Is the fit score AI or a formula?

The 0–100 fit score is a deterministic formula — the same profile and opportunity always produce the same score, and every score shows its factor breakdown. AI is used only to write narrative analysis (like the fit-assessment paragraph in a brief), never to compute the score or fabricate requirements.

Where does AI get used, exactly?

AI rewrites the narrative portions of a bid/no-bid brief and, for Capture Plus, enriches proposal-template narrative from a solicitation you provide. The compliance matrix rows and the fit score are always deterministic; the AI never invents a requirement or a score.

Which AI model does GovConAgent use?

The model depends on your plan: a fast model on Free/Starter, a stronger model on Pro, and the most capable model on Capture Plus. Model choice affects the quality of narrative writing, not the underlying facts or scoring.

How GovConAgent works

GovConAgent turns the noise of SAM.gov into a short, ranked list you can act on. It does that with a transparent scoring model and a deliberate line between what is computed deterministically and what is written by AI. This page explains both.

1. Matching & scoring

When you build a company profile, GovConAgent scores active SAM.gov solicitations against it and keeps the best-fit results. Each opportunity receives a deterministic 0–100 fit score built from weighted factors that sum to 100, with a penalty for disqualifying signals. Every score is explainable — you can see which factors helped and which hurt.

Factor	What it measures
NAICS alignment	Exact and adjacent NAICS-code matches to your registered codes.
Capability keywords	Overlap between the notice's scope and your stated capabilities.
Set-aside / certification	Whether the notice's set-aside may align with your certifications (to verify, never asserted).
Target agency	Whether the buying agency is one you target.
PSC	Product/Service Code alignment.
Past performance	Relevance of your stated past performance to the scope.
Deadline feasibility	Response time remaining vs. your stated capacity.
Geography	Place-of-performance fit.
Exclusion penalty	A negative adjustment when disqualifying keywords appear.

2. Where AI is used

Deterministic (no AI)

• The 0–100 fit score and factor breakdown
• Compliance-matrix requirement rows
• Proposal-template structure (Section-L order)

AI-assisted (drafting)

• Narrative in the bid/no-bid brief
• Proposal-template narrative (Capture Plus, from a solicitation you provide)
• Answers in the assistant chat

The AI never fabricates a requirement or a score. Facts extracted from the notice are always shown separately from AI analysis, and description-derived rows are flagged for you to verify against the full solicitation.

3. Human review is required

AI-generated summaries, analyses, and drafts should be reviewed by qualified humans before use in official submissions. GovConAgent does not guarantee contract awards. Users remain responsible for final compliance and submission decisions.

Want to know where the data comes from? See data sources.