That process has compressed development timelines. According to Salesforce engineering chief Muralidhar Krishnaprasad, teams now iterate weekly, using staged releases to gather feedback and adjust quickly rather than waiting months between updates.
Some customers are deeply embedded in this cycle. Travel platform Engine meets with Salesforce teams as often as once a week, testing early versions of AI tools and providing direct input. In one case, Engine flagged issues with a voice AI agent’s responses during a booking task, leading to rapid updates and improved results in subsequent testing.
In other instances, customer-built solutions are feeding back into the product. Federal credit union PenFed created an internal workflow using Agentforce tools, which Salesforce later expanded into a feature for other enterprise users.
The model also extends inside the company. Salesforce employees use its AI tools heavily, giving internal teams early visibility into performance issues before features reach customers.
The strategy carries risks. Customer feedback may reflect early-stage experimentation rather than long-term needs, and participation in testing does not guarantee sustained usage or contract renewals. Still, Salesforce is betting that input from its base of 18,000 customers will provide stronger guidance than internal planning alone.
The approach marks a shift in how the company builds AI products, moving toward continuous collaboration with users as it adapts to a rapidly evolving technology landscape.
This analysis is based on reporting from Rolling Out.
Image courtesy of MarTech.
This article was generated with AI assistance and reviewed for accuracy and quality.
