DevPulse

• 

  AI incident analysis in action

• 

  Real-time incident investigation and recommendations

Inspiration

We were inspired by the challenges solo developers and small teams face when production systems fail. Without a dedicated SRE team, diagnosing incidents can be stressful and time-consuming. DevPulse was created to automate the first 20 minutes of incident response, helping developers resolve issues faster while staying in control.

What it does

DevPulse automatically investigates production incidents, identifies likely root causes, and sends recommendations to developers through Slack for approval.

How we built it

We used Gemini 3, Google Cloud Agent Builder, Dynatrace MCP, GitHub APIs, Slack, Next.js, and Firestore.

Challenges we ran into

Correlating observability data with deployment history and ensuring safe, transparent automation

Accomplishments that we're proud of

Building an agent that performs real incident triage and recommends actions instead of just answering questions.

What we learned

Building reliable AI agents is about more than just intelligence. it requires the right data, clear reasoning, and human oversight. We learned how observability, automation, and AI can work together to turn complex incident response into a faster and more manageable process.

What's next for DevPulse

Adding predictive incident detection, automated post-mortems, and support for more cloud and communication platforms.

Built With

• dynatrace-mcp
• dynatrace-oneagent
• express.js
• firestore
• gemini-3
• github-api
• google-cloud-agent-builder
• google-cloud-functions
• google-cloud-run
• mcp
• next.js
• node.js
• slack-api
• typescript