Hey Product Hunt! I'm Aayush, the maker of Laya.

I built Laya because I was drowning. Not in work -- in notifications about work.

Every morning I'd open Slack to find a wall of unread messages. Then Jira with a stack of ticket updates. Gmail with threads about the same issues. Bitbucket with PRs related to those tickets. The worst part wasn't the volume -- it was the fragmentation. The same bug existed as a Jira ticket, a Slack thread (called something completely different), a PR on Bitbucket (named after the technical fix), and an email thread asking about the timeline.

I'd spent my morning just building context before I could make a single decision.

So I built Laya. The core idea is simple: by the time you look at a notification, the research should already be done. Laya intercepts events from your tools, links related items across platforms using semantic matching, routes them through specialized AI personas, and presents you with Action Cards -- approve or dismiss, one decision, back to work.

A few things I'm particularly proud of:

- It's truly local-first. Your data stays on your machine. SQLite for structured data, ChromaDB for vector search, OS keychain for API keys. The only external calls are to your chosen LLM provider.
- It learns from you. When you correct a classification (wrong priority, wrong persona), Laya extracts generalizable rules and applies them to future events automatically.
- Context Association actually works. "BUG-1234" in Jira gets automatically linked to the Slack thread where your teammate called it "the payment thing" and the PR titled "fix: null check in payment handler." Three layers: explicit cross-references, semantic similarity, and LLM verification.

It's open source and free. I'd love your feedback -- especially on what integrations and features would make this useful for your workflow.

GitHub: https://github.com/aayushch/laya

Laya

AI command centre - Turn tool overload into one calm feed

AI command centre - Turn tool overload into one calm feed