CortexCI

💡 Inspiration

We wanted to eliminate the manual grind of writing and maintaining tests during rapid development. Our goal was to let AI understand code changes, test them autonomously, and validate results directly in production-like infrastructure — creating a self-sufficient DevOps assistant.

⸻

⚙️ What it does

CortexCI detects code changes, analyzes their impact using OpenAI GPT-4o, generates functional and database tests, executes them on ClickHouse Cloud, and streams all logs and metrics to Datadog for full observability. It acts as an autonomous QA engineer that continuously monitors, tests, and reports — without human intervention.

⸻

🏗️ How we built it

We built CortexCI in Node.js, integrating: • OpenAI for reasoning, test generation, and log analysis • ClickHouse Cloud for running live SQL and data validation • Datadog for real-time logs, monitoring, and analysis Each code push triggers a full pipeline — change detection → AI test generation → functional execution → log feedback loop — creating a closed, intelligent CI system.

⸻

🚧 Challenges we ran into • Balancing AI creativity with consistent JSON outputs from GPT-4o • Handling Datadog API rate limits and formatting structured logs • Safely executing live database queries on ClickHouse during test runs • Designing readable console output that mimics a mini dashboard

⸻

🏆 Accomplishments that we’re proud of • Fully autonomous testing pipeline that runs without manual setup • Real-time Datadog integration with AI-driven log analysis • Functional test execution against actual infrastructure, not mocks • Achieved a working prototype that simulates real CI/CD intelligence

⸻

📚 What we learned

We learned how powerful it is to embed LLMs into DevOps workflows, not just for chat but for reasoning, testing, and decision-making. We also saw how combining AI with real monitoring tools like Datadog enables explainable, data-backed automation.

⸻

🚀 What’s next for CortexCI

Next, we’ll integrate GitHub and AWS Lambda triggers, add self-healing test generation, and expand AI agents to auto-remediate detected issues. The vision is a fully autonomous CI/CD layer — CortexCI, your AI DevOps co-pilot.

Built With

• clickhouse
• datadog
• node.js
• openai