On a recent episode of AI Impact, host Russ Fradin interviewed Bask Iyer, former Global CIO at Johnson & Johnson, Honeywell, VMware, Dell, and Juniper Networks. He is now the founder of BaskMind.com, where you can find his talks and articles about digital transformation and related topics.
Across the conversation, Bask laid out a hard-nosed playbook for enterprise CIOs and CTOs trying to separate AI signal from AI theater. The through-line was accountability.
Bask's most vivid warning was about who owns the bots running inside your company. Most large organizations, he noted, still haven't gotten a clean handle on which workers are full-time, contractors, or contingent workers. Now layer in AI agents talking to other agents, making real decisions in production.
His example: it's Thanksgiving, Amazon drops prices, and your agent discounts your inventory 30 percent to match. Did marketing authorize that? Finance? IT? If revenue craters, who owns the mistake?
Bask argued this is exactly why CIOs, CFOs, and HR leaders have to share accountability. Decision rights and audit trails for digital workers need to be defined the way they are for human ones; but in most enterprise companies, they aren't.
For CTOs evaluating platform strategy, Bask offered a pointed observation: the case for buying a monolithic enterprise suite is weakening fast. If accounts payable is just "see a bill, apply rules, pay a bill," he asked, do you really need a million-dollar system, or do you build a small agent?
He still leans toward buy for commodity functions; somebody has already written that AP agent, and will test and maintain it for you. But he sees the unit of procurement getting smaller and more task-specific. Bask also flagged a warning for the build-vs-buy-vs-partner question that didn't exist a few years ago: data. Pushing all of your proprietary corpus into a public LLM, Bask cautioned, erodes differentiation over time. He'd lean toward hybrid compute and very deliberate choices about which data leaves the building.
The most concrete framework Bask shared was for handling the CEO who is convinced the company is behind. Telling that CEO "you're wrong" doesn't work. Telling them "let's run a nine-month pilot" doesn't work either; by month nine, the question has moved.
Bask's answer is what he calls "pretotyping": Assume the bold AI claim is true, then design the fastest possible test to prove or disprove it. One month, not nine. Pull in the CFO and the relevant function head, define the criteria up front, and run.
Half the time, he said, the CEO themselves will call it: this isn't working. The other half, the team, not just the CIO, gets the credit for finding something real.
For CIOs being asked to evaluate dozens of vendors a month, this is the kind of practice that lets you move quickly, without committing to long pilots that calcify into bad investments or missed opportunities.
Bask was sharp on what counts as an actual AI outcome. Russ shared a story from a recent CIO dinner: every business lead was calling AI tools transformative, but none were willing to give up headcount or commit to a revenue lift as a result. Bask said the CFO has to call the bluff.
"The tool saves me time on weekends" is not a P&L result. Either real costs come down or real revenue goes up. Anything else is cost avoidance dressed up as transformation. CIOs and CTOs who want their AI programs to survive next year's budget cycle need to insist on that bar before the CFO does it for them.
Bask's advice to younger CIOs aiming higher is striking: When it comes to building out your tech and solutions portfolio, don't default to the safe Gartner-quadrant pick. Let's say your competitors lean on reputable names such as Service Now, Salesforce, Cisco, and AWS. If you, as the CIO, do the same, what are you actually bringing to the table? You've also spent your entire subscription budget on incumbents, leaving nothing for the agentic AI bets that could differentiate you.
He recommended carving out structured time for startups, such as a Friday innovation block where the whole team meets new vendors, and taking real chances on early-stage companies whose founders will pick up the phone when you call. (Today, and even five years from now.)
Bask's view of this moment is bullish but unsentimental. The tools are real. The productivity gains are real. But CIOs and CTOs who don't put identity around their agents, push back on cost-avoidance theater, and break the habit of buying what their competitors buy will end up with neither true accountability nor meaningful differentiation.
Listen to the full conversation between Russ Fradin and Bask Iyer on AI Impact.