What this is
The internal platform we built and use to process the RFIs, RFQs, RFPs, and grant solicitations we respond to. It ingests the source solicitation document, extracts metadata and requirements with an LLM, drives a status workflow from intake to delivery, and emits each completed contract solution as a standalone app — independently deployable, no cross-dependencies, ready to be moved to its own repo when the client takes ownership.
The archival-grade document-digitization platform described separately on this page was built and lives as one of these standalone apps.
Why we built it
Government and enterprise contracting work has a long, painful tail: solicitations arrive as PDFs full of mandatory requirements, exhibits, amendments, and Q&A responses; teams ad-hoc respond in Word and email; nothing gets reused; trade-secret marking and public-records discipline are bolted on at the last minute (or forgotten). We wanted to operationalize our own response process — and prove, on ourselves first, that the right software can compress the soft costs of bidding without compromising the rigor.
It also forced us to confront the IP-safety, public-records, and audit-trail design choices we now apply to client systems by default.
What it does
- Ingests the source solicitation document (PDF) and runs LLM-powered metadata extraction at intake — document type, issuing organization, due date, summary, key requirements, category tags.
- Drives an enforced status workflow from intake through requirements, planning, review, building, delivery, and completion — transitions are guarded state changes, not labels.
- Emits every contract solution as its own standalone app under
apps/{solution-name}/with no cross-dependency on the platform — built to be moved to a client repo at handoff. - Treats trade-secret marking and FOIA-aware document handling as first-class concerns in the workflow, not last-minute additions.
- Dual-stores files: an object store for canonical artifacts, plus a git-tracked mirror of the source documents themselves.
Built with
| Layer | Stack |
|---|---|
| Backend | Python, FastAPI (async ORM with migrations) |
| Frontend | Vite + React + TypeScript + Tailwind CSS |
| Database | PostgreSQL |
| Object storage | S3-compatible |
| LLM (intake extraction) | Leading frontier LLM |
| Reproducibility | Docker Compose (clean clone → running stack) |
What this means for you
If you operate a workflow where domain-specific documents come in as PDFs and need to be triaged, scoped, planned, and executed — RFx response, grant administration, regulatory submission tracking, complex insurance intake, customer-onboarding ops with heavy paperwork — we can:
- Build an internal operating system tailored to your workflow, with status-enforced state transitions and LLM-powered metadata extraction at intake.
- Architect a modular app-per-solution model so each engagement / case / grant ends up as a standalone, transferable artifact.
- Engineer trade-secret, FOIA, and audit-trail discipline in from the start.
- Deliver as code your team owns, with the same modern open-source stack we use ourselves.
Want to see the platform run live on a sample solicitation? Contact us.