200 applications. 4 hours of reading.
Now 25 minutes — for the ones that matter.
Bridgepoint Talent's recruiters were drowning in CVs. We built a screening engine that scores, routes, and responds — so humans only read the top 20%.
A recruitment firm where the bottleneck
was the recruiters themselves.
Bridgepoint Talent is a specialist recruitment firm placing mid-to-senior professionals across finance, operations, and risk management in the US and UK. With 40 recruiters managing between 80 and 120 active roles at any given time, the volume of incoming applications had quietly become unmanageable.
Each role received between 80 and 250 applications. Recruiters were expected to screen them all, respond to relevant candidates, brief the client, coordinate interviews, and close placements — while simultaneously managing their pipeline. Most were spending the first two to three hours of every day just reading CVs, before they'd spoken to a single person.
The problem wasn't effort. The problem was that 80% of screening work was low-judgment pattern matching — does this person have the right experience, seniority level, and sector background for this role? That's exactly the kind of task that doesn't need a human. And by keeping humans on it, Bridgepoint was bottlenecking the parts that actually do.
They came to us with one ask: give recruiters their mornings back. What we built gave them a lot more than that.
Five reasons top candidates
were slipping through.
None of these were caused by bad recruiters. They were caused by a process designed for a volume it could no longer handle.
How the screening engine is wired.
Make.com handles the orchestration. Every tool in the stack has one job — and hands off cleanly to the next. No manual data movement between systems.
How every application gets
scored and routed.
OpenAI scores each application against a role-specific rubric. The score determines the response. No human decides who gets a reply — the criteria do.
Three phases. One continuous pipeline.
Built and tested in three weeks. The testing phase deliberately included edge cases — part-time roles, career changers, internal referrals — to make sure the scoring held up under real conditions.
Application intake + AI scoring engine
Every application source — Typeform, LinkedIn Easy Apply, and direct email — was routed into a single Make.com scenario. The first module parses the incoming data and extracts the CV. That gets passed to OpenAI GPT-4o with a role-specific prompt that scores the candidate across five dimensions: relevant experience, seniority match, sector background, tenure pattern, and any flagged deal-breakers defined by the client brief.
The rubric is set once per role when the job is opened — recruiters fill in a structured brief in Airtable that Make.com reads to build the scoring prompt. Different roles weight different dimensions. A risk analyst role weights sector experience heavily. A generalist ops hire weights adaptability. The same infrastructure handles both.
Score-based routing + automated outreach
Once the score comes back, the Make.com router fires one of three paths. Candidates below 60 receive a personalised rejection via SendGrid — not a generic bounce, but a message that references the role they applied for and their approximate experience level. Candidates between 60 and 79 go into a warm hold pool in Airtable with a holding email. Candidates scoring 80 or above are fast-tracked.
For fast-tracked candidates, Twilio sends an SMS within 90 seconds of their application — something along the lines of confirming receipt and asking for availability for a brief call. If they respond, a Calendly link fires. The recruiter gets a Slack notification with the candidate's Airtable card linked. By the time the recruiter looks at it, the interview is often already booked.
Interview logging + automated client briefing
After the recruiter conducts an interview, they update a simple Airtable form — suitability rating, key notes, candidate's stated salary expectation and availability. That triggers the final Make.com flow: a structured client briefing document is generated automatically, pulling from the candidate's Airtable record and the recruiter's post-interview notes, formatted into Bridgepoint's house style, and emailed to the client contact.
What used to be 45 minutes of formatting in Word is now a 90-second automated output. The recruiter's job is the interview and the notes. Everything else is automated.
What changed after go-live.
Measured across the first 90 days in production, compared to the same period in the prior year.
Seven tools. One pipeline.
What's your team doing
that a machine should be?
Tell us the process. We'll tell you whether it can be automated, what it would take, and what it should cost — in 30 minutes, before you commit to anything.
Shivam or Sonam on the call. Not a junior consultant.