Shipping Billing & AI Analytics in Under 3 Weeks

As the solo lead on billing and AI, we had to move fast without breaking things. The billing layer needed to support usage-based metering and proration; the AI chat needed streaming, tool use, and dynamic chart rendering (Recharts) from model output.

We used Vercel AI SDK for the chat and tool-calling flow, and kept billing logic in a dedicated service with idempotent webhooks. Shipped in under three weeks with zero critical bugs in the first month.