AI for staffing agencies that keeps recruiter judgment in charge.
Recruiters read a CV, check notes, compare the role brief, write a summary and prepare a message, many times a day. Most of that information already sits in your applicant tracking system. We build AI that turns it into a factual first draft inside the tools your recruiters use, and the recruiter decides what goes to the client or candidate.
Where recruiter time goes before the conversation.
Reformatting the same facts
Submission packs and client summaries rebuild information that already sits in the ATS, one candidate at a time.
Drafting every message
Follow-ups and updates are personal enough that templates fall short, so recruiters write them by hand.
Conflicting records
Old CVs, duplicate profiles and stale notes make a quick summary risky unless someone checks every fact.
Rules on hiring tools
New York City's Local Law 144 covers some automated employment decision tools, including screening, and discrimination law still applies with a person involved.
AI for the preparation around each placement.
Candidate summaries
A factual summary for a specific role, drawn from the CV, confirmed profile facts and approved notes, with the source of each important claim.
Submission packs
The client's preferred format, narrative or skills table, prepared from the same checked facts for the recruiter to edit.
Follow-up drafts
Updates and outreach written from recorded facts and decisions, never an invented reason or a decision the client has not made.
Conflict flags
Where the CV and the notes disagree, the draft shows the conflict so the recruiter confirms the fact before sending.
Onboarding requests
Once drafting is reliable, requests for onboarding documents tracked against a checklist.
Inside your ATS first
If your ATS already has AI features that cover the job, we set those up and measure them before building anything custom.
The first project prepares work. It does not rank people.
A summary separates demonstrated experience from facts that still need confirming. It does not add a certification because the job asks for one, infer work authorization or availability from an old note, or guess at sensitive personal characteristics. A missing qualification is shown as unconfirmed, not treated as a rejection.
Scoring, screening and rejection are separate decisions with their own legal review, validation and monitoring. We keep them out of the first scope and make sure the assistant does not drift into ranking as prompts are added.
Candidate summaries for a 35-recruiter agency, as an example.
This is an illustrative scenario to show how a first project runs, not a QueryNow client result. Every number is an assumption to replace with your own.
- 01
Scope
Before the build600 candidate summaries and messages a month, prepared from an established ATS. Screening decisions and accommodation requests stay out.
- 02
Build
Two weeksSummaries with sources, conflict flags and follow-up drafts in the recruiter's normal workspace, using the ATS's own AI features where they fit.
- 03
Test
Before go-liveAbout 150 candidate and job pairs, including incomplete CVs, conflicting dates and specialist credentials. Proposed gates: at least 98% factual fidelity on the agreed fields and zero invented qualifications.
- 04
Measure
First monthsEditing time, unsupported claims per pack, important omissions and how often recruiters use the draft. Placements are not the pilot measure.
How the monthly value is worked out.
Hypothetical figures from the example above. Your volume, minutes and editing effort decide the real answer.
| Input | Example value |
|---|---|
| Summaries and messages a month | 600 |
| Minutes per task, before and after including edits | 15 before, 8 after |
| Hours recovered a month | 70 |
| Loaded hourly cost | $45 |
| Share of recovered time that becomes useful work | 65% |
| Monthly software and operating cost assumed | $800 |
| Net monthly capacity value | About $1,248 |
| At half the volume | About $224 |
One workflow, built in two weeks, paid when it works.
We scope one recruiting workflow with you, agree the acceptance tests before we start, and build it in your environment in two weeks. You pay $10,000 only after every test passes. If your ATS already covers the task, the right answer may be configuration and measurement rather than a build, and we will say so.
Bring your current ATS features and one example that still takes too much manual effort. That shows quickly whether the volume and records justify the work.
Sources for the rules on this page.
Checked in October 2026.
What the first project gives your agency.
Checked facts
Every important claim in a summary points to its source.
Faster packs
Client-ready formats prepared from the same record.
Honest messages
Drafts based only on recorded facts and decisions.
Visible conflicts
Disagreeing records flagged before anything is sent.
Measured editing time
Total minutes per task, after recruiter edits.
Clear limits
No scoring or screening in the first scope.
From workflow to working tool in two weeks.
Scope and sign
We define the workflow, the deliverables, and the acceptance criteria, and sign an agreement on them before anything starts.
Build
We build the tool in your environment, with Claude Code and automated evaluation against your own data.
Pay when it works
$10,000, due only after every criterion in the signed agreement is met. Nothing before that.
One recruiting workflow built in two weeks in your environment, paid only after it passes the acceptance tests you signed.
What AI costs, including after the build →
Tell us the workflow →Questions buyers ask.
What is a good first AI project for a staffing agency?
Candidate summaries and follow-up drafts for one type of role, built from your ATS records. The recruiter edits and approves every output, and scoring or screening stays out of the first scope.
Our ATS already has AI features. Why would we need QueryNow?
You may not. We check the native features first. If they cover the task, the work is setup, approved templates and measurement. A custom build makes sense only where a gap remains, for example between two systems.
Does NYC Local Law 144 apply to AI summaries?
It covers certain automated employment decision tools and can reach screening as well as final hiring. Whether it applies depends on how the tool is used. A factual summary the recruiter edits is a different use from a score that ranks candidates, and that difference needs legal review.
Will this reduce time to fill?
It may support faster preparation, but time to fill depends on demand, role difficulty, candidate interest and client decisions. We measure the drafting benefit directly and keep placements separate.
How do you stop the AI inventing qualifications?
The summary uses only the CV, confirmed profile facts and approved notes, shows the source of each important claim, and marks missing facts as unconfirmed. The release test requires zero invented qualifications.
How much does it cost?
One workflow is $10,000, built in two weeks and paid only after it passes your acceptance tests. Model usage and hosting run on your own cloud bill.
Ready to ship AI that actually works?
Send us the workflow. We return a fixed scope, price, and acceptance criteria in 48 hours. You pay $10,000 only after it works.
Tell us the workflow →The mid-market AI advantage: How $50M to $500M companies out-ship the Fortune 500
Why $50M to $500M companies get AI into production faster than the Fortune 500.
Read it free →