Beyond AI adoption: The shift towards measurable business value

What it claims

Varun Rastogi (Partner, AI & Data, Deloitte India) argues that AI returns are multiplicative rather than additive. He spoke at the National AI Expo during Sri Lanka AI Week 2026, organised by Sri Lanka Telecom (SLT-MOBITEL). His claim is that value depends on five factors at once: use case selection, AI autonomy, execution, data and technology foundations, and controls. Weakness in any one of them can undermine the whole investment. In his examples, a strong use case without the right data set "is not going to be useful," and strong technology without adoption and change management "is not going to be relevant."

How it works

The DVF framework ranks AI use cases on three criteria. Desirability is business need, Viability is potential value, and Feasibility is implementation readiness.

Rastogi says returns should be measured through business outcomes, not experiments or model sophistication. His six value areas are:

1. Faster execution.
2. Stronger customer engagement.
3. Lower costs.
4. Innovation.
5. Reduced complexity.
6. Improved risk detection and customer protection.

The four autonomy levels (assist, augment, delegate, automate) are set per process. They decide how much responsibility the AI takes and how much stays with people, which Deloitte calls "human plus." For scaling, he lists leadership support, realistic success measures and regular stage gate reviews. He also stresses recognising when a simpler solution is enough.

Evidence

The examples are stated outcomes, not measured results. In banking, AI cuts business loan application analysis from days to hours. In insurance, AI extracts and validates claims information, checks eligibility and brings processing time down to hours. Neither example names a client, a baseline system, a sample size or a method.

Limits

The page is a short event report with a promotional tone. It gives no criteria or scoring scale for DVF and no definitions separating the four autonomy levels. It has no data behind the multiplicative-returns claim, and no tools, models or implementation details. Mayura Malagala (Deloitte Sri Lanka & Maldives) adds that workforce readiness matters, but gives no method for building it.