How do contractors use AI to triage GovWin IQ opportunities?
Deltek has put AI inside GovWin IQ to sort opportunities, and Costpoint can import the ones you mark. Here is what each layer does, according to Deltek's own documentation, and where the match against your own record still has to be made.
The short answer: In three layers: inside GovWin IQ, Deltek's Dela features do the first pass: Smart Fit Scores rate an opportunity from 40 to 100 against the contracts a company has won before, Smart Summaries condense the record, and Ask Dela answers questions from the documents in that one record. Costpoint's Import GovWin IQ Data screen then carries the opportunities you mark into Manage Opportunities, where the probability of winning is a field a person fills in. What neither layer does is match each opportunity against your own books: the contracts and CPARS assessments already in Costpoint, the funded backlog, the response clock the FAR sets. That cross-system check is what a read-only assistant with sources adds, and the bid decision stays with a person.
Triage is a matching problem. One side of the match is the opportunity: agency, scope, NAICS code, competition type, value, stage, dates. The other side is you: what you have performed for whom, how it was rated, what you are carrying now, and whether the clock leaves room to respond. GovWin IQ supplies the first side in detail. The second side is in Costpoint and the contract files, and nothing in the Deltek pages we read says GovWin IQ reads it.
GovWin IQ supplies one half of the match
Deltek describes GovWin IQ as opportunity intelligence pairing more than 150 market analysts with its AI. When a record is imported into Costpoint, the fields that arrive show what the product tracks. Deltek's GovWin IQ subtask help page lists the opportunity's current status in GovWin IQ ("for example, Pre-RFP or Awarded"), department and agency, solicitation number, estimated value in thousands, duration, competition type, contract type, NAICS code, primary offering ("the primary type of firm that should respond to the opportunity"), potential bidders, a program description with summary, background, requirements and funding, key dates, key contacts, and the date of the last analyst update. "Most fields on this subtask and its child subtasks are not editable as they come from GovWin IQ."
That is a strong description of the opportunity. It says nothing about your fit, because fit lives in your records. FAR 15.305(a)(2) tells evaluators to consider the currency and relevance of past performance, its source and the context of the data, and to consider the problems an offeror reports on those contracts and the corrective action it took. The records that answer those questions sit in Costpoint's Contracts module: the Performance Assessment tab of Manage Contracts holds each CPARS or other assessment with its type, period, status and score, and the contract record holds the agency, contract type and dates. We walked through that tab in Can AI draft government proposals from past performance without compliance risk?
| Opportunity (GovWin IQ), the firm's record (Costpoint) and the draft recommendation | Stage | Fit | Clock |
|---|---|---|---|
| Civilian agency, IT operations support recompete. Est. value $18M over 5 years, NAICS 541512, full and open. Record: one CPFF contract with this agency under this NAICS, ended 2 years ago, final CPARS on file rated Very Good; two active contracts under the same NAICS, no staffing conflict in the new period. Draft: Pursue. Confirm the incumbent and ask the program office for the draft RFP. | Pre-RFP | Fit 5 of 5 | RFP in about 6 months |
| DoD component, engineering and technical services. Est. value $42M, NAICS 541330, set-aside not yet stated. Record: no contract with this component; one subcontract under the NAICS, 3 years ago, no assessment on file. Draft: Answer the RFI only as a teaming partner; do not plan a prime bid. | Pre-RFP, RFI open | Fit 2 of 5 | RFI due in 12 days |
| Civilian agency, financial management support. Est. value $6M, NAICS 541611, small business set-aside. Record: two active contracts for this agency, interim CPARS rated Exceptional; funded backlog on both runs past the new period of performance. Draft: Bid. Check staffing against the two active contracts before committing key personnel. | RFP released | Fit 4 of 5 | Proposals due in 4 weeks |
Sample records with roles, not real companies, agencies or figures. The opportunity line uses the fields Deltek's help page says the GovWin IQ import carries; the record line uses fields on Costpoint's Manage Contracts screen and its Performance Assessment tab. The fit cells are a reading of how much of the firm's own record supports a bid, not a GovWin IQ Smart Fit Score and not a measurement. A full row means recent, rated work for the same agency under the same NAICS with room in the backlog; an empty row would mean none of those. The draft line is the recommendation a person would confirm or overrule.
What GovWin IQ's own AI already does
Use it first. Deltek's June 23, 2021 article on the Smart Fit Score says it "leverages machine learning to 'score' government contracting opportunities based on historical contract award data," from "five influential scoring criteria: NAICS codes, GovWin IQ Smart Tags, opportunity description/keywords, place of performance" and past history with an agency, each weighted and summed into "scores ranging from 40-100." The models are "trained to extract data from historical contracts won by a company," and users can view other companies' scores to see who they might be up against. Deltek's October 17, 2024 post on GovWin IQ data privacy names the suite: Smart Fit, Smart Summaries, Smart Search and Ask Dela.
The June 25, 2025 press release added Ask Dela Opportunity Chat ("Teams can ask questions about government opportunities and receive instant, actionable answers") and AI-Powered Proposal Outlines generated from solicitation documents. Deltek's April 2, 2026 blog says the chat is "only pulling from the source material housed in an individual bid or contract record," and gives sample prompts such as "How many companies have Smart Fit Scores above 70?" and "Explain the requirements in 500 words." On data handling, the 2024 post says GovWin IQ's AI uses "only GovWin proprietary data" and public sources, that "Customer-provided prompts will not be used to contribute to training models," that third-party providers include Microsoft Azure, OpenAI and AWS Bedrock, and that Ask Dela prompts are stored with a timestamp and visible only to that user.
Two things to notice. The score is computed from award history and the record's own text; Deltek does not say it reads your Costpoint contracts, your CPARS or your workload, and the Deltek pages we read publish no accuracy figure for it. And the chat is scoped to one record at a time, which is the right design for trust and the wrong shape for the question "which of these forty should we read today."
The Costpoint import moves the record, not the decision
Deltek's Import GovWin IQ Data help page says the screen is used "to import opportunity records from GovWin IQ into Costpoint," on demand or scheduled through Costpoint's job management, for records marked in GovWin IQ for download. It "imports opportunities from GovWin IQ that do not yet exist" in Costpoint's opportunity table, refreshes the GovWin IQ subtask on later runs, and writes a log. The screen "is available only if you are licensed for Capture & Contracts," and the integration has to be enabled in the Costpoint Configuration Utility.
On Manage Opportunities, the Status tab is where staff "specify opportunity status details, key dates, value of the opportunity, probability of winning the opportunity," and the Assessments tab is where they "rate the opportunity and assess if it is worth pursuing." The probability is a manual entry; the help pages describe no field that GovWin IQ fills with a probability, and no bid decision field at all. So the import puts the opportunity next to your contracts, which is exactly where it should sit. It does not compare them. The comparison, the probability and the decision wait for a person, and that comparison is the part of triage this article is about.
The clock is set by the FAR, not the dashboard
Stage decides how much time you have. FAR 15.201 says exchanges of information "from the earliest identification of a requirement" through receipt of proposals "are encouraged," lists presolicitation notices, draft RFPs, RFIs and preproposal conferences among the techniques, and says "RFIs may be used when the Government does not presently intend to award a contract" and that responses "are not offers." FAR 5.203 requires the notice "at least 15 days before issuance of a solicitation" and, above the simplified acquisition threshold and outside commercial items, "at least a 30-day response time for receipt of bids or proposals." FAR 10.001 has agencies conduct market research before soliciting above that threshold and use it to decide, among other things, whether small business programs apply.
Read against the GovWin IQ status field, that gives a triage rule: an RFI or pre-RFP record is where a response can still shape the requirement and where the match against your record is worth doing carefully; a record that first appears as a released solicitation leaves as little as 30 days and a requirement that is already written. Sorting the list by stage before sorting it by score is a rule a person can apply today, with or without AI.
What a read-only assistant adds
The assistant does the second half of the match and shows its work. For each opportunity in the morning's list, imported or still in GovWin IQ, it reads the record's agency, NAICS, competition type and dates, then reads your side: Manage Contracts for prior and current work with that agency and under that NAICS, the Performance Assessment tab for the CPARS status and score on each, the funded value and period of performance on the active contracts that would share staff, and the proposal archive for the last time you bid this agency. It returns a one-page fit memo per opportunity with every figure traced to its record and run time, a draft recommendation, and the questions it could not answer, such as a set-aside not yet stated. Every answer shows where it came from. No black boxes.
It never enters the probability, never marks the bid decision, never writes to Costpoint or to GovWin IQ, and never shows a reader a contract their Costpoint permissions do not reach; how that last rule is enforced is the subject of How do you keep an AI assistant from showing employees Costpoint data they can't see? The memo is a draft for the capture meeting. AI drafts and flags; people approve and sign. In Deltek's 17th annual Clarity study (May 2026, 917 government contractors), 26 percent named difficulty validating AI-provided insights among their top concerns; a memo whose every line names its record is built for that reader. The question of what the assistant may read from GovWin IQ content under your subscription is one for Deltek and your contract, not for us, so ask it before the build.
How to start
Take the last ten bid decisions your capture team made and the GovWin IQ records behind them. Run the match by hand once: which contracts and assessments in Costpoint would have supported each bid, and what the response window was on the day the record was marked. That is the test set. An assistant that reproduces those ten memos, with sources a contracts director can check, has earned the eleventh. Small proof first, then scale.
Common questions
Does GovWin IQ's Smart Fit Score use our Costpoint data?
Not according to Deltek's description. Deltek's June 2021 article says the score comes from machine learning models trained on historical contracts a company has won, using NAICS codes, GovWin IQ Smart Tags, the opportunity description and keywords, place of performance and past history with an agency, weighted and summed into a score from 40 to 100. It does not mention Costpoint, your CPARS records, your rates or your current workload. Those live in your own systems, which is why the second half of the match is your work.
Should an AI assistant make the bid or no-bid decision?
No. It should assemble the evidence a decision needs, with sources: the opportunity record, the matching contracts and performance assessments in Costpoint, the response deadline from the notice, and the questions nobody has answered yet. The probability of winning on the Status tab of Costpoint's Manage Opportunities and the rating on its Assessments tab are entries a person makes, Deltek's help describes no bid decision field at all, and that is where the decision should stay. AI drafts and flags; people approve and sign.
Do we need the Costpoint import to do this?
It helps, and it needs a license. Deltek's Costpoint 8.2 help says the Import GovWin IQ Data screen and Manage Opportunities are available only if you are licensed for Capture & Contracts, and that the GovWin IQ integration has to be enabled in the Costpoint Configuration Utility. Without it, the opportunity record stays in GovWin IQ and the past performance stays in Costpoint, and a person or an assistant has to read both. With it, the GovWin IQ fields sit in Costpoint next to the contracts they should be compared with.
How early in the cycle should triage happen?
Before the solicitation. FAR 15.201 encourages exchanges with industry from the earliest identification of a requirement through receipt of proposals, lists RFIs, draft RFPs and presolicitation notices among the techniques, and says responses to RFIs are not offers. FAR 5.203 requires the notice at least 15 days before a solicitation is issued and, above the simplified acquisition threshold, at least 30 days to respond once it is. An opportunity that first appears as a solicitation leaves little room; the ones worth triaging well show up as RFI or pre-RFP records.