Introducing Devin Auto-Triage: Your AI first-responder with long-term memory. Devin can monitor incoming bugs, alerts, and incidents, investigate them, and come back with context, next steps, or a PR.

Auto-Triage connects Devin to the systems where issues already show up: Slack, Linear, GitHub, schedules, webhooks, observability tools, tickets, alerts, and code. When something needs attention, Devin starts to investigate automatically.

Devin Auto-Triage gets smarter with every incident. Devin can remember prior investigations, deduplicate related reports, link threads and tickets, map team ownership areas, and keep track of recurring issues.

Teams like @modal are already using Devin Auto-Triage for incidents on their inference team. “Devin Automations feels like a step forward from other auto-triage tools we’ve tried. It monitors our channel, works with our codebase and observability stack, and comes back with useful investigation without us having to prompt it.” — Hari Subbaraj, Member of Technical Staff @ Modal

For a limited time, we're offering $200 in credits when you build your first automation. Try it out today:

We're using it already, and it's great! It's really helpful to have Devin kick off an investigation and return with context before a human has a chance to take a look.

This is insanely useful!!!

sweet

This is huge!!!

pretty sick

Auto-triage is the right shape — investigation memory is high-leverage. The thing that bit us building this for agents: forgetting is harder than remembering. A stale "we restarted the pod" fact poisons the next investigation worse than no memory. We ended up modeling forget as a first-class event (tombstone in the log, suppress at recall, keep audit trail) rather than a delete. Curious how you're handling decay for incidents that recur with new root causes.

Is it also feel snappier to use ?

auto-triage gets scary/useful at the permission edge. can teams set different scopes for read-only incident context, code investigation, and PR creation, or does one automation identity hold the whole bag?

how do I setup the pylon automation?

The alert queue is the part I’d watch. If Devin comes back with a PR, I want the incident log beside it so review doesn’t become archaeology.

Very cool!

“Long term memory” isn’t exactly solved with embeddings that squeeze in top-N semantic matches into the prompt.

triage is a smart first job for an agent. it's bounded and checkable. it either gathered the right context or it didn't, no taste required to grade it. the quiet part is that triage was also where on-call engineers learned the system. you got senior by doing the rung Devin now does. the role doesn't disappear. the on-ramp to it does.

The key design question for auto-triage is evidence quality. A useful first-responder should return more than “I think it is X”: 1. observed signals 2. ruled-out hypotheses 3. files/logs/commands inspected 4. confidence + uncertainty 5. safest next action Without that, long-term memory can preserve bad guesses too.

could be useful, but automated triage still needs human insight. context matters too much to leave solely to AI.

👀👀

cognition is great!

We’re getting closer to self-healing software systems 👀

First-response automation is where agents really shine.

kindly check this out guys

Why all time decline my card some time accept some time not

Devin is still a thing?

@dabit3 are you still doing giveaways

what you are using for agents to web search and scrape ? do you want work together? fastest, lightest single 6mb binary, opencore <3

The come back with next steps or a PR line is doing a lot of work