# AI

How Paul uses AI on shipped work, and which AI systems he has built. This is not an AI product company page.

## How it is used

- Product exploration: prototype features to compare design and architecture before committing
- Structure and boundaries: documentation and standards so code and design stay consistent as a project scales
- Debugging: root-cause tracing, error paths, edge cases
- Acceleration: repetitive coding so planning and review stay human; example: AI-assisted development on Adomo with Convex

## Systems shipped

- Conversational AI: workflow generation from chat, multi-user streaming, tool calls, voice input
- Agent orchestration: approval gates before execution, durable background runs, MCP tool registry, meeting integrations
- Data tooling: query builders for non-technical users, visualization, transformation
- Classification pipelines: categorization, extraction, structured outputs, semantic retrieval

Primary production example: Adomo ([https://adomo.ai](https://adomo.ai), [https://app.adomo.ai/login](https://app.adomo.ai/login)).