Salesforce shifts to weekly customer engagements to drive real-time AI roadmap
By meeting with key clients as often as once a week, Salesforce aims to rapidly adapt its development of AI agents and voice tools, utilising granular feedback to accelerate iteration before broad release.
Salesforce has fundamentally altered its approach to artificial intelligence development by crowdsourcing its product roadmap directly from its enterprise customer base. The company is now engaging with select clients as frequently as once a week, a marked increase from previous annual or quarterly review cycles. This shift allows the firm to gather granular feedback on specific problems and needs, enabling it to rapidly adapt the development of AI agents, voice tools, and Slack integrations.
The strategy operates on a bottom-up philosophy grounded in the assumption that if a solution effectively addresses a specific problem for one large enterprise, thousands of others likely face the same challenge. By classifying real-world problems and solving them at either the large language model layer or through new agentic operating system components, Salesforce aims to build a product roadmap that reacts to technological changes week by month rather than waiting months for feedback.
Jayesh Govindarajan, executive vice president at Salesforce AI, described the 18,000 customers as a wellspring of information essential for customer success. The company has shifted internal labour and resources to create a dedicated AI team, moving away from fixed product timelines to a theme-based strategy focusing on agent context, observability, and deterministic controls. This internal adaptation ensures the firm can respond to advances in agentic AI, a sector in which it has doubled down since launching AI agent management software in late 2024.
The rapid development velocity involves pushing code quickly and utilising gates to test new features with early adopters before a broad release. Muralidhar Krishnaprasad, president and chief technology officer of Salesforce engineering, noted that the company is literally reacting to the environment week by week and month by month. This approach addresses the "last-mile" problem that arose when large language models were introduced, where enterprises lacked the necessary infrastructure to fully utilise the technology, prompting the creation of the Agentforce platform.
Specific partnerships illustrate the efficacy of this model. Engine, a travel management platform, meets with Salesforce weekly, allowing its operations team to provide immediate feedback on issues such as unnatural interactions in AI voice agents. CEO Elia Wallen reported that this direct input led to immediate updates and improved A/B test results, demonstrating how user-built workflows can be refined and rolled out to the broader platform.
Similarly, the federal credit union PenFed developed a custom IT service management workflow using Salesforce's Agentforce tools. Shree Reddy, chief innovation officer and executive vice president at PenFed, highlighted that this collaboration has yielded good results by strengthening the partnership and influencing mutual value. Salesforce subsequently rolled out this user-built solution to its wider customer base, validating the premise that innovations driven by individual clients can serve the entire ecosystem.
While the company relies on the sentiment that the customer is always right, the strategy also incorporates internal testing. Salesforce employees serve as the primary users of their own AI tools, acting as an internal testing ground before external deployment. As the technology continues to evolve, the firm remains committed to adapting its resources to meet the demands of a rapidly changing environment.
