Moosa Memon
← Index of work

Candidate screening that stays accurate after launch

RecruitFlow AI: a conversational screening agent grounded in a RAG knowledge base, automated nurture for candidates who go quiet, and a recruiter-facing pipeline and insights layer. Built for staffing teams that need to scale outreach without adding headcount.

Type
Agents
Stack
FastAPI / Claude API / RAG / n8n / SQLite
Status
Complete; runs end to end with no credentials on a local embedder and templated replies

Problem

Recruiting teams lose candidates in the gap between application and first human conversation. A screening bot fixes the speed problem and creates a new one: answers drift out of date, breakages go unnoticed, and the conversation data nobody looks at is where the actual insight is.

System

A chat widget backed by a screening agent that answers role and process questions from an ingested knowledge base, captures screening answers into a structured pipeline, and hands off to a recruiter with the transcript. Nurture sequences re-engage candidates who go quiet. An insights layer turns conversation data into what recruiters can act on: where candidates drop off, which questions the knowledge base can’t answer, which roles generate the most confusion.

Worth knowing

The project is scoped around ownership after launch, not just the build: the “unanswerable questions” report exists so someone can keep the knowledge base accurate, and the whole system runs on local stand-ins so it can be regression-tested without spending on a model. That’s the shape of the Systems Partner retainer on this site, built as software.

Have a workflow that looks like this?

Most of these start as a messy, manual process someone got tired of. Twenty minutes is usually enough to sketch how I'd approach yours.