Growing Pains
A SaaS startup in Glasgow was scaling fast — from 20 to 45 people in under a year. Their HR team consisted of exactly one person, and she was responsible for everything: job specs, posting, screening, scheduling interviews, offer letters, onboarding.
"I was getting 150 to 200 applications per role," she said. "Screening CVs alone was taking me two full days per position. And we had four open roles at any given time. The maths just didn't work."
The Screening Pipeline
We deployed OpenClaw on their existing cloud infrastructure and built a recruitment screening pipeline:
- CV parsing — the agent ingests CVs from the applicant tracking system, extracting key information: experience, skills, education, location, and salary expectations.
- Requirement scoring — each candidate is scored against the role's requirements on a weighted matrix. Must-haves are non-negotiable; nice-to-haves add bonus points. The scoring criteria are defined by the HR manager for each role.
- Shortlist generation — the top 15-20% of candidates are compiled into a ranked shortlist with a one-paragraph summary explaining why each one scored well.
- Outreach drafting — personalised response emails are drafted for three categories: shortlisted (interview invitation), promising (talent pool), and rejected (polite decline with feedback).
What Changed
Screening time dropped by 75%. What used to take two full days now takes half a day — most of which is the HR manager reviewing the shortlist and making final judgement calls, not reading through hundreds of CVs.
"The agent doesn't decide who we hire. It decides who I should spend my time looking at. That's a massive difference. Instead of reading 200 CVs to find 20 good ones, I'm reading 20 good ones and deciding which 8 to interview."
The Quality Question
The HR manager ran a parallel test during the first month — screening one role manually and the same role with OpenClaw. The overlap between her shortlist and the agent's shortlist was 85%. The 15% difference? Mostly borderline candidates that could reasonably go either way.
"I was worried it would miss good candidates. It didn't. If anything, it caught a few people I might have overlooked because their CV formatting was unusual but their actual experience was strong."
The candidates noticed a difference too — faster response times and more personalised rejection emails. The employer brand improved without anyone setting out to improve it.