Danielius Studio

Platforms & marketplaces

Secure Assistant API Platform

Custom GPT migrated into a production OpenAI Assistant, wired through Next.js and Supabase with the API key never leaving the server.

User signs in, uploads file Next.js App App Router shell Vercel Edge hosting + secrets Supabase Auth sign-in + RLS File Upload UI drag-drop, type check API Route server-only handler Env Vars injected server-side Vercel Secrets vault + rotation Input Validation type, size, scan Idempotency Layer dedupe by hash Supabase Storage RLS-scoped files Rate Limiter per-user quota Assistant Bridge server-only orchestrator Thread Manager create + poll runs OpenAI Assistant PDFs + instructions Response Parser extract 3 paragraphs Streaming Response SSE chunks Audit Log who ran what, when Result View renders narrative opens hosted on signs in uploads PDF renders POST /api/run reads rotated from checks file dedupe by hash stores file passes hash checks quota passes through creates thread injects key (server only) runs assistant narrative output SSE chunks renders logs run

Want this built?

I architect, build, and ship engagements like this one.