The verdict in three sentences
For a temporary staffing agency with 1,500 active temps in London, the bottleneck is CV screening, not sourcing. An AI that reads CVs, extracts skills, location and availability, then proposes a ranked shortlist per assignment cuts screening time by about five while leaving the decision with the consultant. With AI usage costs of roughly GBP 30 to 150 a month and a medium custom application, the investment pays back within months through faster placements.
The hidden cost of manual screening
A London agency receives applications from job boards, its website, LinkedIn and email, in every format. At 6 minutes per CV and 4,000 CVs a month, screening takes 400 hours a month, close to three full-time consultants doing nothing but reading.
| Metric | Manual screening | AI-assisted screening | Difference |
|---|---|---|---|
| Average time per CV | 6 min | 1 min of human review | divided by 6 |
| Screening hours a month (4,000 CVs) | 400 h | 67 h | 333 h freed |
| Average time to shortlist | 3 to 5 days | under 24 h | 2 to 4 days saved |
| Average time to fill a temp role | 4 to 7 days | 48 to 72 h | 48-hour target achievable |
| Roles lost for lack of a timely candidate | 10 to 15% | 3 to 5% | 7 to 10 points recovered |
| Usable CVs resurfaced from the database after 6 months | about 20% | over 80% | a genuinely active talent pool |
These are 2026 orders of magnitude for an agency of this size. The main gain is not time saved but roles won because the agency submits first.
How matching actually works
- Unified intake: job board feeds, website form, email and LinkedIn applications land in one database, with OCR for scanned documents.
- Structured extraction: the AI extracts job titles, years of experience, qualifications, certifications (CSCS cards, forklift licences, DBS status), right-to-work status flags for human verification, postcode and availability.
- Assignment matching: for a client request (say 12 warehouse operatives in Park Royal for Monday), the engine scores skills, commute time, past assignments and ratings.
- Mandatory human review: the consultant approves the shortlist, can remove or add profiles, and every decision is logged.
- Candidate re-engagement: SMS or WhatsApp availability checks with one-tap replies.
Human review is not optional. UK GDPR Article 22 restricts decisions based solely on automated processing that significantly affect individuals, and the ICO expects a DPIA for AI recruitment tools. If you place candidates with clients in the EU, the EU AI Act classifies AI used for recruitment and candidate evaluation as high-risk, with obligations on human oversight, logging and bias monitoring phasing in from August 2026.
Budget: AI usage, application and running costs
| Item | 2026 order of magnitude | Comment |
|---|---|---|
| AI usage (extraction and scoring of 4,000 CVs a month) | GBP 30 to 150 a month | Depends on model and CV length |
| SMS and WhatsApp messaging | GBP 20 to 80 a month | Depends on volume |
| Kolonell simple application | 2,500,000 to 4,000,000 FCFA (EUR 3,810 to 6,100) | Intake, extraction, searchable database |
| Kolonell medium application | 4,000,000 to 7,000,000 FCFA (EUR 6,100 to 10,670) | + assignment matching, scoring, re-engagement, audit trail |
| Kolonell complex application | 7,000,000 to 15,000,000+ FCFA (EUR 10,670 to 22,870+) | + client portal, timesheets, multi-branch |
| Maintenance | 35,000 / 75,000 / 120,000 FCFA a month (EUR 53 / 114 / 183) | Basic, Standard or Premium |
| Hosting and migration of historic CVs | On quote | Depends on volume and formats |
A mainstream ATS with an AI add-on costs in 2026 order of magnitude GBP 40 to 120 per user a month, plus the add-on. Custom work makes sense when your matching rules (shift patterns, client-specific compliance, commute limits) are your edge.
Bias to watch
A poorly tuned model can systematically exclude profiles: women for warehouse roles, older candidates, overseas qualifications. Under the Equality Act 2010 the agency remains liable. Three safeguards: remove age, gender and photo from scoring, audit shortlist distribution every quarter, and let any candidate request a human review.
Mini case study
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
Rachel, managing director of a London staffing agency with 1,500 temps, receives 4,000 CVs a month and employs 4 consultants. She loses about 12% of 250 monthly assignments for lack of a candidate within 48 hours, with an average gross margin of GBP 450 per assignment.
- Assignments recovered: 250 × 8 points × GBP 450 = GBP 9,000 of margin a month.
- Time freed: 333 hours a month, redirected to business development.
- Cost: medium application at 6,000,000 FCFA (about EUR 9,150), AI usage around GBP 100 a month and Standard maintenance at EUR 114 a month.
Year one costs roughly EUR 9,150 + 12 × (EUR 114 + about EUR 115) ≈ EUR 11,900, against over GBP 100,000 of recovered margin. Payback comes in the first two months.
FAQ
Is AI CV screening legal in the UK?
Yes, provided a human makes the final decision, candidates are informed and a DPIA is carried out. The ICO's 2024 audit of AI recruitment tools sets out practical expectations worth following.
Does the EU AI Act apply to a London agency?
It applies if the system's output is used in the EU, for example when placing candidates with EU clients. In that case high-risk obligations apply, including logging, human oversight and accuracy monitoring.
Does the AI score replace the interview?
No. The score ranks, the consultant approves the shortlist and interviews. In practice they review 10 to 15 profiles per assignment instead of 80.
What happens to costs if volume doubles?
AI usage scales with volume: at 8,000 CVs a month, allow roughly GBP 60 to 300. The application itself does not need rebuilding.
How long until the tool is live?
Allow 10 to 14 weeks for a medium application, including 3 weeks of trials on live assignments to tune the scoring weights.
Let's scope your project. We will define with you the scope (intake, extraction, matching, human review) and the go-live plan for your agency. Detailed quote within 48 h. WhatsApp +221 77 596 93 33.
Mohamed Bah
Fondateur, Kolonell
Passionate about digital and entrepreneurship in Africa, Mohamed has been helping Sénégalese businesses with their digital transformation since 2020. Founder of Kolonell, he believes every SME deserves a professional and accessible online présence.
