A product prototype: Yello!
I built Yello!’s initial working prototype myself using AI-assisted development. It made the phone service for kids something we could use and discuss.
I also owned product direction. A cross-functional team delivered the production beta.
Daily business briefings
I used Model Context Protocol (MCP) to connect AI to business data in Snowflake before tools like Claude Cowork were available to us. This let me build daily briefings before self-service analytics was available internally, bringing product and business signals into one place.
When something looked unusual, I used connected tools to investigate the underlying data and check the summary.
Workflow automation with n8n
For a device-financing pilot, I built an n8n workflow that used AI to process and rank application responses collected through SurveyMonkey. It supported the application-review process while we tested the product with customers.
This is another way I use AI in product development: building a working process around an experiment, then learning which steps need more automation as it grows.
Why I do the building myself
I want to understand the tools I make product decisions about. Building a prototype or a workflow lets me try an idea, see where it falls short, and ask more specific questions of the team.