Salesforce is crowdsourcing its AI roadmap with customers

Salesforce is crowdsourcing its AI roadmap in real time. To guide its product development, the company meets with some of its 18,000 customers as often as once a week. Jayesh Govindarajan, executive vice president at Salesforce AI, noted that these customers provide a wealth of information necessary to achieve customer success. Govindarajan added that as large language models (LLMs) and context improve, agent systems will perform increasingly autonomous behaviors, a track the company plans to invest in.

This close collaboration represents a shift away from rigid, long-term product timelines toward a bottom-up strategy led by themes rather than specific release dates. Salesforce launched Agentforce, its agent management platform, in late 2024. According to Muralidhar Krishnaprasad, president and chief technology officer of Salesforce engineering, the rapid pace of artificial intelligence development means the company cannot wait three to six months for feedback. Instead, engineering teams react week by week to customer needs, pushing code quickly and using various feedback gates before releasing features broadly.

Several customers are actively participating in this feedback loop:

• Engine: The travel management platform’s operations team meets with Salesforce weekly. Engine’s founder and CEO, Elia Wallen, noted that the arrangement allows them to influence the product directly, stating, “If somebody is willing to actually help curate and build products that we need, they can help us better and really understand our problem and how they can solve it.”
• PenFed: The federal credit union developed an IT service management (ITSM) workflow using tools and agents in Agentforce. Salesforce observed this success and subsequently rolled the tool out to its broader platform for other enterprises. Shree Reddy, chief innovation officer and executive vice president at PenFed, stated that they spend more time on this strategic relationship because the investment has yielded mutually beneficial results.

While this bottom-up approach accelerates product releases, it carries structural risks. Relying heavily on customer feedback assumes that enterprises—many of which are still figuring out the role of AI in their business—know what they need. Consequently, customers might not be the best source for long-term product development. Furthermore, testing and previewing technology in beta does not necessarily translate to long-term usage habits or future software contracts.

Why it matters

Salesforce is crowdsourcing its AI roadmap in real time with customers to keep up with the rapid pace of AI development, moving away from long-term product timelines to a bottom-up strategy. This shift highlights how enterprise software providers are adapting their development cycles to remain competitive as artificial intelligence technologies evolve.