ArrayTalent: 288,000 candidates scored, and an estimated £447k of sourcing time saved
- Industry
- tech recruitment
- What we built
- LinkedIn sourcing, AI scoring against each role, contact finding, Lemlist outreach and a reply inbox, in one dashboard the whole team uses
- Result
- 287,882 candidates scored in six months, where each sourcer could add 50 to 60 a day by hand
The problem
ArrayTalent hires software, infrastructure and product engineers for tech companies. Their sourcers found candidates on LinkedIn, added them one by one to a contact tool, then judged each profile against the role by eye.
Each sourcer added roughly 50 to 60 candidates a day. As Tom Wetwood, the owner, put it on our first call: "Everything's basically manual." When a client wanted more hires, the only answer was another sourcer.
What we built
One system that runs the sourcing end to end, with the team approving anything that matters.
- Searches run from the dashboard and pull matching LinkedIn profiles in bulk
- AI scores every profile against the role, with written pros and cons, and learns from each approve or decline the team makes
- Shortlisted candidates get contact details found automatically and go straight into Lemlist outreach
- Replies land in one inbox, sorted, with the next step drafted for a recruiter to send
The result
- 287,882 candidates scored and 105,674 shortlisted, across 5,161 searches
- Work that would have taken about 29,900 hours by hand
- An estimated £447k in sourcer time, after scraping costs
That's the dashboard's own estimate, tracked on every run. It uses ArrayTalent's sourcer rate of £15 an hour, plus our assumption of 5.5 minutes to review a profile by hand (from their sourcers' 50 to 60 a day) and 2 more minutes to enrich each shortlisted one.
The quality holds up too. One candidate the system scored 92 was offered the role, and the AI-drafted replies to candidates hold their own:
"Where does Reply with AI get its context from? It's just really blown me away. A candidate from Meta said something along the lines of, 'Why should I be interested in working for Fluidstack over Meta?' And the response was perfect."
Tom Wetwood, Owner, ArrayTalent
"I'm actually really impressed. I've added some random job specs in from other companies and the rubric pulls in relevant target companies (competitors, companies that offer similar services etc.)"
Tom Wetwood, Owner, ArrayTalent
Could this work for your team?
If your recruiters spend their days finding and screening profiles by hand, the same approach applies. See how AI recruiting works, or book a call.
