Solutions

Federal contract search, with the attachments already read.

samsearch is an AI platform for federal contract search. It indexes every notice posted to SAM.gov. It reads the attachments and amendments that come with them. It grades each opportunity against your company profile. And when you decide to bid, it drafts the response.

Search in plain English. Get a summary instead of a 200-page package. Ask the solicitation a question and get a cited answer. Be told the moment an amendment changes something.

Book a Demo

The attachments, read before you open them

The notice is the smallest part of a federal opportunity. Scope, deliverables, security requirements and evaluation criteria live in the package.

AttachmentsRead on arrival5 files · 214 pages

PDFSolicitation W912DR26R001468 ppRead
PDFAttachment 1, Statement of work94 ppRead
PDFAttachment 2, Wage determination11 ppRead
XLSXAttachment 3, Pricing schedule4 ppRead
PDFAmendment 000237 ppRead

What it says

ScopeFacilities O&M across 14 buildings, 1.2M sq ft

EligibilityTotal small business set-aside, NAICS 561210

EvaluationTechnical significantly more important than price

WatchMandatory site visit added by Amendment 0002

214 pages, summarised before anyone opened the package.

The whole package, not the notice

The solicitation. The statement of work. Wage determinations, specification sheets, pricing schedules and every amendment. All of it is read on arrival. Qualifying an opportunity becomes a paragraph, not a folder.

A summary a capture lead can act on

Scope, eligibility, evaluation approach, key dates and anything unusual. All in plain English. This is not an abstract of the notice. It is a read of the whole package, including the parts the notice never mentions.

Sections L and M, shredded

Every requirement is pulled out of the statement of work. Each one is mapped to the Section L instructions and the Section M evaluation factors. The result is a working compliance matrix, and every line cites its page.

FAR Part 15, contracting by negotiation

Verifiable, not just plausible

Every extraction points back to its document and page. You check a citation instead of trusting a summary. That is the only way an AI read is safe to use on a bid decision.

Ask the solicitation anything

Sammy reads the notice and every attachment, then answers plain-English questions with the source and the page beside the answer.

SammyReading 5 attachmentsEvery answer cited

You

Do we meet the eligibility criteria, and what is the page limit on Volume I?

Sammy

Yes. This is a total small business set-aside under NAICS 561210 and your profile is registered small under that code. Volume I is capped at 30 pages, raised from 25 by Amendment 0002.

Section L p.41Attachment 1 p.12Amendment 0002 p.3

Amendment 0002, p.3

“Paragraph L.3.1 is revised to read: the Technical Volume shall not exceed thirty (30) pages, excluding…”

Questions a matrix cannot answer

What is actually in scope here
Do we meet the eligibility criteria
What is the incumbent being paid
Is a facility clearance required
What does Section M weight most

Answered against the package

The solicitation and every attachment
Amendments, at the current version
Your company profile and past performance
Prior awards on the same requirement
Nothing invented from the model's own memory

With the receipts

Document name and page on every answer
The phrase the answer came from, highlighted
A link straight into the source
An honest 'not stated' when it is not stated

Sammy also produces the longer artefacts. A bid or no-bid assessment against your capabilities and past performance. A risk read on the requirement. A client-ready RFP analysis of the kind a consultant spends a day assembling.

Every notice graded against your profile

Fill in a company profile once. Everything published afterwards is read and scored against it, continuously, whether or not you searched for it.

Scored with its reasoning

Each recommendation carries a fit score and a plain-English reason. The reason cites the passage it was scored from, so you can disagree with it. A score with no reasoning is a number nobody trusts twice.

Eligibility applied first

Set-aside code, NAICS, PSC, agency, place of performance and response window are applied before scoring. Work you cannot bid never reaches the feed. Neither does work that closes tomorrow.

Graded on what you have done

Past performance, certifications, contract vehicles, and the work you say you want more of. The strongest matches are usually the ones nobody thought to search for.

Weak profiles are told so

A thin profile produces thin recommendations. samsearch says which parts of your profile are holding the matching back, rather than quietly returning noise.

Search the way you describe your own business

Contracting officers write notices to be legally sufficient, not searchable. The words that identify the work are usually in the attachments.

Natural language search

Type what your company does. No Boolean strings. No procurement vocabulary. No guessing which of a dozen phrasings a contracting officer picked. The AI knows the acronyms, so you do not have to speak contracting-ese.

Semantic and keyword, together

Semantic search matches meaning. A query describing your capability returns notices that share no keywords with it. Keyword search is still there for a solicitation number, a NAICS code or an agency name. That is what it is genuinely good at.

Search inside the attachments

The requirement that identifies an opportunity is often only in the statement of work. samsearch indexes the package, not just the notice. So search reaches it.

Filters that match the domain

Set-aside type, NAICS, PSC, agency and sub-agency. Contracting office, place of performance, contract value, notice type and due-date window. All applied before anything reaches you.

Sources sought and RFIs included

Pre-solicitation notices, sources sought and intent-to-sole-source notices are indexed alongside solicitations. These are often worth more. The requirement can still be shaped.

SAM.gov contract opportunities

Dismiss what you never want to see

Dismissing a result teaches the feed. It is scoped per user and per company profile. One person's clean-up never narrows anybody else's search.

Saved searches that qualify before they email you

A keyword alert is a firehose with your name on it. Qualification happens before the email is sent, not in your inbox afterwards.

Save any search as a monitor
Keywords, semantic description, filters, or your whole company profile. Most teams run two or three with different source sets rather than one that returns everything.
New notices are matched and read
Everything published against that search is picked up continuously and its attachments processed, not just its title.
AI qualifies the matches
Each candidate is graded against your profile and the weak ones are dropped, so the digest is a shortlist rather than everything that contained your keyword.
You get a digest you can act on
Daily, weekly or real-time, with the score and the reasoning on each item. Links go to the opportunity itself, so acting on one is a click rather than a search.

Amendments are tracked, and you are told

The most expensive thing in federal contracting is finding out late that the solicitation changed.

SAM.gov will post an amendment. It will not tell you what changed in it. It will not tell your team either. Deadlines move. Page limits change. Site visits become mandatory. Evaluation factors get reweighted, often after people have started writing.

What happens on an amendment

The whole package is read again
The new version is diffed against the old
Changed fields are listed, not buried
Requirements are re-extracted
Tasks carrying old dates move with them
Everyone watching the pursuit is notified

Fields compared

Proposal and question deadlines
Page limits and format instructions
Evaluation factors and their weighting
Scope, quantities and delivery terms
Set-aside status and eligibility
Site visit and pre-proposal requirements

From found to submitted, without leaving

Search is the front of a longer job. Everything a qualified opportunity needs afterwards is in the same place.

Proposal volumes, drafted

A compliant first draft, structured to Section L. It addresses each extracted requirement. It is written from your approved content and your past performance, not invented. The win themes are still yours.

Grounded in two sources only

The solicitation itself, and your own approved content. Nothing else. Every claim traces back to a document you own or a requirement in the RFP. That is what makes generated text usable on a real bid.

The matrix stays live

The compliance matrix tracks against the draft. What is still unanswered is visible weeks out, not on the final weekend.

The forms as well as the writing

SF 33 and SF 1449. Representations and certifications. The solicitation-specific attachments. The part of a bid package that has nothing to do with writing gets filled in too.

One pipeline for the team

Qualified opportunities move onto a shared board. Owners, tasks, due dates, documents and comments live on the pursuit. A bid is worked by a team, not carried in one person's head.

See Journey Hub

Reporting for whoever asks

Weighted pipeline by stage. Win and loss. When awards are expected to land. All built from the opportunities the team is already working, not a spreadsheet someone refreshes.

What is covered, and what is on every record

Federal is one market of ten

Most teams that search federal work lose bids in the markets they are not watching. Every source below is searched from the same box. Everything found lands on one pipeline.

State, local and education

Thousands of separate portals in one feed, with addenda tracked the same way amendments are.

DLA DIBBS

Defence supply solicitations, with the award history that tells you what the part sells for.

GSA eBuy

Schedule RFQs, graded against the SINs you hold by the same recommendation engine.

Agency forecasts

Work that has not been solicited yet, while the requirement can still be shaped.

Recompetes

Contracts running out, surfaced far enough ahead to do something about it.

Subcontracting

Primes carrying mandated small-business goals they are obliged to meet.

Federal grants

Assistance funding, with eligibility and award ceilings extracted on every record.

SBIR and STTR

Research topics matched to your technology, on a ladder that ends sole-source eligible.

Journey Hub

The CRM everything you pursue lands on: stages, tasks, documents and reporting.

Frequently Asked Questions

What is federal contract search?

Federal contract search is finding the government contract opportunities your company can bid on. Federal opportunities above $25,000 are published on SAM.gov, the official System for Award Management, which is free to search. The difficulty is not access, it is volume and language: tens of thousands of active notices written to be legally sufficient rather than searchable, with the detail that decides whether you can bid sitting in attachments rather than in the notice itself.

Does samsearch replace SAM.gov?

No. SAM.gov is the system of record and you still register there, download the official package and submit through it. samsearch is the search and intelligence layer over it: natural language search across every notice, attachments read and summarised, requirements extracted into a compliance matrix, opportunities graded against your profile, amendments tracked and alerted, and proposal drafting when you bid. If one keyword search a week is all you need, SAM.gov is free and sufficient.

How does AI-powered federal contract search work?

You describe what your company does in plain English rather than constructing a keyword query. Semantic search matches the meaning of your description against the full text of every notice and its attachments, so it returns relevant opportunities that share no keywords with what you typed. Each result arrives with an AI summary of the package, an extracted requirement list, and a fit score against your company profile with the reasoning cited.

Can AI read the attachments, not just the notice?

It has to, because that is where the binding detail is. Scope, deliverables, security requirements, wage determinations and the evaluation criteria in Sections L and M are almost always in the package rather than the notice. samsearch processes the whole package on arrival, including amendments, extracts the requirements, and points every extraction back to the document and page it came from so a human can verify it.

How accurate are the AI contract summaries?

The summaries are extractive rather than inventive: they are built from the official documents and every claim points back to the page it came from, so accuracy is checkable rather than something you have to take on trust. That is the important property. Where the package does not state something, the summary says so instead of filling the gap, and anything load-bearing on a bid decision should still be confirmed against the citation.

What is Sammy?

Sammy is the AI agent inside samsearch. It reads a solicitation and every attachment, then answers plain-English questions about scope, eligibility, deadlines, evaluation criteria and wage rates, with each answer linked to the source document and the phrase it came from. It also produces longer artefacts: bid or no-bid assessments against your capabilities, risk reads on a requirement, and client-ready RFP analyses.

Does samsearch track solicitation amendments?

Yes, and it is the reason many teams keep it. Any solicitation you track is watched from then on. When an amendment posts, samsearch re-reads the whole package, diffs it against the previous version, and shows what actually changed: deadlines, page limits, evaluation factors, scope, set-aside status, added or removed requirements. Requirements are re-extracted, tasks carrying the old dates move with them, and everyone watching the pursuit is notified.

Will I get alerted when a new federal opportunity matches?

Yes. Any search can be saved as a monitor, and your company profile is matched against new notices continuously. Alerts arrive daily, weekly or in real time by email, and they are qualified with AI before they are sent rather than being a raw keyword dump, so the digest is a shortlist with a score and a reason on each item.

How do the AI recommendations work?

You fill in a company profile once: capabilities, past performance, certifications, set-aside status, NAICS codes and the work you want more of. Every new notice is then read and graded against it, whether or not you searched for it. Each recommendation comes with a fit score and a plain-English explanation citing the passage it was scored from. A thin profile produces thin recommendations, and samsearch tells you which parts of the profile are holding the matching back.

Can samsearch write my proposal?

It produces a compliant, grounded first draft: structured to Section L, addressing each extracted requirement, and written from your approved content and past performance rather than invented. What it cannot supply is the win themes and the discriminators, which come from knowing the customer. Generation is grounded in two sources and nothing else, the solicitation and your own approved content, so every claim traces back to something checkable.

Which federal agencies are covered?

Every agency that posts to SAM.gov, which is effectively all of them: the Department of Defense and its components, DHS, GSA, NASA, Energy, Health and Human Services, Veterans Affairs, State and the civilian agencies. Beyond SAM.gov, samsearch also covers DLA DIBBS, GSA eBuy and PIEE, which carry federal work that never appears on SAM.gov at all.

How quickly do new federal opportunities appear?

New notices are ingested continuously and are searchable within minutes of posting, with the attachments processed shortly after. That matters most on short response windows, where a day spent finding a notice is a meaningful share of the time available to respond to it.

Is samsearch useful for small businesses and set-asides?

Yes, and set-aside filtering is applied before anything else. You can restrict the feed to 8(a), HUBZone, WOSB, EDWOSB, SDVOSB and small-business set-asides, or search full and open competition alongside them. Small teams get the most from the qualification layer, because reading a hundred-page package only to reject it is the cost they can least afford.

Can I search sources sought and pre-solicitation notices?

Yes. Sources sought, requests for information, pre-solicitation notices and intent-to-sole-source notices are all indexed and searchable. These are frequently more valuable than the solicitation itself: a sources sought is the government asking whether the market can do the work, which is the last point at which the requirement and the set-aside decision can still be influenced.

Can my team work opportunities together?

Yes. Anything you pursue moves into Journey Hub, the CRM side of the platform: a shared pipeline with stages and win probabilities, tasks with owners and due dates, the solicitation and its attachments attached to the pursuit, comments with mentions, and reporting for leadership. The proposal team inherits the capture team's reasoning rather than a handover call.

How much does federal contract search cost?

Searching SAM.gov directly is free. samsearch is a paid platform priced per team, and the honest way to decide is whether the qualification, the amendment tracking and the drafting save more hours than they cost. Pricing is on the pricing page, and a demo runs against your own profile rather than a prepared dataset.

See it run against your own profile.

A demo searches real federal notices against your capabilities, reads a package you choose, and shows you the draft it produces.

Book a Demo

Federal contract search, with the attachments already read.

Search in plain English. Get a summary instead of a 200-page package. Ask the solicitation a question and get a cited answer. Be told the moment an amendment changes something.

Book a Demo