Federal Capability Search · Company Database

How do you find every contractor that’s actually done this work — not just the ones who say they can?

Type the work the way you’d actually describe it — “systems engineering support for fighter aircraft sustainment programs” — and the platform reads back the companies with the strongest federal evidence behind that description. No NAICS code to guess and no capability statement to take on faith: 3,047,746 demonstrated capabilities, read off 1,011,617 federal awards, ranked by how many of those awards there are and what they were worth.

See it work ↓

Included starting with the Emerging plan. There are two kinds of number on this page and we label every one: the corpus figures just below are real, measured against the platform’s own published figures. The product screens further down are real captures of the running platform, over a fictional company set that is safe to publish.

What The Corpus Shows

3,047,746 demonstrated capabilities, read off 1,011,617 federal awards.

Every one of those capabilities is a company that demonstrably performed that work — not a NAICS code it happens to hold, and not a line from its own capability statement. Read from the government’s own record and never asserted without the award that proves it. Measured 4 September 2026.

3,047,746
Demonstrated capabilities on file
1,011,617
Federal awards behind them
55,690
Companies profiled
What Counts As A Match
One capability, demonstrated by one company, on one federal award — the platform’s smallest unit of evidence, never a rolled-up estimate.
Read from the government’s own award record — FPDS descriptions and, where a Performance Work Statement or similar scope document was available, the document itself — with AI assistance (Claude, by Anthropic). Every count and dollar behind it is then computed from the federal record.
Every capability sits on a named evidence ladder — document-verified outranks demonstrated, which outranks contracted-only, which outranks coded-only — so a stronger claim always wins, and on the company’s own record every capability shows which kind of evidence sits behind it.
Every company with a federal registration has a page here, even one we haven’t profiled yet — you’re told which kind you’re looking at, and you never dead-end on a name you already know exists.
What This Number Is Not
Not a company’s own marketing. A capability statement is written to win a proposal; this is read from what the government already paid for.
Not a percentage or a match score — see “Ranked by what they’ve actually done,” below.
Not a claim that a company can’t do something. Absent evidence means the platform hasn’t found it yet — never that the work wasn’t done.
Not blended across every company the same way. A small business’s five awards and a prime’s five hundred are shown as what they are, side by side, never smoothed into one score.

The Workflow

From a plain-English description to a ranked, evidence-backed list.

Every step below runs off the exact same search surface and evidence tables the rest of the platform reads — nothing here is a simplified demo path.

1Describe The Work, Get The Ranked Answer

Write it the way you’d say it out loud — and get back two facts you can check yourself, not a match score.

No NAICS code to guess and no boolean syntax to learn. Type the scope the way a program manager would describe it, and the platform understands related terms on its own. Quote a phrase only when you need an exact match — a program name, usually; a capability description is almost always better left unquoted. Add a second, narrower term in the optional Secondary box to see who demonstrates both without losing everyone who only demonstrates the first.

Most search tools hand you a single invented number and ask you to trust it. This one shows the two facts that actually decide the order — how many federal awards demonstrate the capability, and what they were worth — right on the row, named after the exact words you searched. Highest award count leads; matched dollars break the tie. Add a Secondary term and the list doesn’t blend the two capabilities into one score: where the second term is worth grouping on, it splits into two labelled groups, Both first, then Primary only, each ordered the same way. Nobody who only matches the primary is dropped — just banded below. When grouping would tell you nothing — the second term matched nobody, or almost everybody — the list stays whole and says so on screen instead.

intelligent.win/company?mode=capability&capPrimary=Aviation+Logistics&capSecondary=Sensors

The Capability tab mid-search, running the platform's own worked example: a plain-English Primary term, a narrower Secondary beneath it, and the ranked answer below. Each row names the two facts the order is built from — how many federal awards demonstrate the term you searched, and what they were worth — and, where a company also demonstrates the Secondary, how many of those awards overlap. The list is split into labelled groups: Both first, then Primary only further down, each ordered the same way.

In this image: companies are fictional and dollar figures are illustrative. Agencies, buying offices and codes are the real federal record; any contract number shown is altered so it resolves to nothing.

Showing the top of the page — full capture 6,053px tall; expand for the rest of the ranked list and the Primary-only band.
2See The Real Scale Beside The Match

The match is one fact. The company’s actual federal footprint is another — shown, never blended.

Prime portfolio, contract count, average and largest award: real numbers pulled straight from the federal record, sitting beside the match, never folded into it. A capability match tells you they’ve done the work. The scale beside it tells you whether they can do it again, at the size you need — and the strip omits itself entirely on a company with no aggregate on file, rather than printing a fabricated zero.

Open any row and you land on that company’s own record, where the same evidence is laid out in full: every capability it has demonstrated, ranked by how much federal work sits behind each one.

intelligent.win/company/M4RD8ANF57XK

The company record behind a result row: the KPI strip across the top — prime portfolio, contract count, average and largest award — and below it every capability the company has demonstrated, ranked by the federal work behind each one, with the award count and dollars beside every bar. This is the same evidence the search ranks on, laid out in full for one company.

In this image: companies are fictional and dollar figures are illustrative. Agencies, buying offices and codes are the real federal record; any contract number shown is altered so it resolves to nothing.

Showing the top of the page — full capture 6,174px tall, expand for all of it.

What You Actually Get

Everything the capability search shows — and what it deliberately refuses to.

Two facts, not one invented number. Evidence, not a company’s own marketing.

In the Product
A plain-English capability search across the full company database — no NAICS code or taxonomy lookup required
Companies ranked by matched-award count, then matched dollars — the two facts that actually decide the order, shown on every row
Every match traces to the government’s own award record — FPDS descriptions and, where extracted, the actual Performance Work Statement — never a company’s own capability statement
A “Your Search” panel on every row, naming the exact capability set driving its count
Real federal-award scale — prime portfolio, contract count, average and largest award — shown beside the match, on the companies that have it on file
Deliberately Not Here
A composite match score, percentage, or letter grade — retired from this exact search; two visible facts replaced one invented number
A company’s own capability statement, marketing copy, or self-description anywhere in the ranking
A claim that a company is or isn’t a small business — SAM’s own on-file certification is shown, never our verdict
A ranking boosted by a vehicle a company doesn’t actually hold — vehicle position shows as a plain fact, never folded into a hidden score
Any conclusion about who should win. The evidence is assembled here; the team decision is still yours.

Stop reading award descriptions one at a time to find out who’s actually done the work.

Describe it once, in plain English, and see companies with real federal evidence behind it — ranked by what they’ve done, not a black-box score.

Or see how the whole platform works first →

Common Questions

Three things people ask before they trust a search like this.

Is this just a keyword search?

No — it understands related terms on its own, and every word in your description has to match somewhere in the evidence, not just one of them. Put a phrase in quotes only when you need the literal words together, which is usually the right call for a program name and usually the wrong call for a capability description — quoting too much narrows the search further than most people mean to.

Where does the “matched” evidence actually come from?

From the federal award record — FPDS descriptions and, where a Performance Work Statement or similar scope document was available, the document’s own language. Capabilities are read out of those government documents with AI assistance (Claude, by Anthropic); every count and every dollar behind them is then computed from the federal record — deterministic, and the same answer if you run it twice. None of it comes from a company’s own capability statement.

Why isn’t there a match score or percentage?

Because a single invented number hides more than it tells you. This search is ranked on two facts you can check yourself — how many federal awards demonstrate the capability, and what they were worth — shown right on the row instead of compressed into a score nobody outside the platform can verify. If a company shows no evidence for something, that means the platform hasn’t found it yet, never that the work wasn’t done.