Incident reports from the autonomous agent
Detailed incident report of slow query
Autonomous Claude update based on incident report

Optimera

Inspiration

Spending countless hours staring at logs and telemetry in order to find out where that one malicious query is hiding is something we have personally wasted many hours on. We don't want to be limited by the fear of potentially locking the production website; we want to push features. Developers are often paralyzed by this "Production Fear" We built Optimera to turn telemetry from a "diagnostic tool" into an "automated cure."

What it does

Optimera is an AI powered autonomous system, that continuously listens to your application's logs, providing actionable insights and solutions, without requiring coding knowledge. It doesn't just show you graphs, it fixes your source code.

Optimera lets you focus on what's important; velocity and providing features for your users - all without scaling bottlenecks, technical depth, or infrastructure complexity holding you back.

How we built it

The main agent loop continuously polls an Open Telemetry source to find problematic areas of the code. Upon finding an actionable item, Claude will create a pull request with a potential solution, visualizing it in our dashboard. We monkey patch common libraries in order to provide as granular telemetry as possible without requiring user intervention.

Challenges we ran into

Open Telemetry is a powerful protocol, but with many gotchas. One of the hardest part of the project was monkey patching libraries in async environments.

Accomplishments that we're proud of

The quality of product we managed to create under such a short time and the potential for further development.

What's next for Optimera

We will continue to explore both vertically and horizontally. How many libraries are possible to monkey patch? Where can our product reach the most amount of people?