Insights
Your agents make the same calls over and over. They re-decide things you already settled, re-introduce bugs you already fixed, and forget the conventions your project grew into.
Insights reads your indexed chat history and distills it into the durable signal — then feeds it back so your agents stop repeating themselves.
What you get
Open the 💡 Insights tab in the dashboard and hit Generate insights. Conare reads your past chats and sorts what it learns into three buckets:
| Bucket | What it captures |
|---|---|
| Decisions | What you decided and why — naming, architecture, product calls. |
| Bugs to avoid | Mistakes you already hit, kept as explicit guardrails. |
| Playbook | The conventions and dos that emerged across your work. |
Each insight shows how often it came up (seen 4×) and expands to reveal the original chat it came from, the reasoning behind it, and the concrete steps.
How it feeds back
The decisions don't just sit on the dashboard. Conare automatically pulls the relevant ones into every recall and search, so a new conversation starts already knowing what you decided last time — without you re-explaining it.
You generate insights once. From then on, your agents pick up the playbook on their own.
When to use it
- After a few weeks of real work, when patterns and decisions have piled up.
- When agents keep re-litigating choices you've already made.
- When you're onboarding a teammate (or a new agent) and want it to inherit your project's hard-won context.
Generation runs in the background and usually finishes in a couple of minutes. Re-run it anytime to fold in newer chats.
Insights is in beta. If you don't see the tab yet, it's being rolled out — index some chat history first and check back.