Sep 9, 2025
AI has not simply accelerated recruiting. It has rewritten parts of the playbook. Recruiter intuition still drives hiring, and following the old rulebook now works against building a high-performance team. Eight rules changed: candidate AI use, identity verification, scorecards, agent-run chores, speed, compensation, published fairness, and privacy.
Teams have flopped despite hiring "A-players" on paper, because visionary solo players clash and customers get an incoherent outcome.
Company6 min to read
Candidate fraud detection software for remote interviews detects and blocks fraud in real time, saving companies from bad hires. Learn how.
Company6 min to read
6 common recruitment challenges costing companies top talent in 2026, and how a unified agentic AI platform fixes the entire hiring pipeline.
Company6 min to read
Meta has stated it is considering allowing candidates to use AI on some coding tests.
Fraud has moved past proxy interviews to deepfake lip-syncing, so identity verification is no longer optional.
High-performance teams run interview sprints, making decisions the same day or within days.
"Hiring people is an art, not a science, and resumes can't tell you whether someone will fit into a company's culture." — Howard Schultz, founder of Starbucks
Recruiters have relied on intuition, process and hustle, then on data, automation and better tooling. AI is doing something different from acceleration. It is rewriting parts of the playbook.
# | Rule | The change |
|---|---|---|
1 | Design for candidates working with AI | Test how candidates think and adapt, rather than blocking the tools |
2 | Put a trust layer in the stack | Identity verification becomes standard, delivered humanely |
3 | Turn JDs into scorecards automatically | AI parses the job description into competencies and evaluation criteria |
4 | Let agents run the chores | Scheduling, reminders, sourcing and note summaries move off recruiters |
5 | Make speed a promise, not a hope | Interview sprints with same-day or within-days decisions |
6 | Calibrate compensation to reality in real time | Live market signals instead of static bands |
7 | Publish fairness, don't bury it | Report pass rates, adverse impact, response time and more |
8 | Make privacy a trust builder | On-device options, minimal retention, clarity on data use |
Assume they already do, and test how they think rather than whether they had help.
Candidates are preparing with ChatGPT, Claude and other copilots. Blocking the tools is not always the right response. Meta has stated it is considering allowing candidates to use AI on some coding tests.
The process should test how candidates think, adapt and learn in real time. Three formats do that:
Work samples. Tasks that mirror the role.
Pair exercises. Collaboration with a future teammate in real time, which shows how they communicate, push back and adapt.
Live problem-solving. An ambiguous scenario with missing data, judged on how they clarify, structure and iterate rather than on reaching a "correct" answer.
The point is not to trip candidates up. It is to see whether they think clearly when the terrain is uncertain.
A trust layer is identity and integrity verification built into the hiring process rather than bolted on afterwards.
Fraud is no longer a rare edge case. Proxy interviews and deepfake lip-syncing have both arrived, and the cost of a wrong hire is high enough that identity verification cannot stay optional. So is the reputational risk of being duped.
The solution also cannot be draconian. In practice that means three things:
Humane identity checks. Biometric or multi-factor verification that confirms identity while staying quick, transparent and respectful.
Test integrity safeguards. Fraud detection that flags second screens or suspicious behaviour, with a clear explanation of what is monitored and why. Candidates should understand it is about fairness rather than surveillance.
Work provenance validation. Confirming that work samples belong to the candidate, through real-time exercises, pair tasks or version tracking. This shifts the focus from suspicion to authenticity.
A process that balances rigour with humanity signals the kind of workplace people want to join.
Have AI parse the job description into competencies, required skills and behavioural markers, then map those into specific evaluation criteria.
Panels too often walk into interviews with a vague job description and their own gut feel. The result is unstructured conversation, subjective judgement, and candidates judged on personality fit or surface rapport rather than demonstrated capability.
With the framework in front of them, the panel stops debating what they are looking for. Three advantages follow:
More structured interviews. Every candidate is asked similar questions tied to core competencies, which creates consistency and avoids bias-prone small talk.
Less bias, more data. Rubric-driven scoring creates a paper trail showing why one candidate rated higher on collaboration, problem-solving or technical skill. Bias does not disappear, and it becomes easier to spot when the data is structured.
Clear comparisons. Panels weigh strengths and gaps against agreed criteria instead of recalling who "felt right".
Training matters alongside the tool. Panels have to learn to anchor feedback in observable evidence rather than impressions, or the rubric becomes a checkbox exercise instead of a living framework.
Scheduling, reminders, sourcing from skills graphs, and summarising interview notes.
Recruiters should be selling the mission rather than chasing calendar slots. Moving the chores frees them to calibrate culture fit, sell the vision and make the final call.
Because candidates do not wait, and a slow process leaks in three separate ways.
High-performance teams run interview sprints: tight loops with decisions made the same day, or within days at worst.
Speed protects your pipeline. The longer a process runs, the leakier it gets. Fast loops keep the best prospects engaged and signal momentum.
Speed signals respect. A drawn-out process tells the candidate they are not a priority. A streamlined one says their time is valued.
Speed is brand equity. Candidates share experiences on Glassdoor, Reddit and within their professional networks. A reputation for slow, opaque hiring damages the brand the way a clunky product experience does.
Against live market signals, continuously.
If every single offer needs an exception, your pay ranges are fiction. The best teams connect to live market signals, update benchmarks continuously, and empower talent acquisition teams to make calls within clear guardrails. This is not about overpaying. It is about keeping credibility with candidates and reducing negotiation fatigue.
The ones that would change behaviour if they moved.
Most companies claim they are fair. Few show the receipts. Candidates are open to AI in recruiting and want assurance of fairness. High-performance teams track and publish:
Pass rates by stage
Adverse impact
Time to response
90-day productivity
Manager NPS
Internal fill rates
And they retire metrics that do not actually change behaviour.
Because candidates are more privacy-aware than ever, and an invasive process loses them.
The new baseline: on-device options where possible, minimal data retention, and absolute clarity on how candidate data is used. Privacy is a signal of respect as much as a compliance obligation.
Recruiting in the AI era is not about bolting tools onto an old process. It is redesigning for trust, speed and fairness at scale. The recruiters who thrive will use AI to create more humane, transparent and decisive hiring journeys.
Do you have a new rule we have not considered?
Assume they already are, and design around it. Meta has stated it is considering allowing candidates to use AI on some coding tests. Work samples, pair exercises and live problem-solving on ambiguous scenarios test how a candidate thinks and adapts, which AI assistance does not fake well under real-time probing.
A trust layer is identity and integrity verification built into the hiring process: humane biometric or multi-factor identity checks, fraud detection with a clear explanation of what is monitored and why, and validation that work samples genuinely belong to the candidate.
AI parses the job description into competencies, required skills and behavioural markers, then maps them to specific evaluation criteria. The panel gets a rubric instead of a debate about what they are looking for, which produces structured interviews, a scoring paper trail, and comparable candidates.
High-performance teams run interview sprints and make decisions the same day, or within days at worst. Speed protects the pipeline from leaking, signals respect for the candidate's time, and builds brand equity, because candidates share slow experiences on Glassdoor, Reddit and in their networks.
Pass rates by stage, adverse impact, time to response, 90-day productivity, manager NPS and internal fill rates. Publishing them is what converts a fairness claim into evidence, and metrics that do not change behaviour should be retired rather than reported.
Against live market signals rather than static annual bands. If every offer requires an exception, the ranges are fiction. Continuous benchmark updates plus clear guardrails for talent acquisition teams preserve credibility with candidates and cut negotiation fatigue.
Because candidates are more privacy-aware than ever and will walk away from a process that feels invasive. The baseline is on-device processing where possible, minimal data retention, and clarity about how candidate data is used.
Next step: Answer the security review before it starts. Get the compliance pack
Interviews you can prove
Screening, interviews, and live identity verification in one platform. Built for high-volume technical hiring.
Book a demoOriginally published on Substack.
Automated resume screening parses resumes, compares them to job requirements, and ranks candidates, so recruiters review only top talent.