PromptPilot

The problem#

Prompt workflows become harder to manage when instructions, preferences and earlier results are scattered across different places.

What I built#

I built a Next.js application that brings prompt tools, account preferences and history into one workflow. Supabase supports authentication and persistence, while the AI layer can use a mock provider or OpenAI.

My contribution#

• Designed and implemented the application end to end.

• Built the interface and the prompt and developer workflows.

• Connected authentication, settings and history to Supabase.

• Implemented provider selection, fallback behavior and server-side request guards.

Behind the decisions#

Keep the AI provider replaceable#

The provider factory separates the workflow from the service that generates the response. Mock mode also supports development without a live API.

Make persistence a user choice#

Account preferences control history saving, while row-level security policies define data ownership.

What I learned#

Connecting an interface to authentication, persistence and an external AI service means thinking about failure states and account boundaries throughout the application.

What comes next#

• Complete the closed-beta checklist and verify the core flows in a configured deployment.

• Validate account isolation, provider fallback and release rollback.

• Gather feedback before treating the application as a released product.