Salesforce subsidizes AI implementation help

Salesforce is offering free or heavily discounted technical help to help some customers deploy Agentforce, according to a May 22 report, as the company tries to convert interest in AI products into working implementations. The effort comes days before Salesforce reports first-quarter fiscal 2027 results on May 27. The same push is paired with a stronger emphasis on data-organization tools and data cleanup before customers switch on AI features. That combination puts the work of deployment, integration and governance at the center of Salesforce’s AI sales pitch. ### Why is Salesforce putting services behind Agentforce now? May 27 is Salesforce’s next earnings date, and the company said on May 6 that it will release first-quarter fiscal 2027 results after the market closes and host an investor webcast at 2 p.m. Pacific time. UBS said ahead of the report that enterprise-software demand remained mixed and that there were limited signs of near-term acceleration in bookings trends for Salesforce. That backdrop gives investors a near-term test of whether AI products are translating into revenue and usage. ### What does Salesforce say customers need before Agentforce works? (investor.salesforce.com) Salesforce’s own Agentforce materials describe the product as running on existing workflows, data and integrations, not as a stand-alone tool. The company says Agentforce lets customers build and deploy AI-powered agents across teams and systems. Trailhead training published by Salesforce says customers must identify needed data, locate its source, connect it to Data 360, and determine whether they need real-time access before implementing Agentforce. (finance.yahoo.com) The same training says teams should outline data-processing requirements such as identity resolution, transforms and insights, and assess whether the data has quality issues that need to be fixed. (salesforce.com) ### What does that tell customers about the bottleneck? Trailhead’s implementation module says data preparation can include resolving inconsistent names and formats, mapping cleaned data into the data model, and linking records to unified profiles through identity-resolution rulesets. In Salesforce’s example, those steps happen before the agent is built to answer customer questions. (trailhead.salesforce.com) Those details line up with the reported decision to subsidize technical help. If customers need data cleanup, model mapping and profile unification before an AI agent can perform reliably, the main barrier is less likely to be access to the feature itself than the work required to make enterprise data usable. That is an inference based on Salesforce’s published implementation steps and the reported support offers. (trailhead.salesforce.com) ### How is Salesforce packaging the AI offer? Salesforce says Agentforce pricing can be consumption-based, with flex credits or conversations, and can also be sold through per-user licensing. The company also says customers can build, customize, test, deploy and monitor agents on the platform. March product material from Salesforce’s developer site also tied Agentforce to Data Cloud in a new Developer Edition, giving users a way to explore building agents while harmonizing data. (trailhead.salesforce.com) That pairing reinforces Salesforce’s message that AI deployment sits alongside data unification, not apart from it. ### What should investors watch next week? (salesforce.com) May 27 is the next concrete milestone. Salesforce said it will publish first-quarter fiscal 2027 results after market close that day, and management will discuss the numbers with investors on a webcast at 5 p.m. Eastern time. Investors will be looking for any updated commentary on Agentforce adoption, services demand and the pace of customer deployment. (investor.salesforce.com) (developer.salesforce.com)