Mojo the Recruiter

works your pipeline · A university's team

Working today

The challenge

Recruitment has become an arms race nobody is winning. Applications get easier to send every year, so volume climbs while the share of genuinely eligible candidates falls, and AI has made a polished enquiry cost nothing to produce. Meanwhile the students you would most want are drowning in outreach from institutions that never checked whether they qualify. So you are left with an enquiry list four thousand names long, a quiet certainty that perhaps forty genuinely fit, and a team that cannot read four thousand files. So you market to all of them, hope, and find out in September.

What it does about it

Now imagine having a colleague who has already read all four thousand enquiries against your actual entry requirements, and can simply tell you which forty matter. The Recruiter works through the enquiries you already have, sees only your own institution's, and answers in plain questions — who fits, who stalled, who is close.

How it works

It works inside your own organisation and only over students who chose to be seen, and every figure comes from your live pipeline rather than an estimate. It can read whether a student meets your published requirements, and it cannot touch what that student is shown about themselves — a separation enforced by a build check rather than by policy.

What you get

Your enquiries as they really are, only ever your own, and answerable in plain questions — so effort goes to the students who could actually enrol. It works alongside your Forecaster on the same pipeline, and reaches students only through the consent they gave — never around it.

The machinery

• Producesodin_runs (engine 'university_chat') — read-views over consented leads; no student-file writes by design
• Can act whenlib/odin/engines/university-chat.ts (the prompt that names it) · app/api/u/odin/chat/route.ts

Read from the same manifest the product runs on, so this cannot describe machinery that does not exist.