Launched this week
Struct
AI agent that root-causes engineering alerts
691 followers
AI agent that root-causes engineering alerts
691 followers
Struct is an AI agent that root-causes engineering alerts using logs, metrics, traces, and code. Resolve incidents faster with a composable, customizable system that deploys in minutes and works with your existing DevOps workflows.
Coefficient.io
Struct
@kurtjheinrich Thanks Kurt! So glad we've been able to help you guys with your oncall process!
Congrats @deepan_m and @nimeshmc! We've been piloting Struct for the past couple months. It's been fun working alongside them and watching the product transform.
Struct
@deepan_m @william_namen thanks William! Your feedback has been so instrumental to how much better Struct has gotten the past few weeks
I'm a software developer who works with AI. Could you give me a specific example of how your software could help me? I've read through your site, but I don't quite understand what exactly you do.
Struct
@daniyar_abdukarimov Hi Daniyar! Of course. Many customers have a slack channel that they have alerts posted to when their monitoring tells them something might be wrong. We monitor that, investigating each one based on your code and the logs + metrics outputted in production and deliver an understanding of how it is impacting your customers and what specifically caused the issue. One example:
Alert comes in: out of memory crash occurred
Struct pulls the logs + metrics relevant to this alert and correlates it with what your code does
In 5 minutes, it posts a message back to the Slack thread with "Commit abcd1234 introduced a memory leak in the caching layer that increased memory usage over 30 minutes, eventually causing container restarts. 214 users received 5xx errors during container restarts."
I understand, thank you
Does Struct actually build a dependency graph across your infra or just pattern-match on the alert text? I've got microservices that can fail in three different systems depending on the blast radius, and I'm curious if it traces downstream impacts or just points to the loudest screaming dashboard.
Struct
@lliora Yes! It builds an internal understanding of how your services work together by utilizing your code and observability data. This isn't a simple surface level attempt at pattern matching known exceptions and alerted symptoms. Our agents trace through what actually happened in production, going iteratively through layers of services until they find definitive evidence of the root cause. Most of our customers have several services deployed in production with observability data in different places. We tie them all together.
This is exactly what on-call engineers need. The gap between "alert fires" and "understanding what actually broke" is where most incident response time gets wasted, and automating that correlation across logs, metrics, and traces is a huge win.
How does Struct handle alert deduplication when multiple monitors fire for the same underlying issue? Does it group them into a single investigation automatically?
Struct
@borrellr_ yes you got it! Our agent has access to the ongoing stream of alerts and is able to evaluate historical and ongoing patterns to pull together multiple signals that inform the actual underlying root cause.
Congrats on the launch! Curious how you'd position this vs the AI SRE features that observability platforms are starting to bake in natively (e.g. Better Stack's AI incident summaries). The key difference I'm guessing is that those are limited to the data within their own platform, while Struct can pull from logs + metrics + traces + code across multiple tools and actually trace the root cause end to end?
If that's the case, the composability angle is the real moat here, most real-world incidents span 3-4 tools and no single vendor sees the full picture. Nice work!
Struct
@maks_bilski the composability is a critical part of what makes this fantastic, as you said. It makes it possible to actually analyze issues across services, cloud boundaries, etc.
Other important pieces unique to our platform: we are able to utilize historical and ongoing alerts to evaluate patterns and better root cause the real underlying issues, and you can use our MCP to pull context from historical incident analyses to inform projects in claude code, cursor, etc.
Autumn
Great product that we’re really enjoying using. Congrats on the launch!!
Struct
@ay_ush We're so glad you guys are loving the product! Amazing customers also working on a great product ❤️