Workspace.

Last month Google cancelled my Workspace subscription and locked me out of my own email, contacts, and everything instantly. No warning. Early Termination on our grace period.

That day I decided: never again.

I built Hansel in under 30 days with my AI engineering partner Bob. It's a sovereign communication platform everything gmail or outlook gives you, except YOU own it.

Your email runs on your own server. Your messages encrypt on your device before they ever leave. We built our own cryptographic protocol called Wicked Crumbs — rotating identities, tamper-evident message chains, spam solved at the architecture level. Without your key, a message simply cannot arrive.

No company can lock you out. No company can read your data. No company can cancel your account. Because none of it lives with us.

This is what I wish existed before Google flipped that switch on me. I hope it helps someone else never feel that powerless again.

Download for Windows ... macOS and mobile coming soon.

Happy to answer any questions 🙏
— Kemone Phillips II

If this is what I think it is then why would we ever need standard email ever again? The idea of complete ownership definitely is enticing. Congrats on the launch team.

Did someone say launch day?! 👀

It’s been a journey watching Hansel come together and I’m excited to finally see it out in the wild. The more I’ve learned about what we’re building, the more I’ve appreciated the idea behind it. Having more ownership and control over something as important as your email instead of handing all of that over to a platform just makes sense.

There’s a lot that’s gone into getting Hansel to this point, and now I’m really excited to see what people think once they get their hands on it.

Go check it out and let us know what you think! 🙌

As a spreadsheet hell rider, this sounds like heaven. Nice work @leslie_depond !

Hi Product Hunt! 👋 I’m Leslie, the builder behind Bulbthings.

We noticed a massive irony: scaling businesses are building cutting-edge products yet they’re still tracking their physical assets (e.g. IT equipment, high value-value prototypes, A/V gear, machines, vehicles, furniture, facilities) across messy spreadsheets and chaotic WhatsApp threads.

Notion lacks the asset-centric workflows and analytics needed for scaling teams, while legacy IT Asset Management (ITAM) and Computerized Maintenance Management (CMMS) platforms are clunky, enterprise-priced, and bloated.

So, we built Bulbthings: a modern, AI-powered asset inventory, tracking and maintenance management platform designed specifically for fast-moving teams.

Here is why we think you'll love it:

• ⚡ Zero-Friction Onboarding: Select your industry and Bulbthings instantly auto-configures your workspace with the exact features and templates you need to automate any assets you rely on for your operations. Plus, our smart import module turns your messy spreadsheets into structured data in clicks.

• 🤖 Unified AI Copilot (paid feature): Faster onboarding and automation. Just tell Bulb AI what you need: snap a photo or scan a QR code to add gear, assign a MacBook to a new hire, book equipment for a project, report an issue, or pull instant reports—all via natural language (chat or voice).

• 🛠️ Developer-First (Headless AMS/CMMS): Your data, your rules. Use our open API and SDK to build complex workflows or new apps.

• 🔓 Un-Gated & Fair: Tired of upgrading to Enterprise just to use one key feature? We don't gate functionality. Mix and match feature packs as you grow.

🎯 Who is Bulbthings for?

We designed this for growing teams that have outgrown spreadsheets to automate their asset inventory and workflows but aren't ready for complex, legacy enterprise software.

🎁 The Offer for the PH Community

We have a highly capable "Free Forever" tier so you can get your operations out of spreadsheet hell today without pulling out a credit card. Get started for free here 🚀

What's Next?

We are on the verge of launching our AI Developer Agent, designed to help you customize your workspace or build asset-centric features and applications even faster. We would love your feedback on the core platform, API, and these upcoming AI capabilities.

Hey Product Hunt 👋

I'm Deniz, PM at Text2Test.

AI writes our code in minutes now. That's the easy part. The hard part: you can't fully trust it, and you can't easily validate it. Fast code you can't verify isn't actually faster, it just moves the risk downstream.

Testing should close that gap. But it stayed manual, slow, and brittle, breaking on every UI change. So we built Text2Test.

You describe a test in plain text, and it compiles into a deterministic test that runs the same way every time on real browsers. Consistency you can count on, not flaky results you re-run. When your app changes, tests self-heal, so you stay in control instead of maintaining them.

We are ambitious about hitting the needs. Try it, tell us what's missing, we welcome all of it and read every comment. Thanks for checking us out 🙏

@deniz_colakoglu1 - Interesting product, would this work for native ios and macOs swift test using xcode?

Julien here, cofounder of Oriane.

The problem we're solving with this free tool is simple. Every brand and creator watches a rival account. But comparing two feeds properly means watching weeks of videos on both sides, so nobody actually does it. Strategy decisions end up running on vibes and follower counts.

So this tool does the watching for you. Drop in two Instagram or TikTok accounts. The AI watches both feeds, breaks down each account's hooks and content mix, and lines up the stats side by side: views, interactions, engagement rate, posting rhythm. You can even send the whole report as PDF slides.

Free. No signup. Under a minute.

We ran it on @cocacola vs @drinkpoppi. Coca-Cola has 3.2M followers and pulled 15.5M views over the period. Poppi has 662K followers and 5.6M views. But poppi gets a 3.1% engagement rate against Coke's 0.6%, and about double the interactions. The giant wins reach, the challenger wins attention.

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 5 weeks?

Thibaut here, CTO of Oriane.

Technical angle on this one. The tricky part isn't the side-by-side UI, it's making two accounts you know nothing about actually comparable. Different posting frequency, different formats, wildly different audience sizes. Raw stats alone won't tell you who's winning. They just tell you who posts more.

So each account gets processed independently through the same pipeline (vision, audio, caption, all fused per video), then lined up on the same axes so you can actually compare apples to apples. Topic mix, hook patterns, performance trends over time. That's what turns "who's really leading, and on what" from a guess into something you can actually answer.

Same setup as everything else we've shipped this month, running on the Oriane API, across both Instagram and TikTok, on demand.

Happy to go deeper on how we handle accounts with very different posting volumes, or how we normalize engagement across platforms that measure things differently. Fire away!