Case studies · Building with AI agents

Case studies: shipping AI products with agents

Three short write-ups from building Quantlys: how releases are gated, how model costs are held to a budget per build, and how a team of AI agents and human teammates is run. Each one covers the problem, the approach, what happened (including what broke), the lessons, and what I'd tell another team.

Context

What these are built on

Quantlys is a native app for Apple Silicon Macs. Inside it, Conclave runs a council of AI models and a PM, Architect, Developer and QA crew through four stages (Draft, Peer-rank, Arbiter and Ship) to plan an app, then build it, verify it and deploy it to your own GitHub, Vercel and Supabase accounts. The idea in one line: from idea to live in minutes, no one fixing code.

Apps made this way are on Built with Quantlys. For how Quantlys compares with other AI app builders, see Best Lovable alternatives. For the orchestration pattern underneath, read Multi-agent LLM orchestration, explained.