# Selling Trust

**Track:** Go-to-Market in the Model Era — AI for Entrepreneurship — complete (29)
**Framework / surface:** venture strategy
**Level:** Intermediate
**Prerequisites:** Ownership, Licensing & Provenance

**In one line:** Privacy, liability, and accuracy commitments — the enterprise buys the guarantee, not the demo.

## Theory, aesthetics & inspiration

The enterprise does not buy the demo; it buys the guarantee. Objections to AI purchases are by now a stable catalog — where does our data go, is it trained on, who is liable when the model is wrong, how do you know how often it is wrong — and each has become a sellable artifact: data-processing agreements and no-training commitments, SOC 2 audits, accuracy SLAs backed by published eval results, and insurance-shaped contract language. The liability question stopped being hypothetical in 2024 when a Canadian tribunal held Air Canada responsible for a bereavement-refund entitlement its website chatbot invented: the company argued, in effect, that the bot was a separate entity responsible for its own actions, and lost. For startups this catalog is an opening, not a burden — a prepared founder walks procurement in days while unprepared competitors stall for quarters, and trust artifacts compound like code. Sell the ceiling too: buyers adopting you are underwriting the agent you will ship next year, and they need to trust the trajectory, not just the release.

**Founder question:** Which artifact — audit, SLA, or agreement — would unblock your hardest procurement conversation?