Salesforce Adopts AI Without Alternative Plans
Salesforce integrated AI into its revenue engine, removing alternative plans and changing incentives. This approach led to successful outcomes, but also presented challenges.
Original reporting · forrester.com
Original: “Salesforce Shows What Happens When AI Has No Plan B”
Read “Salesforce Shows What Happens When AI Has No Plan B” at Forrester →
How this was written. Voyager Commercial wrote this article with a language model from Forrester’s reporting (model: @cf/meta/llama-3.3-70b-instruct-fp8-fast); it describes that reporting rather than adding to it. The original, linked above, is the full story.
Key points from Forrester’s report
- Salesforce integrated AI into its revenue engine without a plan B, changing its incentives and approach to customer engagement.
- The company had to expand observability and use new evaluation techniques to assess agent responses and improve behavior.
- Technology leaders should prioritize discipline in agent knowledge organization and governance to build confidence in agent behavior.
Salesforce's adoption of Agentforce is a notable example of a company committing to AI without a plan B, according to Forrester. This approach, known as goal-shielding, can push competing alternatives into the background and change the incentives around making a plan work.
The company's website is its lead generation engine, producing 200 signals that feed attribution models and determine the next best experience for visitors. As part of the Agentforce launch, Salesforce made agents touchpoints in the customer journey, aligning its incentives and removing a comfortable path back to business as usual.
The integration of AI into Salesforce's revenue engine was successful, but it also presented challenges. The company had to expand observability beyond traditional analytics and use LLM-based evaluation techniques to assess agent responses and improve behavior.
Technology leaders should understand that knowing what agents do is not the same as knowing whether they are producing good outcomes. Building confidence in agent behavior requires greater discipline in how agent knowledge is organized, accessed, and governed.
Salesforce also found that agents need knowledge that traditional publishing processes were not designed to provide. The company had to create agent-specific knowledge assets, including content stored in its content management system but never published to the public website.
The use of a headless CMS can provide part of the knowledge substrate for agents, making it a valuable tool for companies integrating AI into their revenue engines. Forrester clients can read more about Salesforce's customer zero experience and its implications for agent strategy and digital experience platforms.
Questions and answers
What is goal-shielding?
Goal-shielding refers to the commitment to one goal that can push competing alternatives into the background, changing the incentives around making a plan work.
How did Salesforce integrate AI into its revenue engine?
Salesforce made agents touchpoints in the customer journey, aligning its incentives and removing a comfortable path back to business as usual.
What challenges did Salesforce face in integrating AI?
Salesforce had to expand observability and use new evaluation techniques to assess agent responses and improve behavior, and create agent-specific knowledge assets to support agent behavior.
