Applicant tracking · agencies of 1–10
Proof you can hand a client.
Recruit Copilot reads each application against the criteria your client actually asked for, and returns one Submission Record — every finding beside the passage it came from, under your own logo.
- 5
- criteria, in the order your client wrote them
- 0
- scores, percentages, or ranked shortlists
- 1
- document that leaves the building
Submission Record
Sarah Whitfield · draft
Criterion 2 of 5
Five years in-house payroll for a UK entity of 200+ staff
Evidenced
“I ran monthly payroll for 240 UK employees at Halliday Group for six years, including RTI submissions and year-end reporting.”
CIPP qualification
Not found
The path an application takes
Four stages, and nothing hidden between them
Stage 1
The Brief
Write the criteria the client actually asked for, before applications arrive.
Stage 2
Review
Each application read against those criteria, every finding beside the sentence that produced it.
Stage 3
Crosscheck
Identity and consistency signals: contact details, timeline arithmetic, document metadata, duplicates across your own history.
Stage 4
Submission Record
All of it in one document carrying your logo.
The part most tools get wrong
There is no score.
No match score. No percentage. No star rating. No ranked shortlist. You get five criteria in a fixed order, and for each one either a citation or a stated absence. Compare the shape of the row, then read the citations that matter to you.
Candidate
Criteria, in fixed order
Evidenced
Sarah Whitfield
4 of 5 evidenced
Daniel Osei
3 of 5 evidenced
Priya Raman
4 of 5 evidenced
- Filled
- evidenced, with a citation
- Hollow
- not found in the application
- Cell 1 · Payroll for 200+ staff
- Cell 2 · CIPP qualification
- Cell 3 · Sage 50 Payroll
- Cell 4 · Managed one or more reports
- Cell 5 · Commutable to Leeds
Slice by slice, on camera
The build, in public
The product is being built on camera, slice by slice, from an empty repository. Nothing is skipped and nothing is pre-baked: the schema, the tests, the model boundary and the failures all happen in view. Someone following along can build their own; someone who would rather not can buy this one.
Skeleton
Repository, local Postgres, this page in production
Org, user, auth
Multi-tenant from the first commit, with tests that assert failure
The Brief
Every ATS starts with a job. This one will not open one without criteria.
Person and Sighting
No resolving source, no record — a database refusal, not a request
The model boundary
Evaluation runs inside the product, on the customer's key
Candidacy
The pipeline, and the moment a person becomes a candidate
Documents
Two-column PDFs, tables, scans, and DOCX from 2011
Review
Every finding cited against pinned criteria
Crosscheck
Integrity signals, traceable to artifacts
Submission Record
The artifact that leaves the building
Episodes publish as each slice lands. Module 0 of the course starts where this leaves off.
The course
Build your own
Seven modules, each one shipping a working piece of an AI-forward recruiting process — and each one tied to a number it is supposed to move. You start by finding your constraint, so you do not spend six weeks automating a stage that was never the problem.
| # | Module | The build | The metric it moves |
|---|---|---|---|
| 0 | The constraint map | Map your funnel, find the constraint | MovesThroughput — stops you optimising the wrong stage |
| 1 | Market mapping | A market-map agent (conference scrapers, Scholar, LinkedIn) | MovesLeads/week, coverage |
| 2 | Tier definition | Personas from your own hires, both traps, and the four-fifths gate | MovesLead quality score |
| 3 | Sourcing | The sourcing agent, tiered output | MovesQuality score ≥ 75, % to outreach |
| 4 | Screening & assessment | Structured evidence for recruiter and hiring-manager screens | Moves% reaching onsite, HM NPS |
| 5 | Coordination | The scheduling constraint | MovesScheduling cycle time, candidate experience |
| 6 | Measurement | The metric set, before and after | MovesAll of the above, made visible |
What you end up with
- Your own agents, running on your own keys, against your own market.
- A before-and-after number for every module you finish.
- The four-fifths gate applied to your tiers, so the process survives scrutiny.
The build series is free and publishes as each slice lands. The course is the longer version, with the market mapping and sourcing agents you would run yourself.
Take the course
Seven modules, each shipping a working build and a number it moves. Tell us where to send it when enrolment opens.
Pricing
Every price is on this page
There is no quote to request and no call to book. Start on Free without a card.
Free
$0
no card
- 100 applications / month
- Pre-flight
- Watermarked Submission Record
Solo
$79
per month
- 500 applications / month
- 1 seat
- Evidence-linked Review
Firm
$299
per month
- 2,500 applications / month
- 5 seats
- Unbranded Submission Records
Agency
$699
per month
- 10,000 applications / month
- Unlimited seats
- Compliance audit pack
Annual billing costs ten months rather than twelve.
Identity verification checks are sold separately as credit packs, never bundled.
Early access
Two ways in. Pick the one you mean.
Follow the build and run your own agents, or take the finished product and get back to placing people. Which form you fill in is what tells us what to send you — there is no call to book either way.
I would rather build my ownBuy this one instead
Design partner seats open before the build is finished, at the prices published above.