Home/Connect/Intercom Fin × Stripe
The agent
The system

Make Intercom Fin act in Stripe — safely

Fin decides the refund is warranted — Upware issues it in Stripe, within your policy limits, with a complete audit trail, and posts the entry to your finance system.

Request a demoSee the platform
Governed by policy & RBAC Full audit trail Outbound-only on-prem bridge No rip-and-replace
The gap

Intercom Fin stops at the Stripe boundary

Fin can decide what should happen, but it can't safely make it happen — issuing the refund, updating the account, and recording the outcome in your systems of record are all outside its reach.

Three ways to close it

Only one is governed

Intercom Fin alone
Acts in StripeNo
Governedn/a
Auditedn/a
Custom API glue
Acts in StripeSort of
GovernedYou build it
AuditedYou build it
Intercom Fin + Upware
Acts in StripeYes
GovernedPolicy + RBAC
AuditedEvery action
In practice

From conversation to refund, governed end-to-end

A customer contacts support about a duplicate charge. Fin confirms the error, determines a full refund is appropriate, and passes the action to Upware. Upware's governed execution layer checks the refund against your configured policy limits, issues it directly in Stripe, and records the transaction in your finance system of record — all without a human in the loop for amounts under your threshold. Above that threshold, dual-control kicks in and a finance reviewer gets a queued approval request. Every step lands in the audit trail.

The harness

AI to learn. Deterministic to execute.

Upware learns the process once and encodes it as a mostly-deterministic workflow: system interactions replay exactly, and any LLM steps are wrapped in policy, verification, and audit — so execution stays governed and drift-proof.

O
Observe & Learn
Records the real workflow once, across every system it touches.
G
Generate
Builds the API, data flow, and logic automatically — no integration project.
E
Encapsulate
Wraps it as a governed AI skill over secure MCP or API.
R
Run
Executes deterministically — RBAC, audit, and scale built in.
FAQ

Common questions

How does Upware know what Fin is allowed to do in Stripe?
You define the rules: refund caps, eligible charge types, which Stripe accounts are in scope, and what requires a second approval. Upware's RBAC layer enforces those policies at execution time, so Fin never exceeds what you've sanctioned. If a request falls outside the defined bounds, Upware blocks the action and surfaces it for human review rather than guessing.
Stripe has an API — why do we need Upware at all?
The Stripe API handles the money movement, but that's only part of the problem. Upware wraps Stripe actions in the governance layer your enterprise requires: policy-bounded limits, dual control above thresholds, cross-system orchestration (e.g. posting to your GL after a refund), and a complete audit trail tied back to the original agent conversation. You also get a single governed interface across every system Fin touches, not a bespoke integration for each one.
Is there a risk that the AI will issue refunds incorrectly or hallucinate a decision?
Fin makes the decision; Upware executes it deterministically. Once the action is handed off, there's no open-ended LLM improvising at runtime — Upware replays the exact steps it learned, against the exact Stripe endpoints, under your defined policies. A refund either meets the criteria and goes through, or it doesn't and a human is notified. The audit trail captures every action so your compliance team can review what happened and why.

Connect Intercom Fin to Stripe

See a governed action go from recorded workflow to live in days.

Request a demo