We believe in building in public. This is our first monthly update, covering what we shipped in January 2026.

What We Shipped

Ottly Chat (v1.0)

Our AI assistant went live with full support for:

Web search with real-time results and citations
Multi-model support (GPT-4o, Gemini 2.5 Flash, Claude Sonnet)
Code generation with syntax highlighting
Document and image analysis
Conversation history with search

Infrastructure

Deployed on AWS ECS with automated rollback on failed health checks
Set up CI/CD pipeline with integration testing before every production deploy
Configured Cloudflare DNS with automatic SSL for all subdomains

Developer Experience

Established our monorepo structure with independent git repositories per service
Set up pre-push hooks that require passing builds before code reaches remote
Created comprehensive CLAUDE.md documentation for AI-assisted development

What We Learned

Keep the architecture simple. We started with a microservices-heavy design and quickly simplified to three core services: API server (Go), frontend (Next.js), and worker (Python). Complexity should be earned, not assumed.

Multi-model support matters from day one. Adding provider abstraction after the fact is painful. We are glad we did it early.

What is Coming in February

Ottly Desk (AI journalism workspace) launch
Workflow automation MVP (later became Ottly Automate)
Improved streaming performance for long AI responses

Thanks for following along. See you next month.