I Fed 3 Months of SAM.gov Listings Into an AI. Here's What Actually Came Back.

AI filters noise — it does not find contracts

The government contract discovery problem is not that opportunities are hidden. SAM.gov is a public database, updated daily, searchable without registration. The problem is volume.

A search for NAICS code 541611 (management consulting) on SAM.gov for a 30-day window returns hundreds of solicitations — full and open competition, set-aside, pre-solicitation, sources sought, and more. A new contractor without staff dedicated to business development cannot read and evaluate hundreds of solicitations per month while also running the business they just acquired.

AI tools — ChatGPT, Claude, Gemini — are good at processing structured data quickly and applying multi-criteria filters. They are not good at telling you whether you can win. They do not know your past performance, your team, your relationship with the contracting officer, or the 40 other vendors who are looking at the same opportunity. What they can do is reduce 200 opportunities to 5, so you can spend your 2 hours on analysis instead of filtering.

The 4 scoring criteria that eliminate 90% of irrelevant opportunities

Every opportunity gets scored on 4 criteria. Each is a binary question. An opportunity that fails any one criterion is eliminated from the shortlist.

Criterion 1: NAICS code match. Does the solicitation NAICS code match one of your registered SAM.gov NAICS codes? Opportunities in adjacent NAICS codes (one digit different) are borderline — include them for manual review.

Criterion 2: Contract size vs. bonding capacity. What is the estimated contract value? If you do not have bonding in place, any contract that requires a performance bond above your surety's limit is out. If you have no bonding at all, contracts requiring bonding are eliminated entirely.

Criterion 3: Incumbent presence. Is there an existing contractor on this work? SAM.gov often indicates whether a contract is a recompete (incumbent renewal) or a new requirement. Recompetes where the incumbent has a strong CPARS history are difficult to win without a price advantage or differentiated capability.

Criterion 4: Set-aside eligibility. Is the solicitation a set-aside for a certification you hold? Full and open competition solicitations are accessible but have the lowest win rates for new contractors. Set-aside solicitations narrow the competitive field significantly.

The actual prompt sequence — copy and use this

This workflow assumes you have exported SAM.gov search results as a CSV or copied the listing text into a document. Both methods work; CSV is faster for large volumes.

Step 1 — Set context: > "You are analyzing government contract opportunities for a small business that was recently acquired. The business is registered in SAM.gov under NAICS codes [list your codes]. It does not have small business certifications yet. It has a bonding capacity of $[amount] per project. I will give you a list of SAM.gov opportunities and I want you to score each one on 4 criteria and give me a shortlist."

Step 2 — Define the scoring criteria: > "Score each opportunity as Pass or Fail on: (1) NAICS code match — Pass if it matches one of our registered codes exactly, or a partial match if it differs by one digit. (2) Contract size — Pass if estimated value is under $[your bonding limit]. (3) Incumbent — Pass if no incumbent is listed or the listing says 'new requirement.' (4) Set-aside type — Pass if it's full and open competition (any contractor can bid) or a set-aside type we'll specify separately. Return only the opportunities that Pass all 4 criteria. For each, give me one sentence on why it passed."

Step 3 — Paste the data and run.

Expect 85%–95% of opportunities to be eliminated at this step.

10 sample SAM.gov opportunities through the filter

Of 10 opportunities: 4 shortlisted, 5 eliminated, 1 sent to manual review. The AI triage took approximately 8 minutes.

What most articles get wrong: AI is a triage tool, not a business development tool

SAM.gov AI content tends to fall into two categories: breathless coverage of AI's potential to "transform government contracting" and equally breathless warnings about AI replacing business development professionals.

Both miss the practical point. For a new contractor without past performance, without agency relationships, and without a dedicated BD team, the problem is not strategy — it is time. Reading 200 SAM.gov opportunities per month is 20–30 hours of work that produces a 5-item shortlist that an experienced contractor could generate in 2 hours with a simple filter.

AI compresses that to 2 hours for anyone. It does not make you more competitive in the proposal phase. It does not substitute for past performance. It does not tell you whether the contracting officer has already pre-selected a vendor. It is a screening tool — no more, no less.

Once you have a shortlist, the work of winning a contract is entirely human: understanding the requirement, building a differentiating technical approach, and writing a proposal that the evaluation team scores above the competition. For that part, see AI proposal writing.

Setting up the 2-hour/week workflow

The workflow requires two one-time setup steps and two weekly recurring steps.

One-time setup 1: SAM.gov search saved filters. Create a search with your NAICS codes, a dollar range under your bonding limit, and the opportunity type set to "Solicitation." Save the search as a custom alert with daily email delivery. Approximately 30 minutes.

One-time setup 2: AI triage prompt template. Build the 3-step prompt sequence above into a reusable document. Save it in Google Docs or Notion. Approximately 30 minutes.

Weekly recurring: Monday morning SAM.gov export. Export the prior week's opportunities that matched your saved search to a CSV. 15 minutes.

Weekly recurring: AI triage run. Paste the CSV into your AI tool of choice, run the prompt sequence, review the shortlist. 30–45 minutes.

Total weekly time after setup: approximately 60 minutes. The remaining time goes to analyzing the shortlisted opportunities and deciding which to bid.

Frequently Asked Questions