Editor’s note: AI Impact is Newsweek’s weekly newsletter where we explore how business leaders are creating real value with artificial intelligence. Subscribe to receive this newsletter in your inbox.
One phrase from Balkrishan “BK” Kalra, president and CEO of Genpact, stands out: “We are proud engine room people. We are not the showroom people.” Kalra refers to the finance, procurement, supply chain, insurance, and risk operations Genpact has expertly managed for years. His point is that understanding how these processes fail, where problems occur, and where judgment is necessary becomes even more important as AI takes on more execution tasks.
Research with Genpact and HFS Research, a global research and advisory firm, highlights the issues companies face when integrating AI into their operations. 85% of leaders believe that technology, data, process, and talent limitations reduce AI’s value. Meanwhile, 51% either don’t have a plan to tackle these problems or haven’t acted on one; only 6% have established and measured successful programs.
Genpact’s solution is termed agentic operations. This concept involves redesigning processes so agents execute tasks while humans manage the system, address issues, and hold responsibility. Kalra summarizes, “There is no Artificial Intelligence without Process Intelligence.”
On September 17 at 11 a.m. ET, Newsweek’s Dr. Ranjit Tinaikar will interview Kalra in the upcoming “AI Impact Forum.” The discussion will cover Genpact’s internal changes and why some firms struggle to derive value from AI investments. They’ll also explore how agentic operations could transform jobs and business models. Join us as we approach this event in a few weeks.
My recent conversation with FICO’s Nikhil Behl digs deeper into how an AI model’s output leads to a decision affecting a customer.
Signal Capture: Frontlines of AI Adoption
FICO President: Keeping AI Decisions Explainable
A loan denial or false fraud alert may originate from the data given to an AI model and the systems using its output. Nikhil Behl, president of software at FICO, notes the need for infrastructure connecting input data with model results and subsequent business decisions. This must include thorough documentation of the process to ensure traceability.
An undocumented input makes for an indefensible output,
Behl emphasizes.
The decision layer must manage diverse outcomes, where some decisions are autonomously made within preset boundaries, while others incorporate AI advice with further business rules and human input. When models shift across teams, visibility may suffer, requiring meticulous tracking of model inputs, outputs, and their subsequent uses.
Lineage is essential,
Behl asserts, outlining the need to trace all aspects of model operations.
As these systems increasingly influence customer decisions, false data or problems can have significant repercussions. In regulated domains, unexplained AI models present considerable liabilities.
Stronger governance enables organizations to integrate AI swiftly without compromising oversight. As Behl explains, robust governance ensures accountability remains central as AI transitions into active roles.
Upcoming Webinars
Overcoming Barriers to AI Transformation in Legacy Industries: Hosted by Newsweek, Gabriel Snyder of Newsweek will moderate a discussion on integrating AI into longstanding organizational frameworks. Panelists include Matt Sanchez, COO at Yahoo.
The Inflection Point: Agentic Operations and Enterprise Value: Dr. Ranjit Tinaikar will discuss Genpact’s approach to agentic operations with leader BK Kalra, analyzing its impact on business processes and economic models.
What Makes a Smart Hospital? On September 3, participate in exploring digital maturity in hospitals, with experts from Statista and NYU Langone Health.
Prompt Injection
What’s one recent insight you’ve learned about AI?
Leaders are advised to prioritize precision over output quantity. As organizations manage AI-driven security, scaling responses can inadvertently increase noise, rather than ensuring signal clarity.
Run Log
AI Use Case of the Week: Contract Review
Ivo’s AI was tested against a general-purpose AI and a lawyer to review contracts. Though not empirically definitive, results showcased Ivo’s proximity to human-level output achieved with less time. The AI reviewed contracts almost six times faster than human efforts, highlighting potential AI efficiencies in legal workflows.
Context Window
- Enterprises are ill-equipped against AI security risks, emphasizing proactive management. (Computerworld)
- Bill Gates calls for government adaptation to AI disruptions. (Gates Notes)
- McKinsey reports 37% of companies see EBIT benefits from AI, unchanged from last year. (McKinsey)
- OpenAI addresses model vulnerabilities discovered in recent evaluations. (OpenAI)
- CompTIA finds 59% of organizations integrating AI but report data improvement needs. (CompTIA)
Transfer Protocol
Key executive movements in AI companies, including AJ Saha’s new role at AIA Contract Documents.
Magic Moment
Share your fun AI applications:
One user asked an AI to roast them, effectively showcasing AI’s ability to recognize human behaviors humorously.
Subscribe to receive similar content in your inbox.