Home/Connect/Decagon × Salesforce
The agent
The system

Make Decagon act in Salesforce — safely

Decagon closes the conversation; Upware closes the loop — writing the refund, case update, or stage advance into Salesforce with a full audit trail and no manual handoff.

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

Decagon stops at the Salesforce boundary

Decagon resolves conversations, but the action that closes the loop — refund, update, fulfillment — lives in systems of record it must reach safely.

Three ways to close it

Only one is governed

Decagon alone
Acts in SalesforceNo
Governedn/a
Auditedn/a
Custom API glue
Acts in SalesforceSort of
GovernedYou build it
AuditedYou build it
Decagon + Upware
Acts in SalesforceYes
GovernedPolicy + RBAC
AuditedEvery action
In practice

From resolved chat to updated record

A customer contacts support about a duplicate charge. Decagon identifies the issue, confirms eligibility, and reaches the resolution — but the actual refund credit and case closure must land in Salesforce. Upware's on-prem bridge takes over: it replays the exact sequence it learned from a recorded agent workflow, updates the case status, logs the resolution note, and marks the opportunity record with a policy-approved activity. The action is attributable to the session, bounded by RBAC, and written to the audit trail before the customer's chat window even closes. No engineer touched a custom integration; Upware learned the Salesforce workflow from a recorded desktop session.

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

Salesforce is SOX-relevant for us. How does Upware make sure Decagon's writes are attributable and reversible?
Every action Upware executes carries the identity of the initiating session, the policy rule that permitted it, and a timestamped audit entry. Writes to revenue and forecast objects are scoped by RBAC so Decagon can only touch what it's been explicitly granted. Upware also supports triggering approved Salesforce Flows rather than writing directly to objects, which keeps your existing approval and reversal logic intact.
Does this require Salesforce to expose new APIs or custom development on our side?
No custom API work is required. Upware learns the Salesforce workflow by recording how a human agent already performs the action — UI interactions, web requests, and screen state — then replays that sequence deterministically. If a supported REST API exists, Upware will prefer it; if a particular action only lives in the UI, Upware can execute it there instead. Either way, your Salesforce configuration stays as-is.
Decagon already handles the conversation well. What exactly does adding Upware change?
Decagon resolves the customer interaction, but the downstream record — the case update, the logged call, the stage advance — still needs to reach Salesforce correctly and safely. Without a governed execution layer, that step either falls to a human or relies on a fragile point-to-point script with no policy enforcement. Upware provides that layer: Decagon signals the intent, and Upware carries out the action in Salesforce with the same deterministic, audited precision every time.

Connect Decagon to Salesforce

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

Request a demo