Relic Lab
Structured generative AI for speculative product dossiers
Generative AI application that turns a present-day problem into a future artifact dossier with typed outputs, a local vault, export workflows, demo mode, and request-level tests.
Visual evidence
System visualization
Case study
Problem, Build, Outcome
Problem
Most generative AI experiences stop at an unstructured chat response, making creative output difficult to reuse, compare, validate, or export.
Build
Built a Node.js and Express application around the OpenAI Responses API with Zod-validated artifact schemas, local persistence, deterministic demo data, and test coverage for generation failure modes.
Outcome
Transforms an open-ended problem into a repeatable, reviewable artifact dossier that can be saved, deduplicated, copied, and exported as Markdown or JSON.
Architecture
System Flow
Architecture Notes
- API layer turns an unstructured prompt into an OpenAI Responses request with strict output requirements.
- Validation layer parses the output against Zod schemas before anything reaches storage.
- Persistence layer manages saved dossiers, deduplication, and file exports.
Engineering choices
Technical Decisions
Used OpenAI Responses with structured schemas instead of raw completion text to guarantee the output structure.
Added deterministic demo mode so the application remains fully testable and demonstrable even without API credentials.
Added request-level validation tests to verify that invalid responses fail predictably instead of poisoning state.
Evidence
Proof Points
Zod-validated artifact schemas for structured generation
Local vault with duplication prevention and Markdown/JSON export
Integration tests verifying parsing, schema guardrails, and demo fallbacks
Next read