Your consultants are typing when they should be talking
AI helps a recruitment agency in three places: sourcing from the database you already own, turning screening calls into structured notes that survive staff turnover, and keeping every candidate informed without consultants typing the same message forty times a day. We build these systems for UK agencies with the GDPR work designed in from the start, not bolted on after.
Rather build it yourself? We will show you where to start, free.
What we would build
- 01A CRM that answers in plain English
- 02Screening-to-shortlist engine
- 03Candidate comms that never go quiet
Where the fees leak
Two activities win fees. Everything else is overhead.
Screening write-ups
Every call needs a structured note in the CRM. Most get a rushed paragraph, or nothing, and the candidate knowledge your agency is built on walks out of the door with each leaver.
Re-keying and reformatting
Candidate details copied between systems, CVs reformatted to each client’s template, shortlist summaries assembled by hand. Pure reshaping of information you already hold, which is exactly what a model is reliable at.
Candidates left in silence
Candidates who hear nothing go quiet, take counter-offers or sign with a competitor. And they remember which agency ignored them.
The unreachable database
Years of candidates sit in your CRM, findable only if whoever wrote the original note used the right keywords. You are paying job boards for people you already know.
What we would build
What we would build for your agency
Three systems with the same split in every one: AI handles reading, writing, formatting and chasing. Consultants handle judgement, relationships and accountability.
A CRM that answers in plain English
The biggest sourcing win is not a new database. It is the one you already own, made searchable.
- Ask for "embedded software engineers near Leeds who interviewed well but lost at offer stage" and get a ranked list with a sentence of evidence per name.
- Long lists re-ranked against a live brief, in an order a consultant can check rather than trust.
- Stale records revived: the model reads an old CV, infers where the candidate likely is now, and drafts a re-engagement message for approval.
- Nothing silently excludes anyone. A model that removes a person from consideration is making a decision you are accountable for, so it does not get to.
Screening-to-shortlist engine
The highest-return workflow for most agencies: the largest block of admin, attacked directly.
- Calls recorded with the candidate’s clear consent and transcribed automatically.
- A structured note in your house format lands in the CRM before the consultant finishes their coffee: situation, motivation, salary, notice, flags against the spec, next action.
- The consultant corrects and approves every note. That step is not optional; the note informs a placement decision.
- The same source material drives CV formatting to client templates, shortlist summaries and interview prep packs.
Candidate comms that never go quiet
Automate the messages where speed beats personality. Keep humans where the relationship is the point.
- Acknowledgements, interview confirmations, reminders the day before, and status updates when a process stalls, so silence never reads as rejection.
- Post-placement check-ins at the end of week one and month one, sent every time, not when someone remembers.
- Offers, rejections of final-stage candidates and counter-offer conversations stay with consultants. Always.
- Compliance chasing in the same machine: right-to-work documents requested and tracked, expiring certifications flagged.
Do it yourself
You could build this yourselves. Here is how to start.
The screening and database wins do not need a platform purchase. They need structure, consent and a fortnight of discipline.
The guides are free and they do not hold anything back. If you get partway and want it finished fast, or built properly first time, that is the other reason this page exists.
How the free route works- 01Write your house note format down once: situation, motivation, salary, notice, flags against the spec, next action. The format is the system; the AI just fills it.
- 02Record one screening call with the candidate’s clear consent, transcribe it, and have AI draft the structured note. Compare it with what would have reached the CRM otherwise.
- 03Dedupe and tag the CRM before buying anything. Ranking is only as good as the records underneath it.
- 04Set up the silence-killers: acknowledgement, interview confirmation, day-before reminder, and a status note when a process stalls.
- 05Before anything candidate-facing is automated, do the GDPR work: impact assessment, privacy notice, and a real human review on every rejection.
Straight answers
Questions we get from this industry
- Is using AI in recruitment legal under UK GDPR?
- Yes, done properly, and the regulator has been explicit about what properly means. The ICO audited AI recruitment tools in 2024 and issued almost 300 recommendations. The practical list: a data protection impact assessment before deployment, privacy notices that tell candidates what the AI does, a real human in any decision that rejects someone, minimised data with a genuine retention period. We do this work as part of the build, not as an afterthought.
- Does the AI decide who gets rejected?
- No. Solely automated decisions with significant effects engage Article 22 of UK GDPR, and a consultant skim-clicking approve does not count as meaningful review. Anything that affects a candidate’s chances keeps a person in the loop by design.
- Our CRM data is a mess. Does that kill the idea?
- It is the first thing to fix, not a reason to stop. Ranking is only as good as the records underneath it, so deduplication and structure come before any model. Matching quality depends on data structure, not model cleverness: tag the records properly and simple ranking works.
- Will clients care that we use AI?
- Increasingly they ask, because they are running their own AI governance reviews. An agency that can explain exactly how candidate data is processed wins terms. Compliance done well is a sales asset, not a tax.
- Where should an agency start?
- One workflow, usually screening notes or candidate status updates. Automate it end to end with consultant approval built in, measure hours saved per week, then expand. The second workflow reuses most of the first one’s plumbing, which is where the returns stack up.
Tell us your worst admin job
Thirty minutes, no pitch deck. We will tell you which workflow to automate first, what it depends on, and what we would refuse to automate at all.
Or email dc@operosus.com and tell us what is eating your week.