Neel Sundaresan is candid when asked how most developers use AI coding tools when left to their own devices.
“People will just say, ‘What model do you want to use?’ And people will pick the latest Sonnet 4.7 or whatever. And they might be running a simple prompt, but it will cost $40 for a million tokens,” he says. “It’s like taking your Ferrari to go buy milk. You don’t need to.”
Sundaresan, GM of Automation and AI at IBM Software, founding engineer of Microsoft GitHub Copilot, and former researcher at IBM, just saw the debut of IBM Bob, the company’s new agentic development platform. Bob has been running internally at IBM since June 2025, scaling from 100 developers to more than 80,000 users across IBM’s global workforce.
As Sundaresan tells TNS’s Darry K Taft in today’s lead story, IBM’s Bob doesn’t expose the underlying model to users. It routes tasks automatically — to Anthropic Claude, Mistral open-source models, IBM Granite, or one of several proprietary, fine-tuned models built specifically for Bob’s environment — based on what the task actually requires.
When it’s time to save the Ferrari for another day, IBM claims Bob helps developers do that.
Go deeper: Most AI coding is “like taking your Ferrari to buy milk”: IBM’s Neel Sundaresan
More stories worth reading:
▻ Palo Alto Networks makes a $700M-class AI bet on Portkey gateway
▻ Arize AI and Google Cloud lay down a standardized telemetry mandate to keep enterprise agents in check
▻ “To us, it’s just a tool”: How SAS is selling AI to the Fortune 500
