Launching today
Tiden
Intent verification for coding agents
247 followers
Intent verification for coding agents
247 followers
Tiden verifies your coding agents built what you actually asked for. Most checks run inside the session that wrote the code, so the tests agree with it by construction. Tiden keeps your intent as requirements that branch and merge like code, and checks every change against that record. The verdict comes from tests run in your environment; no model grades the diff. You decide what merges, and production errors trace back to the release. Works beside Claude Code, Cursor, Codex and Copilot.
Joined the Tiden team not from the very beginning, so there already was a lot of functionality, taken decisions and other things that make me nervous when I join new teams. Especially in this AI era, where even the most frontier model might want to shape the reality around itself in order to finish its job.
But building Tiden while using Tiden makes me worry much much less, I know Tiden will force my agent to take into account previous decisions and existing specs and no adjacent features will be broken because of that.
The ability to trace production errors back to a release is a useful addition. It could make debugging much less frustrating.
@anthony_adams_ Yes debugging is easier and fixing those errors is a few clicks away, so any post production issue can be responded and fixed in minutes.
I like that the final verdict comes from actual tests instead of an AI judging the code. That feels more reliable.
@maali_baali Yes, tests are enforceable boundary that protects against agents claiming their own victory.
Great launch! With agents writing most of the code, keeping up with quality is becoming the real bottleneck. Specs that branch and merge like code is a smart touch
@rodrigo_montemor Exactly product behaviour which was either a tribal knowledge or deep in code is now first class artifact that Tiden takes care of building and maintaining so no one has to spend out hours figuring out cross impact of changes.
A green test suite can still be answering the wrong question. I like that Tiden keeps the request as a separate record for checking what ships.
@oleksii_sekundant The tests act as enforceable boundary to reduce the risk of regression and take a away load from CI-CD pipeline. Tiden's ability to maintain test coverage as product evolves on its own takes away lot of maintenance overhead engineering team faces today.
@oleksii_sekundant Exactly, its a system or record that keeps whole context of a release right from requirements to post production errors so finding what went wrong becomes lot easier when each piece is seperate.
Hi Product Hunt Community,
I am Nitin, and I am product manager for Tiden. Today, after six months of building, we are finally sharing it with you.
Here is the problem in one line: your agents are fast, but your delivery is not.
Coding agents can write a change in minutes. Yet that same change often takes days to reach production. The PR waits because nobody is sure what else it touches. CI catches a regression on the second push. A reviewer spots a function the agent invented. The code got faster. The delivery did not.
The data backs this up. In Stack Overflow's 2025 Developer Survey, trust in AI accuracy fell to 29%, and 66% of developers said they spend more time fixing "almost right" code. Faros AI found that teams with high AI adoption merge 98% more PRs, while review time rises 91%. Google's DORA research put it simply: AI amplifies what is already there.
The market's answer so far has been a point AI tool for every phase: one for context, one for tests, one for PR review, one for release signals. They are faster cars on old roads. The lifecycle underneath was designed for humans who read specs and follow instructions. Agents sometimes do, sometimes do not, and sometimes invent things that were never there. Throwing more AI at the problems AI creates is not a solution. It is an illusion.
So we started from a blank slate. Tiden is a continuous verification system for agentic coding teams, built around one idea: an enforceable boundary.
Here is how it works:
1. Before a single line of code: Tiden builds a context graph of your system, linking features, requirements, tests and components. When a change comes in, the agent walks that map, finds what is affected and fills the gaps. The result is captured as intent, and that intent becomes the boundary the code must deliver against.
2. Clear context, not all the context: the coding agent gets exactly what is relevant, so it ships in fewer iterations and gets it right more often.
3. Before the PR, not after it: Tiden picks the functional and regression tests that matter for this change, based on requirements and risk, and runs them before a PR exists. Only when that quality gate passes can the agent open the PR.
4. After the merge: release evidence is assembled automatically, and production issues are tracked, feeding the next change.
In our own use, cycle time from starting a change to releasing it came down by almost half. Not because code was written faster, but because the most of rework loops simply stopped happening.
Tiden is built by the team behind Qase, so almost a decade of quality assurance thinking has gone into how verification should work when agents write the code.
We would love your honest feedback:
- Where does your agentic workflow break down today?
- What would you need to trust an agent's change without reading every line?
- What is missing?
You can start a 14-day trial at tiden.ai. I will be here all day answering questions.
Agents build. Tiden verifies. Your team decides.
Love how Tiden is addressing the gap where AI coding assistants and agents can generate 1000s of lines of code in seconds, but it’s very difficult to validate the quality of that code, especially at enterprise scale. Tiden with Claude looks like a killer combination.