Before you sign

AI Vendor Claim & Contract Scrutiny Kit

Every AI vendor's deck reads great. This kit scores their claims one by one on two things that actually protect you — is it proven, and is it in the contract — and tells you which ones to get in writing before you commit a dollar.

The verdict follows the weaker axis, so a claim that's proven but not in writing can't read “substantiated.” And an ROI or “agentic” promise that's neither proven nor in the contract forces the whole vendor to WALK AWAY — no matter how clean the rest of the pitch.

instant download · .xlsx · yours to keep

This kit is a scrutiny aid for your own vendor evaluation — not legal advice. It scores the vendor's pitch artifact, not people, and does not interpret your contract or determine legal rights. Substantiation of marketing claims is governed by the FTC; have counsel review any agreement before you sign.

The problem

The pitch is polished. The exposure is in what isn't written down.

“Fully agentic.” “Cuts cost 40%.” “Your data is never used for training.” Vendors lead with the claims that close the deal — and the ones that matter most are often the ones with the least behind them. A claim is only as good as the proof you can check and the contract language you can enforce. This kit makes that difference visible per claim, before you sign, instead of after something breaks.

See it work

Five clean claims. One unbacked ROI promise. Still WALK AWAY.

Northwind's uptime, security, data-handling, integration, and support claims all hold up. Its “cuts cost 40% in 90 days” ROI claim has no proof and isn't in the contract — so the high-stakes gate fires and the vendor reads WALK AWAY. Drag the proof or contract marks up on that claim and watch the gate release.

Vendor: Northwind AI Platform

Mark each claim 0–3 on proof and on contract backing. The verdict follows the weaker axis.

WALK AWAY
99.9% uptime SLAmaterial
SUBSTANTIATED
min 3
SOC 2 Type II certifiedmaterial
SUBSTANTIATED
min 2
Your data is never used for trainingmaterial
SUBSTANTIATED
min 2
Native Salesforce + Slack connectors
SUBSTANTIATED
min 2
24/7 human support, 1-hour response
SUBSTANTIATED
min 2
Cuts ticket-handling cost 40% in 90 daysmaterialgated
RED FLAG
min 0
5
substantiated
0
get in writing
1
red flags

High-stakes gate: a ROI or autonomy claim is unproven and not in the contract — the vendor is WALK AWAY no matter how clean every other claim reads. Add any proof or any contract language and the gate releases.

The standard

The weaker axis wins, and the gate only ever holds back.

Two axes, MIN-capped

Verifiability and contract backing, each 0–3. The per-claim verdict follows the weaker of the two — proven-but-unwritten and written-but-unproven both fall short of SUBSTANTIATED.

A high-stakes gate

An ROI or autonomy claim that's both unproven and not in the contract is forced to RED FLAG and forces WALK AWAY — regardless of how the other claims score. Worsen-only: it never promotes a vendor.

Your marks, no benchmark

The verdict comes from your own reading of the pitch and the draft agreement. No baked-in industry multiplier, nothing uploaded, the vendor never contacted.

How it works

Three tabs, one afternoon.

  1. 01

    On Claim Scrutiny, list every claim the vendor makes. Tag its category and whether a purchase decision rests on it.

  2. 02

    Mark each claim 0–3 on proof and 0–3 on contract backing. The verdict computes live.

  3. 03

    Read the Dashboard: each vendor's posture and the one claim to fix first. Take the GET IN WRITING list into the negotiation.

What this is — and isn't

It grades the pitch you assemble — not your systems, not people.

It is

  • A claim-by-claim scrutiny of a vendor's marketed promises before you sign.
  • A way to separate what's proven and enforceable from what's just sales talk.
  • A GET IN WRITING list you carry into the negotiation.

It isn't

  • A contract review or legal opinion — that's the Inbound Contract & SOW Risk Gate plus counsel.
  • A reliability scorecard of a tool you already run on your own usage data.
  • An assessment of any person; it scores the pitch artifact only.

This kit is a scrutiny aid for your own vendor evaluation — not legal advice. It scores the vendor's pitch artifact, not people, and does not interpret your contract or determine legal rights. Substantiation of marketing claims is governed by the FTC; have counsel review any agreement before you sign.

Who it's for

Anyone signing an AI vendor on the strength of a pitch.

For

  • Founders and operators evaluating AI tools without a procurement team.
  • Anyone comparing two vendors whose decks both look airtight.
  • Teams who've been burned by an “agentic” or ROI promise that didn't hold.

Not for

  • Reviewing the contract clauses themselves — use the Contract & SOW Risk Gate.
  • Deciding whether to renew a tool you already use — use the Reliability Scorecard.
  • Anyone wanting a guaranteed outcome; this surfaces risk, it doesn't remove it.

Pairs well with

The full vendor path: scrutinize, sign, govern, renew.

Common questions

The questions buyers actually ask before they sign a vendor.

That scorecard grades a vendor you already use, from your own usage and reliability data, to decide renewal. This kit works one step earlier: it scrutinizes the claims in a vendor's pitch before you sign, on whether each claim is verifiable and whether it's actually in the contract. One is renewal from your data; this is the pitch before you commit.

Get the kit

Know what to get in writing — before you sign.

  • One .xlsx · three tabs · live formulas · worked example.
  • Opens in Excel, Google Sheets, or Numbers.
  • Yours to keep — no login, nothing uploaded.

← Browse all Quick Kits

Sold by RedHub AI LLC · Secured by Stripe · redhub.ai