The problem
Agency recruiters get paid on placements, then lose the day to everything that is not one: screening fifty resumes for a single req, rebuilding a Boolean string from scratch, retyping a candidate write-up the account manager can actually forward. The tools built for them are full ATS platforms with a setup tax a small desk never earns back, and generic AI writers hand back markdown that turns into literal asterisks the moment it is pasted into Outlook.
What we built
Candid.ai is a lightweight ATS with Claude wired into every step. Each req is a project: paste the job description, batch-score the resumes, and every candidate comes back with a score out of 100 across five weighted categories and a Strong Submit through Pass call. Notes from the hiring manager's kickoff override the job-description boilerplate and produce a separate manager fit score. From one candidate card a recruiter generates tiered Boolean variants, pre-screen probe questions, an internal write-up, a client submittal and a contact-scrubbed Word resume, all in plain text built to survive a paste into Outlook. Rejections and placements are captured back per client, so the desk builds its own record of what that client actually hires.
The result
Screening, submittals and spread in one desk.
Candidate scoring, Boolean variants, internal and client submittals, pre-screen probes and contract margin tracking all live in the same workspace as the pipeline.
Also shipped
- Contract spread tracker with an all-time-high watermark
- Aria, an AI chat agent for enterprise lead capture
- Skill assessments with trust scoring and PDF reports