How we run engagements

We run every engagement the same way. The Pack is a productized service — the deliverables are defined, the timeline is defined, the price is defined. The methodology below is what makes that possible.

1. Intake and qualification

Every engagement starts with the intake form. It is a structured questionnaire, not a sales call.

We ask about: headcount, industry, geographic footprint, which AI hiring tools are in use, the stages of recruitment those tools touch, your annual hiring volume, whether you have completed a bias audit, whether you have your vendor's technical documentation, whether legal is aware, and what triggered the inquiry.

Inside one business day we respond with one of three outcomes: a yes (with proposed scope and price), a no (with a referral if we can offer one), or a request for one or two clarifying questions.

We say no when:

• You have not yet selected an AI hiring tool. Start with HireAIScore and come back.
• Your hiring AI use is incidental and the regulatory exposure is low.
• You need full SR 11-7-style model risk management. That is a different SKU and a different firm.

2. Kickoff

Kickoff is a 30-minute call with everyone who will be in discovery — typically a sponsor (Chief People Officer, GC, or CPO), an HR or Talent Acquisition lead, an engineering or IT owner of the system, and one representative from legal. We confirm scope, in-scope regimes, system boundaries, and the stakeholder map.

Kickoff is where we lock the engagement contract. After kickoff, scope changes are managed as change orders — there is no scope creep priced into the base.

3. Discovery

Discovery runs across weeks 1 and 2. It is structured around four threads, run in parallel:

• Engineering and system inventory. What does the system actually do? What inputs does it take, what outputs does it produce, where does it sit in your hiring workflow, who has access?
• Vendor evidence. What documentation has the vendor produced — model cards, technical files, bias-audit reports, security attestations? We request what is missing in writing.
• HR / Talent Acquisition. How is the system used day-to-day? Who oversees the outputs? What training has the team received? What are your existing escalation paths?
• Legal. What is your existing position on AI in hiring? What contracts govern the vendor relationship? What jurisdictions are you currently telling regulators about?

Discovery produces an evidence ledger — a structured inventory of what exists, what is missing, and what is in dispute.

4. Drafting

Week 3 is drafting. The seven deliverables (see the Pack) are drafted in parallel against the evidence ledger. We use a fixed table of contents for the conformity file that maps to Annex IV section numbers; we use the Article 27 framework for the FRIA.

Where evidence is missing, we mark the gap inline — "[gap: vendor has not produced training-data documentation per Annex IV §4]" — and lift the gap into the remediation roadmap. We do not paper over evidence we do not have.

5. Review and sign-off

Week 4 begins with an internal review with your team. We meet with engineering, HR, and legal — typically over two or three working sessions — and incorporate one round of revisions.

We do not certify compliance. No consultant can. What we do is produce a file that your counsel can sign off on, and that a future regulator can navigate.

6. Handoff

Handoff is a one-hour walkthrough call. We deliver the final files — typically a structured Word document for the conformity file, a PDF of the FRIA, and supporting Word documents for the RMS, monitoring plan, and remediation roadmap. We answer questions. We do not retain the artifacts; the file is yours.

A second round of revisions is available within 30 days at no additional cost. Beyond that, additional revisions or updates are billed as add-ons.

7. Post-engagement support

We do not run an MSP-style retainer. We do offer:

• Annual conformity file refresh — a fixed-scope, fixed-price engagement to update the file each year as the system and the regulations evolve.
• Regulator response support — hourly, billed in 30-minute increments, if a notice arrives.
• Add-on tools — the Pack covers one AI hiring tool. Additional tools follow the same methodology and pricing.

Conflicts of interest

Casework runs engagements for AI deployers. Our sister property, HireAIScore, evaluates AI hiring vendors. The scoring rubric and evidence framework HireAIScore uses are documented publicly at hireaiscore.com/methodology. The two properties are operationally independent and the firewall is enforced as follows:

• Casework engagement leads do not contribute to HireAIScore scoring.
• Vendor evaluations on HireAIScore are not modified based on Casework engagement findings.
• Casework does not accept fees, kickbacks, or referral arrangements from AI hiring vendors. If a client introduces us to their vendor for evidence collection, we do not pay or receive payment for that introduction.
• Where we recommend a tool, technique, or external expert by name in a Casework deliverable, we disclose any relationship we have with that party.

Full disclosure on the About page.