# TrustedPAI pitch

## One sentence

TrustedPAI gives merchants a policy decision and an evidence record before an
AI agent’s payment is executed.

## Problem

Agentic payment flows introduce delegated authority, unfamiliar protocol
objects and less human context. Existing fraud products remain essential, but
may not verify whether an agent’s action matches a signed mandate across new
payment rails.

## Product

One API call normalises a supported payment authorization, verifies available
protocol evidence, applies the merchant’s policy and returns APPROVE, HOLD or
REJECT with reasons. A customer-scoped evidence record preserves the decision
context.

## Why now

AP2, ACP, x402 and MPP show that agent-led payments are becoming a distinct
integration surface. The standards are still changing, so buyers need a safe
way to observe and control them before committing to a single rail.

## Customer

Start with merchant platforms, PSPs, procurement automation providers and
agentic SaaS products carrying acceptance-side payment risk.

## Pilot

Six to eight weeks in shadow mode. TrustedPAI does not execute payments or
replace the client’s PSP or fraud system. The pilot measures coverage,
review usefulness and willingness to pay.

## Differentiation to prove

Neutral multi-rail policy, protocol-aware verification and evidence that is
useful for a dispute or review. This is a hypothesis to validate, not a claim
to make without a paid design partner.

## Ask

One scoped workflow, a technical owner, anonymised outcome feedback and a
commercial decision at the end of the pilot.