S&P Global Tech Chief Says AI Efficiency Won’t Be Enough—Try Humility

S&P Global's upstream energy portfolio consists of 16 products that provide intelligence about oil and gas exploration. Customers often had to switch between applications with different interfaces to complete their work. Rather than developing another application to bring those functions together, S&P Global built a flexible platform designed to support various workflows using AI agents.

When the company introduced the platform to a few customers, the response was encouraging, but S&P Global's executive vice president and chief technology and transformation officer, Firdaus Bhathena, was cautious about claiming success. "By incredibly successful I don't mean that it suddenly drove a huge amount of revenue out of the gate," Bhathena stated.

The real measure of success was the high quality and speed of customer feedback, which helped S&P Global prepare the platform for broader adoption.

Bhathena emphasized that faster, cheaper work can create value, but lasting competitive advantage depends on what businesses can accomplish for their customers. He highlighted that S&P Global's advantage lies not only in the data it possesses but also in the methods it applies to that data and the relationships it has nurtured over time. Bhathena explained that these relationships allow the company to rethink how clients receive and use information as AI reshapes their work.

Bhathena recounted S&P Global's early AI experimentation phase, when the company had hundreds of projects in progress. While encouraging employees to explore the technology was valuable, the company eventually needed to focus on initiatives capable of delivering significant value. "I'd rather work on three things that are going to deliver significant value than 300 things," he said.

Improving productivity can free up resources for other investments, Bhathena argued. However, competitors will eventually adopt similar capabilities, diminishing how long efficiency alone can distinguish one business from another. Bhathena questioned how companies measure their progress, emphasizing that AI adoption is an input metric, not an output metric. He posed a more probing question: "What are you doing for your customers today with AI that you couldn't do two years ago?"

As customer expectations evolve, businesses are reevaluating longstanding processes, creating opportunities for S&P Global to help redesign workflows rather than merely accelerate existing tasks. One approach is forward-deployed engineering, where engineers participate directly in customer discussions to understand their problems and develop solutions. Sales and marketing teams must also deepen their understanding of the technology to engage effectively in these conversations.

S&P Global has also implemented an agentic product development life cycle (PDLC), which differs from the more limited software development life cycle. One significant change has been the role of product managers, who must work more closely with engineers as development cycles shorten and customer feedback arrives more rapidly. Bhathena cautioned against evaluating AI initiatives primarily through cost savings and urged executives to assess product quality, customer needs, and market success.

Ensuring security, compliance, and resilience from the outset is another critical challenge. Bhathena stressed that safeguards traditionally viewed as obstacles to rapid progress should be considered integral features of a product, along with its functionality and analytical capabilities. He emphasized that companies cannot treat these safeguards as add-ons to be implemented later.

As customer expectations shift, businesses increasingly want transparency regarding cybersecurity threats, vendor oversight, and vulnerability responses. Bhathena advocated for radical transparency, ensuring customers are proactive and direct about potential risks while still avoiding disclosing details that could create additional security concerns.