Grant & RFP Fit Scorer (AI) — Grants.gov, SAM.gov Bid/No-Bid

What it does

Government Grant & Contract Opportunity Fit Scorer ranks federal funding and contract opportunities by how well they fit your organization. Describe your company once (what you do, NAICS codes, set-asides and certifications, size, where you work, keywords, past awards, budget range), then give it either specific opportunities (Grants.gov IDs or URLs, SAM.gov notice IDs, or JSON records from any feed) or search terms. For every opportunity it writes one dataset item with the normalized record plus a fit block: a fitScore from 0 to 100, a fitVerdict (strong, possible, weak, disqualified), a recommendedAction (pursue, review, skip), fitReasons, hard disqualifiers (eligibility, set-aside, deadline passed, already awarded), the keyRequirements pulled from the synopsis, and a one-paragraph fitSummary a grant writer or capture manager can act on. The run's SUMMARY record carries the top-10 ranking.

Search terms are run against the official Grants.gov API (no key), the SAM.gov Opportunities API (your free personal key) and USAspending (no key; recent awards, scored as "who won similar work" intelligence). Every fetched opportunity is enriched with its full synopsis or notice text before scoring.

It has three scorers, chosen with the scorer input and recorded on every item as scoreMethod. AI scores are the default: when scorer is left empty the Actor uses the hosted Claude model; set scorer=rules for a free, model-free run, or scorer=byok to use your own key.

• hosted (default): Claude on the Actor's own key, tokens included, from $0.025 per opportunity. Nothing to set up. Pick a model tier: fast (Claude Haiku 5.5, $0.025), balanced (Claude Sonnet 5.5, $0.30) or best (Claude Opus 5.5, $0.60). If the model cannot be reached, the opportunity gets the rules score instead, marked scoreMethod: "rules-fallback" with a scoreWarning, and is charged the rules price; a run never fails because of the model.
• byok (bring your own key): Claude scores the opportunity using the anthropicApiKey you supply (stored as a secret, sent only to api.anthropic.com), on the same three tiers. You pay Anthropic for the tokens and $0.005 per opportunity here; each item records tokensIn, tokensOut and modelCostUsdEstimate.
• rules: a deterministic scorer with documented weights (keywords 30, NAICS 20, set-aside 15, eligibility 15, geography 10, deadline 10, budget 10, status 5), scaled by relevance so an open, eligible grant about something else still scores low. No language model, no key, runs in milliseconds, costs $0.003 per opportunity. Every model-mode item also carries this score as rulesFitScore with the per-component rulesBreakdown.

Built for grant writers, nonprofit development teams, SBIR/STTR applicants, small government contractors and capture managers who get dozens of keyword hits a week and need the ten worth reading, and for AI agents that need a bid/no-bid judgment on demand. Results are available as JSON, CSV, Excel, through the API, webhooks and the Apify MCP server.

How to use

1. Fill in the Company profile (JSON): description, capabilities, NAICS codes, set-asides, organization type, locations, keywords, budget range. Only a description or a few keywords are strictly needed; the more you give, the sharper the score.
2. Either list Opportunities (Grants.gov IDs/URLs, SAM.gov notice IDs, or JSON records) or enter Search terms and pick Sources; add your SAM.gov API key to include contract notices.
3. Leave Scorer empty for hosted AI scores and pick a Model tier (fast is the default); or choose byok and paste your Anthropic API key; or choose rules for a free run. Click Start. Items appear as each opportunity is scored; the SUMMARY record lists the top 10. Schedule the run weekly to triage every new opportunity automatically.

Quick start input

The same input the Store runs in its daily automated test, so it is cheap and always available; paste it into the JSON tab of the input form or send it as the body of an API or MCP call:

{
  "companyProfile": {
    "name": "Cascadia Civil Engineering",
    "description": "Civil engineering firm with 40 staff, NAICS 541330, SBA small business, serving the Pacific Northwest; past EPA and USDA rural water and wastewater grants.",
    "capabilities": [
      "water infrastructure design",
      "wastewater treatment",
      "stormwater management",
      "rural water systems",
      "environmental engineering"
    ],
    "naicsCodes": [
      "541330",
      "237110"
    ],
    "setAsides": [
      "SBA"
    ],
    "organizationType": "small business",
    "businessSize": "small",
    "locations": [
      "WA",
      "OR",
      "ID"
    ],
    "keywords": [
      "water",
      "wastewater",
      "infrastructure",
      "drinking water",
      "rural"
    ],
    "pastAwards": [
      "EPA Water Infrastructure Improvements for the Nation",
      "USDA Rural Development Water and Waste Disposal"
    ],
    "budgetRange": {
      "min": 50000,
      "max": 5000000
    }
  },
  "searchTerms": [
    "water infrastructure"
  ],
  "sources": [
    "grantsGov"
  ],
  "oppStatuses": [
    "posted"
  ],
  "postedWithinDays": 30,
  "maxOpportunities": 25,
  "scorer": "rules"
}

Input

Field	Type	Required	Description
companyProfile	object	yes	name, description, capabilities[], naicsCodes[], setAsides[] (SBA, 8A, HZC, SDVOSBC, WOSB, EDWOSB, VSA or their names), organizationType, businessSize, locations[] (states, regions or "nationwide"), keywords[], pastAwards[], budgetRange{min,max}
opportunities	array	no	Grants.gov IDs or URLs, SAM.gov notice IDs or URLs, or JSON records (a gov-opportunity-feed item, or any object with title, description, agency, naicsCode, setAside, closeDate, url…). Supplied records are scored without a fetch
searchTerms	string[]	no	Keywords searched in each enabled source; required when opportunities is empty
sources	string[]	no	grantsGov (default), samGov (needs samApiKey), usaspending (recent awards)
oppStatuses	string[]	no	Grants.gov statuses for searches (default posted)
postedWithinDays	integer	no	Search window, 1–365 days (default 30)
maxOpportunities	integer	no	Cap on opportunities scored per run, 1–1,000 (default 100)
samApiKey	string (secret)	for SAM.gov	Free personal key from sam.gov; never logged
samFilters	object	no	{"naics": "541330", "setAside": "SBA", "state": "WA", "ptype": "o,k"}
scorer	string	no	hosted, byok or rules. Empty = hosted when the Actor's model key is configured, else rules
anthropicApiKey	string (secret)	for byok	Your Anthropic API key; used only in byok mode
modelTier	string	no	fast (default, Claude Haiku 5.5), balanced (Claude Sonnet 5.5) or best (Claude Opus 5.5); applies to hosted and byok
minRulesScoreForModel	integer	no	Model modes: opportunities whose rules score is below this get the rules score at the rules price and skip the model (default 0 = model for all)
maxInputTokens	integer	no	Per-opportunity input cap for model modes, 2,000–8,000 (default 8,000)
maxConcurrency	integer	no	Opportunities processed in parallel, 1–5 (default 3)
webhookUrl	string	no	POST target for the run summary (counts, verdicts, top 10, token totals)

Output example

A record the buyer supplied, scored in rules mode for a small civil engineering firm in the Pacific Northwest:

{
  "source": "provided",
  "recordType": "opportunity",
  "opportunityId": "RFP-22",
  "title": "Wastewater Treatment Plant Engineering Services",
  "agency": "City of Yakima",
  "closeDate": "2026-12-15",
  "url": "https://example.org/rfp/22",
  "description": "Design of rural drinking water and wastewater infrastructure in Yakima, WA. The contractor must provide a licensed PE. Total small business set-aside.",
  "naicsCode": "541330",
  "setAside": "SBA",
  "setAsideCode": "SBA",
  "placeOfPerformanceState": "WA",
  "fitScore": 100,
  "fitVerdict": "strong",
  "recommendedAction": "pursue",
  "fitReasons": [
    "Matches your keywords \"water\", \"wastewater\", \"infrastructure\", \"rural\", \"wastewater treatment\" and partially \"water infrastructure design\" (\"wastewater\", \"wastewater treatment\" in the title).",
    "NAICS 541330 matches your code 541330.",
    "Set aside for SBA, which your profile qualifies for.",
    "Place of performance WA is in your service area.",
    "Closes on 2026-12-15, in 59 days."
  ],
  "disqualifiers": [],
  "keyRequirements": ["The contractor must provide a licensed PE."],
  "fitSummary": "Wastewater Treatment Plant Engineering Services (City of Yakima) scores 100/100 for Cascadia Civil Engineering, a strong fit. Matches your keywords \"water\", \"wastewater\", \"infrastructure\", \"rural\", \"wastewater treatment\" and partially \"water infrastructure design\" (\"wastewater\", \"wastewater treatment\" in the title). NAICS 541330 matches your code 541330. Set aside for SBA, which your profile qualifies for. Place of performance WA is in your service area. Scored by deterministic rules (keywords, NAICS, set-aside, eligibility, geography, deadline, budget, status); no language model was used.",
  "daysToClose": 59,
  "rulesFitScore": 100,
  "rulesBreakdown": {
    "keywords": { "score": 1, "weight": 30, "note": "Matches your keywords \"water\", \"wastewater\", … in the title)." },
    "naics": { "score": 1, "weight": 20, "note": "NAICS 541330 matches your code 541330." },
    "setAside": { "score": 1, "weight": 15, "note": "Set aside for SBA, which your profile qualifies for." },
    "eligibility": { "score": null, "weight": 15, "note": "No applicant-type list on the record." },
    "geography": { "score": 1, "weight": 10, "note": "Place of performance WA is in your service area." },
    "deadline": { "score": 1, "weight": 10, "note": "Closes on 2026-12-15, in 59 days." },
    "budget": { "score": null, "weight": 10, "note": "No award amount on the record." },
    "status": { "score": 1, "weight": 5, "note": "Status not stated; treated as open." }
  },
  "scoreMethod": "rules",
  "profileName": "Cascadia Civil Engineering",
  "fetchedAt": "2026-10-17T10:26:20.692Z"
}

Opportunities fetched from Grants.gov carry the feed fields too (number, agencyCode, status, postedDate, assistanceListings, awardCeiling, awardFloor, estimatedTotalFunding, expectedAwards, costSharing, fundingCategories, eligibilities, agencyContactName, agencyContactEmail, the synopsis as description); SAM.gov notices carry noticeId, solicitationNumber, department, subTier, office, type, responseDeadLine, naicsCode, setAside, placeOfPerformance, pointOfContact and the notice text; USAspending awards carry recipientName, awardAmount, awardingSubAgency and always a disqualifier saying the award is already made. Items written in byok or hosted mode have the same fit fields (the model writes the reasons and summary in prose, with recommendedAction as its bid/no-bid call) and additionally carry modelTier, model, tokensIn, tokensOut, modelCostUsdEstimate (USD at Anthropic list prices) and inputTruncated. If a model call fails (rate limit, API error, timeout, refusal, invalid JSON) or the hosted model is not configured, the item falls back to the rules score with scoreMethod: "rules-fallback", records the reason in scoreWarning, and is charged the rules price.

Error items look like { "recordType": "error", "source": "grantsGov", "input": "…", "error": "NOT_FOUND", "errorMessage": "…" } (INVALID_OPPORTUNITY for an entry that is neither an ID, a URL nor a record; NOT_FOUND; SAM_KEY_REQUIRED; FETCH_FAILED, SAM_AUTH_FAILED, SAM_RATE_LIMITED for a source or search term that could not be read) and are charged at the fetched price, because the work to discover the problem was done. A byok run without a key, or with a key Anthropic rejects, fails with a clear message before any further opportunity is charged; an empty profile fails before any charge at all.

The SUMMARY record in the key-value store holds scored, errors, byVerdict, byMethod, topOpportunities (the 10 best by fitScore, with title, agency, close date and URL), model token totals and charges.

If the output saves you time, a review on the Store page helps other buyers find this Actor; the Store invites you to leave one after you have run it a few times.

Pricing

Pay per event. You only pay for what the Actor produces; set a maximum cost per run in the run options to cap spend.

Event	Price	Notes
Opportunity fetched	$0.001	Per opportunity fetched from Grants.gov, SAM.gov or USAspending (search result or pasted ID), including its synopsis or notice text; not charged for records you supply. Error items are charged at this price
Opportunity scored (rules, no model)	$0.003	Per opportunity in rules mode, below the minRulesScoreForModel pre-screen, or on a rules-fallback
Opportunity scored (your Anthropic key)	$0.005	Per opportunity scored by Claude in byok mode, any tier; your tokens are billed by Anthropic, not here
Opportunity scored (hosted, fast tier)	$0.025	Claude Haiku 5.5 on the Actor's key, tokens included, up to 8,000 input tokens per opportunity
Opportunity scored (hosted, balanced tier)	$0.30	Claude Sonnet 5.5, same caps
Opportunity scored (hosted, best tier)	$0.60	Claude Opus 5.5, same caps
Actor start	$0.00005	Apify platform event

Worked examples. A weekly run that fetches 200 Grants.gov matches and scores them on hosted fast (the default) ≈ $0.20 + $5.00 = $5.20; the same on balanced ≈ $60.20, on best ≈ $120.20; rules only ≈ $0.80. Bring your own key: 200 opportunities ≈ $1.20 here plus roughly $0.13 of Claude Haiku 5.5 tokens on your Anthropic bill (about 3,800 input and 500 output tokens per opportunity at $0.10 and $0.50 per million); about $2.50 on Sonnet 5.5 ($2 and $10 per million) or $5.00 on Opus 5.5 ($4 and $20 per million). With minRulesScoreForModel: 35, a typical broad search sends a third of the matches to the model and the rest stay at $0.003. Scoring 50 records you already have (no fetch) on fast ≈ $1.25.

Spending limit and model tiers. The hosted model is called before its per-opportunity event is charged, so before every model call the Actor checks that your run's maximum total charge still covers that event. When it does not, the model is not called: the opportunity gets the rules score instead, charged at the rules price, with a scoreWarning that names the tier and the limit to set (the run's SUMMARY counts these as modelSkippedForSpendingLimit). When not even a rules-scored item fits, the run stops cleanly with a log line and nothing more is charged. A run with no limit set behaves as before. In practice: set the maximum total charge to at least the tier price times the number of opportunities you want scored by the model, plus $0.001 each for fetched ones.

How the hosted prices are set. Against the worst case, not the average. The opportunity text is capped at 8,000 input tokens (about 28,000 characters, more than any Grants.gov synopsis or SAM.gov notice text, which is cut at 20,000 characters), every call carries about 1,700 fixed tokens of profile, instructions and schema on top, and output is capped at 1,000 tokens, so the most one opportunity can cost in tokens is: fast 9,700 × $0.10/M + 1,000 × $0.50/M ≈ $0.0015; balanced 9,700 × $2/M + 1,000 × $10/M ≈ $0.029; best 9,700 × $4/M + 1,000 × $20/M ≈ $0.059. Apify pays developers 80% of the event price ($0.02, $0.24 and $0.48 respectively), so every tier clears its worst-case token cost even at the cap, by about 13×, 8× and 8× (at least 85% margin at the cap, the bar every model Actor in this account meets); a typical 2,000-character synopsis costs a fraction of that. Reasoning tokens are kept out of the output cap (thinking is turned off on Haiku and Sonnet; Opus runs at low effort). A model failure is never charged at the hosted price: the item falls back to the rules score at $0.003.

Limits and fair use

• Sources are read with plain HTTP from Apify datacenter IPs through their official public APIs only: Grants.gov search2 and fetchOpportunity (no key, about 2 requests per second), SAM.gov Opportunities v2 and its notice-description endpoint (your personal key; SAM.gov allows about 1,000 requests per day per non-federal key, and a search is one request per 1,000 notices plus one per notice for its text), USAspending spending_by_award (no key). No scraping of the websites, no browsers, no proxies.
• SAM.gov notice IDs are looked up within the last 365 days of postings (the API's maximum window). Grants.gov IDs are fetched directly, any age; closed opportunities are scored and disqualified.
• The rules scorer matches your keywords and capabilities against the title and description with light stemming, maps NAICS codes by shared prefix (6, 5, 4, 3, 2 digits), expands set-aside implications (an 8(a) firm also qualifies for total small business set-asides), reads Grants.gov applicant types against your organizationType, and treats missing data as "not applicable" rather than as a penalty. It is deterministic and documented; it does not read the fine print. Model modes do, and get the rules findings as facts in the prompt.
• Model modes cut the description to maxInputTokens (at most 8,000 tokens) before the call and say so on the item; output is capped at 1,000 tokens per opportunity. The model is set by modelTier in both hosted and byok modes.
• At most 1,000 opportunities per run (maxOpportunities); search results are capped by that number across all terms and sources and de-duplicated by source and ID within the run.
• Time budget. Each search term on each source gets 10 minutes (paging included), and a search that delivers no new candidates for 60 seconds is stopped. Candidates fetched before the stop are kept, scored and charged as usual and the unit is listed under truncated in the SUMMARY record (TIME_BUDGET); a search that produced nothing returns a charged error item (TIME_BUDGET or the source's error code). Source requests time out after 20 seconds (30 for SAM.gov, 45 for USAspending) including retries, and Retry-After pauses are capped at 10 seconds. After two search terms in a row fail on the same source, the remaining terms for that source are not requested and each gets one charged SOURCE_UNAVAILABLE error item (pasted IDs behave the same after three failed lookups). If three enrichment requests in a row fail, the remaining candidates are scored from their search data (detailError/descriptionError says so). A model call that takes longer than 60 seconds is cancelled without a retry and the opportunity gets the rules score at the rules price.
• Memory: default run memory is 512 MB (256 to 1024 MB allowed); the Actor holds one opportunity per worker at a time.
• Only public data is accessed: the three government APIs and, in model modes, the Anthropic API. Your anthropicApiKey and samApiKey are stored encrypted by Apify and are never logged or written to the dataset.

Related Actors

• US Government Grants & Contract Opportunities Feed (same account): the raw Grants.gov, SAM.gov and USAspending records with new/updated flags for scheduled monitoring, from $0.001 per record. Its dataset items can be pasted straight into this Actor's opportunities input.
• Link 1–2 more sibling Actors from this account once published.

Use from AI agents (MCP)

Pin this Actor in the Apify MCP server with one link: https://mcp.apify.com?tools=owlabee/gov-opportunity-fit-scorer, or let the agent find it with the search-actors tool. It is pay-per-event with limited permissions, so agents can discover it, inspect the input schema with fetch-actor-details and pay per run through agentic payments. Prompts that map directly onto the input:

1. "Score the Grants.gov opportunities posted in the last 30 days for 'water infrastructure' against our company profile and give me the top five with reasons."
2. "Rank these three SAM.gov notices for a HUBZone IT firm in Virginia and tell me which to bid on."
3. "Check whether we are eligible for Grants.gov opportunity 364099 and summarize its requirements."

The agent receives the same dataset records shown in the output example above.

FAQ

Which scorer should I use? The default, hosted on the fast tier, reads the synopsis and writes reasons, disqualifiers, requirements and a paragraph for a flat $0.025 per opportunity with nothing to set up; move to balanced or best when eligibility fine print and subtle capability fit matter more than price. Use byok when you already have an Anthropic key and want to control the model and the token spend yourself. Use rules for bulk triage, monitoring and schedules: it is free of model costs, needs no key, and its reasons are verbatim facts (keyword hits, NAICS and set-aside matches, days to close). A good pattern: hosted fast with minRulesScoreForModel: 35, so the model only reads what the rules could not rule out.

Does this need proxies or an API key? No proxies. Grants.gov and USAspending are public APIs read over plain HTTP from Apify's standard infrastructure. A free SAM.gov personal key is needed only for SAM.gov notices; an Anthropic API key only for byok mode. hosted and rules on Grants.gov need nothing.

How is cost controlled? Cost = opportunities × (the fetched price when the Actor fetches them + the price of the scorer mode and tier you chose). Limit opportunities with maxOpportunities, narrow searchTerms, postedWithinDays and samFilters, pre-screen with minRulesScoreForModel, lower maxInputTokens in byok mode to cut your own token bill, and set the run's maximum total charge; the Actor checks it before every model call and stops cleanly when it is reached. Every model-mode item records tokensIn, tokensOut and modelCostUsdEstimate, and the run's SUMMARY record totals them.

How is the fit score computed? In rules mode: eight weighted components (keywords 30, NAICS 20, set-aside 15, eligibility 15, geography 10, deadline 10, budget 10, status 5), each 0–1 or not applicable, averaged over the applicable ones and scaled to 0–100; the result is then multiplied by 0.45 + 0.55 × relevance, where relevance is the keyword and NAICS part, so an open, eligible grant about something else cannot score high. Hard disqualifiers (closed, already awarded, an applicant-type list or set-aside that excludes you) cap the score at 20. Verdicts: strong ≥ 70, possible ≥ 45, weak below, disqualified when a disqualifier applies. In model modes the model scores 0–100 against the same definitions, with the rules findings in its prompt; the deterministic score is kept on the item as rulesFitScore.

Can I score opportunities from another feed or my own list? Yes. Paste JSON records into opportunities: a gov-opportunity-feed dataset item is recognized as is; any other object needs at least a title or description, and uses agency, naicsCode, setAside/setAsideCode, closeDate (or deadline), state, eligibilities, awardCeiling and url when present. Supplied records are not charged the fetched event.

Can an AI agent use this? Yes; see "Use from AI agents (MCP)" above for the pin link and example prompts. Schedules and webhooks work the same way from the Console or the API.

What happens on errors? A search term or ID that cannot be fetched produces an error item with a clear code instead of failing the whole run; other opportunities are unaffected. A Grants.gov detail or SAM.gov notice text that cannot be downloaded leaves detailError or descriptionError on the item and the opportunity is scored from the fields that were fetched. A model call that fails falls back to the rules score (scoreMethod: "rules-fallback", reason in scoreWarning) and is charged the rules price, so a hosted run never fails because of the model. Only a byok run with a missing or rejected key stops, with a clear message.

Can I try it for free? Yes. The Apify free plan comes with monthly platform credit, and the quick-start input above costs a few cents per run, so you can test it on your own inputs before paying anything.

How do I find this Actor again? Bookmark it on the Store page (the bookmark icon next to the title); bookmarked Actors are listed in your Console, and the Store page shows new versions as they ship.

Legal

Use the data in compliance with each source's terms and applicable law. The Actor reads only the official public APIs of Grants.gov, SAM.gov (with your own key, under SAM.gov's API terms; it never scrapes the SAM.gov website) and USAspending.gov, keeps request rates polite, and in model modes sends the opportunity text and your profile to the Anthropic API solely to produce the score. Scores, reasons and summaries are advisory and machine-generated; they are not legal or eligibility determinations, may contain errors, and must be verified against the notice of funding opportunity or solicitation before a bid or application decision. Agency contact names and emails are reproduced only as the agencies publish them in the notices. The Actor is not a consumer reporting agency and its output must not be used for employment, credit, insurance, or tenant screening decisions.