Salesforce targets last-mile challenge in enterprise AI adoption
Srini Tallapragada, president and chief engineering and customer success officer at Salesforce, told TOI that the past two years have revealed a widening gap between how LLMs perform on benchmarks and how they behave in real-world business settings.“LLMs are foundational technology and will be relevant for many years,” Tallapragada said.
“But enterprises are discovering that strong benchmark performance doesn’t automatically translate into consistent business outcomes.”According to Tallapragada, most large firms spent 2024 and early 2025 running AI pilots and demonstrations, only to find that few systems could be pushed into full production.
The challenge, he said, lies in the “last mile”, where AI systems must operate predictably across edge cases, over time, and under regulatory scrutiny.LLMs, by design, are probabilistic systems.
While they excel at understanding intent, language, and context, they do not always follow fixed instructions with absolute certainty.
“They may comply 97% of the time, but enterprises need workflows that work 100% of the time,” he said, particularly in areas such as financial services, customer refunds, and policy enforcement.To address this, Salesforce is combining generative AI with deterministic systems that enforce non-negotiable rules and standard operating procedures.
In practice, this means using LLMs where flexibility, reasoning, and empathy are required, while relying on rule-based logic for compliance-heavy or audit-sensitive steps.“People initially tried to use the same tool for everything,” Tallapragada said.
“But sometimes a simple ‘if-then’ rule is the right answer.
The challenge is making these different approaches work seamlessly together.”Tallapragada also cautioned against over-reliance on industry benchmarks, noting that many tests are theoretical and can be gamed.
“A perfect score doesn’t mean the system will perform reliably in the real world,” he said.Despite this more disciplined approach, Salesforce is not reducing its use of LLMs.
The company works with multiple large and small models and continues to increase overall usage, optimising for performance, cost, and sustainability.Looking ahead, Tallapragada said 2026 is likely to mark a turning point for enterprise AI adoption.
“The focus is shifting from excitement to outcomes,” he said.
“Our job is to turn powerful models into systems that deliver real value for businesses—consistently and at scale.”Salesforce CEO Marc Benioff has previously said the company’s AI strategy is aimed at augmenting human decision-making rather than replacing it, with AI agents handling routine tasks while humans retain judgment-driven roles.
