What is an AI twin used for? Seven real jobs

We've defined what an AI twin is — an AI version of a real person, grounded in their own material. The more useful question is what one is for. Here are the seven jobs we actually see twins doing, roughly in order of how often they're the reason someone builds one.

1. The expert link-in-bio#

Instead of a Linktree of destinations, one link where your audience can ask you things — and get your real answers, cited from your real work. For anyone whose inbox is their audience's search engine, this is the default use.

2. Pre-sales answering#

Coaches, consultants, and freelancers put their twin where prospects are deciding: methodology questions answered from your writing, prices quoted verbatim from your services list, and stuck prospects converted into leads that arrive with their question attached.

3. Office hours that never close#

Professors, community leads, and mentors use a twin as the always-open first stop: the twenty logistical and conceptual questions every cohort asks get answered instantly, and the genuinely novel ones get forwarded to the human. The twin handles frequency; the person keeps the interesting part.

4. Product support, in a neutral voice#

The same grounding engine pointed at a product's docs instead of a person's writing: it answers how-do-I questions on your own site, never sells, and hands what it can't answer to your inbox with the user's email — nobody dead-ends. (This is a distinct mode — support twins — because a product should not have a personality crisis.)

5. Internal team knowledge#

An invite-only twin trained on your project notes and decisions, for colleagues. The fourth "where are the Meridian call notes?" gets answered without you — with the note itself cited, and a hard rule that the twin reports what you wrote but never decides on your behalf.

6. The course companion#

Course creators load the curriculum plus every Q&A thread they've answered. Students ask at the moment of confusion instead of waiting for the weekly call — and the gap queue shows the creator exactly which lesson needs a better explanation.

7. Candidate and audience screening#

A twin loaded with your open roles, your "how we work" docs, and your FAQ answers the fifty identical questions every posting attracts — and the questions it couldn't answer tell you what your job ad forgot to say.

The pattern underneath#

Every one of these is the same loop wearing different clothes: repeated questions → grounded answers with receipts → honest gaps → the human answers once → the twin gets permanently better. If some version of that loop is your week, a twin is buildable from what you already have — or poke a finished one first.