Proposal Generation
Generate Proposal kicks off an AI-assisted first draft: it reads the solicitation, plans out the required sections, writes each one, and reviews its own work before handing it back to you.
This produces a draft, not a submission-ready document. Every generated section is meant to be reviewed by a person before it goes out — the platform flags anything it isn’t confident about rather than guessing silently.
How it’s built
A LangGraph state machine runs parse -> compliance -> plan -> write -> review -> assemble, with conditional error edges. Every node persists to Postgres before handing off, so a run is resumable from any completed node, not just re-runnable from scratch.
- The review step checks generated text for unsupported factual claims against the section’s own cited sources — a section that fails this check is escalated for human review automatically.
- Regulatory clause citations are grounded against the real FAR/DFARS catalog — see Clause Library.
- Real generation requires the platform’s live-calls gate to be enabled; otherwise you’ll get clearly-labeled mock output, never invented content presented as real.