Placewise AI

Placewise AI is a voice-native recruitment screening platform. It phones candidates, holds a real conversation, scores the answers against a rubric you control, and hands recruiters a shortlist instead of a calendar full of repeat calls. I am the co-founder and CEO, and I wrote the system.

The problem with first-round screening

Screening is the most repetitive work in hiring and the tightest bottleneck in it. A recruiter spends most of the week on calls that follow the same script: confirm what the CV already says, ask the same ten questions, chase the people who did not pick up, then write notes nobody reads twice. Around 70% of a recruiter's time goes to screening, roughly 40% of booked calls end in a no-show, and the average time to hire stretches past three weeks. None of that work is judgement. All of it delays the part that is.

How the voice agent works

The agent calls the candidate and runs the interview as a conversation, not a phone tree. It asks follow-up questions generated from what the candidate actually said, so a vague answer gets probed instead of scored blind. If nobody picks up, it retries on its own schedule rather than dropping the candidate.

Every interview is scored against a rubric the hiring team defines, producing a hire or no-hire recommendation with the reasoning attached, so a human can audit the call rather than trust a number. Campaign-level analytics roll that up per role: completion rates, conversion, where candidates drop out.

The rest of the pipeline

An agent alone does not replace an ATS, so Placewise carries the pipeline too:

• Resume parsing that extracts skills, experience and contact details from whatever format the candidate sent
• Bulk import and application tracking with tagging and bulk operations
• Branded career portals per company, so the candidate never sees our name
• GDPR-compliant data exports, because recruitment data is personal data
• A company knowledge base the agent answers from, using retrieval over each company's own documents, so answers stay brand-accurate and policy-consistent

Under the hood

Python and FastAPI on the backend, Next.js on the front, PostgreSQL and Redis underneath, with LLM orchestration driving the conversation and RAG grounding the knowledge base. It is multi-tenant from the first commit: role-based access for owners, admins, recruiters, interviewers and viewers, organisation scoping on every query, and department management for larger teams.

Free, Starter at £49 a month and Professional at £149 a month, with an enterprise tier for anyone who needs it. Teams running it report around 70% of manual screening time gone.

Lessons

1. Voice AI lives or dies on the unglamorous parts. The retry logic that calls a no-show back politely at the right hour converts better than any prompt tweak.
2. Multi-tenant from day one costs a week up front and saves a rewrite later. RBAC, org scoping and GDPR exports are features, not chores.
3. Selling it and building it at the same time is the fastest feedback loop I know. Every sales call rewrites the backlog.