Featured Project
SeederKit
A Rails developer tool for creating realistic seed scenarios. SeederKit explores how AI can support product workflows when it is constrained by schema awareness, validation, and predictable execution.
Schema-Aware Planning
Use Rails schema and relationship context to constrain scenario generation.
Structured Outputs
Treat model output as data that must be validated before execution.
Deterministic Execution
Generate plans probabilistically, then execute through predictable application code.
Rails-Native Tooling
Build around Rails conventions, ActiveRecord models, validations, and workflows.
The Thesis
Seed data tooling breaks down when product teams need believable, connected scenarios instead of isolated fake records. SeederKit explores a more useful path: let AI assist with intent and planning, but keep validation and execution inside predictable Rails code.
What The Devlog Will Cover
- Why scenario generation is different from fake data generation.
- How schema readers and domain graphs shape safer AI workflows.
- Why structured validation matters before anything touches a database.
- How predictable execution keeps AI-assisted tooling debuggable.