Salesforce iconSalesforceSep 15, 2026 ~7 min source read

How Salesforce and OpenAI Customers Use Agentforce and ChatGPT Work to Move from Answers to Action

Shared customers combine OpenAI models with Salesforce context, orchestration, and governance to automate tasks, prep sales work, and route complex cases to humans — turning AI responses into finished work.

Salesforce and OpenAI Customers Are Putting AI to Work

Share this story

Send the public story page.

Useful takeaways from this story.

Agentforce agents use OpenAI models plus company data to answer customers, escalate to humans when needed, and automate onboarding and support workflows.

Customer results include Uber for Business’ onboarding agent (fast build, 30% incremental revenue lift) and West Monroe’s automated weekly reporting (13,104 annual hours saved).

Salesforce’s Headless Toolkit, MCP/APIs, and a Tableau plugin extend these connections to developers and analysts while preserving enterprise identity and governance.

Salesforce and OpenAI are being used together by customers to move beyond single-answer outputs and get real work done. The combination pairs OpenAI's reasoning with Salesforce context, orchestration, and governance so agents and employees can act on AI outputs safely and at scale.

Agentforce puts OpenAI models inside workflows that know about customers, accounts, and company knowledge. In service scenarios, agents can draft responses that reference a customer's account and escalate to a human representative when issues exceed the agent's scope. That lets service teams focus on higher-complexity cases.

Sales and internal workflows in ChatGPT Work and Codex

Customers connect Salesforce into OpenAI products to prepare for calls, write follow-ups, and update CRM records without leaving the chat interface. Agentforce Sales in ChatGPT Work is already used by hundreds of organizations. Users can pull Salesforce records into ChatGPT, draft account briefs and follow-ups, and rely on the platform's existing permissions to control access to data.

A Tableau plugin for ChatGPT and Codex enables teams to explore business questions against governed data and build or validate workbooks before publishing to Tableau. Salesforce's Headless Toolkit, MCP, and APIs extend these integrations for developers, preserving enterprise identity, permissions, and governance while enabling new surfaces like voice.

Uber for Business: Built an Agentforce onboarding agent in six weeks to follow up with companies that start but don't finish signup. The agent handles roughly 2,000 leads per month, routes complex questions to humans, and produced a 30% increase in incremental revenue versus the prior manual onboarding process.

West Monroe: Rolled out ChatGPT Enterprise and Work/Codex across the firm. Sellers use Salesforce inside ChatGPT to prepare account briefs, spot deals that need attention, and draft follow-ups and CRM updates. West Monroe automated weekly reporting for 252 top accounts, saving one hour per week per account, or 13,104 hours annually.

These integrations maintain existing Salesforce identity and permission controls. Organizations keep enterprise governance and data definitions while enabling AI to act on records. The Headless Toolkit and MCP provide developer-level tools to extend access to custom surfaces and applications without weakening governance.

Customers are deploying agents and integrations that combine generative reasoning with enterprise data and controls. The results shown in customer examples are about practical outcomes: speeding onboarding flows, automating reporting, and letting employees use Salesforce records inside OpenAI tools while preserving access controls and governance.

More context around this story.

Loading more related stories...

Keep reading in the app

Open the app view to save this story, compare related coverage, and continue from the same source.

Open in app