AI SystemsSpecialist recruitment

Candidate Intelligence Platform

Cut candidate cover letter turnaround from 45 minutes to under 2, without losing consistency between consultants.

Built for Quay Group. Interface and data shown here are illustrative, not live client data.

PythonWhisperOpenAIReportLabFastAPI

Outcome

Cover letter generation dropped from 45 minutes to under 2 minutes per candidate, fully automated. Output quality improved because the system captures every detail from the call rather than relying on note quality, and the agency handles higher candidate volumes with the same team.

Product Interface

๐ŸŽ™

Record Call

๐Ÿ“

Transcribe

๐Ÿ”

Extract Profile

โœ

Generate Letter

๐Ÿ“‹

Add to Flipbook

Transcript

awaiting recording...

Extracted Profile

awaiting transcript...

Cover Letter

awaiting profile...

Product interface ยท Representative data

The Challenge

Consultants were spending 45 minutes per candidate writing cover letters from handwritten interview notes. Output quality varied between consultants, key profile details were sometimes missed, and the process bottlenecked the business during high candidate volumes.

Existing Process

Manual note-taking during interview and reference calls, followed by manual drafting of cover letters and profiles with no shared structure or quality baseline.

The Approach

Built a pipeline that records candidate calls, transcribes them, and uses structured extraction to identify experience, specialisations, qualifications, technical skills, career goals and notable achievements. A generation step turns the extracted profile into a professionally formatted cover letter matched to the role, appended automatically to the candidate's flipbook PDF.

What Was Built

A workflow covering interview and reference call processing, transcription, structured candidate data extraction, combining calls, CVs, notes and references into one profile, and generating polished candidate profiles and cover letters with consistent quality across consultants.

  • Interview and reference call processing
  • Automated transcription
  • Structured candidate data extraction
  • Combines calls, CVs, notes and references
  • Candidate profile and cover letter generation
  • Client-ready, consistent outputs

Thimoth's Role

Designed and built the transcription-to-profile pipeline end to end and worked with consultants to tune the extraction and writing style to match the business's tone.

Need something similar built around your business?

Discuss it with CacheStack.

Discuss it with CacheStack