Stop Token-Maxing: Use AI Agents Without Busting Your Budget
▼ Summary
– Token consumption for agentic AI is rising unsustainably, and businesses must develop a “tokenomics” strategy to measure and manage costs.
– Steve Lucas of Boomi warns against rushing into agentic AI without cost planning, noting his own token spending increased tenfold in one year.
– Leaders like Snowflake’s Sridhar Ramaswamy advocate giving staff freedom to explore agents within guidelines, as creative use can unlock business opportunities.
– Matt Luizzi of Whoop emphasizes enabling staff with guardrails and monitoring to ensure efficient token use without stifling innovation.
– Executives recommend setting clear context and objectives for agent projects, and using smaller models when possible, to limit token consumption and control costs.
A year ago, the big concern was that CIOs were diving into generative AI without a clear roadmap. Now, the same rush is happening with agentic AI, and the cost of token consumption is exploding. Steve Lucas, CEO of integration firm Boomi, warns that IT professionals are repeating the same mistake, only this time the stakes are higher because the scale of usage is unsustainable. “Whether you work inside or outside a company, it feels like your core hustle is to use AI and you token max the heck out of that technology for your own job,” he told ZDNET.
A token is the basic unit of data an AI model processes, and research shows that agentic AI consumes orders of magnitude more tokens than earlier systems. Not long ago, companies celebrated token leaderboards to show who used the most AI. Today, those leaderboards are obsolete. No one can afford to waste tokens. In the agentic era, token maxing is a sign of excess, and tokenomics , the practice of measuring, pricing, and managing token consumption , has become a critical business discipline. Lucas noted that his own spending on Claude at Boomi increased tenfold in the past year. “That’s not sustainable,” he said. “I can’t do that every year.”
The responsibility now falls on businesses and professionals to build their own enterprise-grade tokenomics strategy. Agentic AI promises to transform operations, but creating a cost-effective approach to exploring agents is urgent. Lucas emphasized that the core question for any organization is simple: “Can I operate AI at a return?” That is the fundamental challenge.
Rather than clamping down on agent use, business leaders suggest a smarter path: give people guidelines and context to make cost-effective model decisions. Snowflake CEO Sridhar Ramaswamy acknowledged that token consumption is rising inside his own company. “Are we worried about how much we are spending on AI inference across our different internal teams? Absolutely,” he said. “But do I see that spend as a reason not to use AI? Absolutely not.” At Snowflake’s Summit 2026 event, he argued that creative AI use can unlock new business opportunities, boosting efficiency and speeding up service delivery.
Matt Luizzi, VP of analytics at wearable tech firm Whoop, takes a similar stance. His company invests in agentic exploration to find a competitive edge, and he encourages staff to push boundaries. “If we want people to push themselves out of their comfort zones, we’re going to need to be OK with them taking risks and understanding that you can’t break anything,” he said. However, that doesn’t mean unchecked spending. “We have guardrails in place, and monitoring and observability to let people know what they’re spending,” Luizzi explained during a panel. “But more likely than not, we just need to sit down and enable them on, ‘How could you be doing this more efficiently and what are you trying to accomplish?'”
The secret to success, Luizzi suggested, is carefully managed enablement. Let people experiment with agents and tokens, but don’t let them go overboard. “At the end of the day, we’re leaning in hard because we think there’s an ROI to this. We are seeing people become more efficient. We are seeing work get done faster,” he said. “Those positive results mean we’re not trying to be super keen on how many tokens people use but also understanding that this is something that needs to scale, so not enabling people to go crazy either.”
Sriram Sitaraman, CIO at Synopsys, echoed that sentiment during the same panel. “You can’t innovate with constraints; you can only go so far,” he said. He recommended setting clear boundaries through project context. For example, asking an agent a vague question like “What should I do tomorrow?” can consume massive amounts of data and tokens with little value. Instead, defining a specific task , such as creating a North America sales ops agent with precise objectives and data , limits token use and delivers higher value.
Francois-Xavier Pierrel, group chief data and ad tech officer at French TV network TF1, warned against treating AI like a sugar addiction. “Everything was cheap, very cheap, and we got all into it,” he said, reflecting on the early days of generative AI. Now, agent obsession presents a more costly proposition. “The number of tokens used can quickly become a big number. And at some point, if we look at all the companies investing crazy money into agentic AI, the bill will come back.”
Pierrel advocates a common-sense approach to token control. In his organization, professionals are asked to consider smaller, more efficient models for narrow use cases. “For use cases that are very narrow or have strong boundaries, we say, ‘Can we use a small model, something that will consume fewer tokens than expected or planned, and that will still do the same job?'” As companies deploy more agents, this kind of tokenomics strategy will be essential to value creation. “I think we are going to be more agile on this one, trying stuff, and then adapting to the use case to avoid having crazy bills,” he said. “We’re not shooting a rocket to the moon. So, let’s be reasonable , don’t deploy a million agents if you can do it on a smaller scale and it will still work.”
(Source: ZDNet)
