AI is moving out of the experimental phase and into the everyday rhythm of work. Teams are no longer using it occasionally for novelty or quick wins, but instead are exploring more robust use cases to investigate issues, answer questions faster, surface context, and help them move through complex workflows with more confidence. That’s the shift that most organizations’ leadership teams have been asking for. The call to action was for AI to become a part of how work gets done, but as adoption deepens, a problem is becoming harder to ignore: many of the pricing models behind enterprise AI discourage the very behavior they are supposed to enable. The more useful AI becomes, the more expensive it can feel to use at scale. That tension came through clearly in a recent Fortune story on Microsoft, Uber, and rising enterprise AI bills, which showed how quickly broad usage can turn into budget pressure.
This is why the conversation around AI costs needs to be framed carefully and intentionally. There is no issue with enterprises using AI, and in fact, in many cases, organizations should probably be using more of it. They should be asking more questions, exploring more paths, investigating problems more deeply, and making AI assistance a regular part of how work gets done. AI is a highly effective tool when applied to the right problems. But the issue with AI implementation is that too many pricing models quietly discourage that behavior. When every prompt feels metered, and every workflow carries an incremental usage cost, teams naturally begin to self-regulate, becoming more selective in their usage and oftentimes hesitating at exactly the moment AI is supposed to help them move faster and think more clearly.
At Selector, we have a different approach. We never want to penalize people for using AI, which is why we don’t charge token costs. Our belief is simple: if AI assistance helps teams work faster, investigate with more confidence, and get to better outcomes, then the commercial model should encourage that behavior, not punish it. The point of AI in operations is not to create a new source of hesitation, but to remove friction, reduce noise, and give people a better way to understand what is happening across complex systems.
Why AI token costs create the wrong behavior
The Fortune piece captures the economics clearly. Microsoft reportedly began pulling back on direct Claude Code licenses after broad internal adoption, and Uber’s CTO said the company had already burned through its annual AI coding budget in only a few months. The article also points to a broader pattern across the market: companies are pushing for more AI usage while discovering that token-based pricing can make success feel expensive. Even if token costs were to decline, total spend can still rise as usage expands and agentic workflows consume more compute.
The practical effect of all this is bigger than the bill itself. A metered experience changes use behavior. When there is a cost associated with every action, people start thinking about whether they really need to ask one more question, run one more investigation, or lean on AI in one more part of the workflow. That’s a bad habit to build. AI often creates the most value when users feel free to engage with it naturally, especially in moments where speed and context matter. If the way a product is designed is teaching users to hold back, then it’s working against its own purpose.
That’s the part of the market we think deserves more attention. It shouldn’t be about giving teams access to AI and then making them cautious about using it. The goal should be to make AI assistance something teams can rely on as a part of everyday work.
AI adoption works best when teams can use it freely
This sentiment is especially true in operational environments. During incidents, investigations, escalations, or moments of uncertainty, no one wants to be calculating the marginal cost of asking another question. Teams need to be able to move from noise to clarity quickly, understand root cause, connect signals across domains, and decide what to do next without getting buried in dashboards, alerts, or disconnected tools.
This is the larger philosophy behind Selector’s platform. We help organizations reduce alert fatigue, cut ticket volume, accelerate root cause analysis, and go from an alert to action in a matter of minutes. Using AI, we bring together logs, metrics, configs, topology, and more across domains into a unified context that helps our customers understand what’s actually happening in their environment and act with more confidence. That kind of value only compounds when users are encouraged to engage with the system fully.
We think this matters more than many vendors acknowledge. AI assistance should feel like part of the operating model, not just as an optional add-on that needs to be rationed out. If users are worried that curiosity carries a cost, they will naturally use the system less than they should. But if they know the platform is there to help them investigate, validate, and move faster without penalty, the behavior changes. Adoption becomes more natural, and the product becomes more valuable over time.
Pricing is only part of the story
Pricing is not the only issue. Some AI deployments struggle because companies try to force AI into workflows that were already too fragmented and inefficient. While AI can solve a lot of problems, it is not yet the catch-all that marketing makes it out to be. A lot of first-time AI users ask the system to mimic a person step-by-step inside a process full of handoffs, rework, and context gaps, and then wonder why the economics feel disappointing. In those cases, the problem is not only the model cost, but that the workflow itself was never redesigned to make the most efficient use of the AI.
That’s an important lesson, but it shouldn’t distract from the main point. Even a well-designed workflow will underperform if the commercial model teaches users to be cautious. Produce design and pricing design have to reinforce the same outcome. Teams should be encouraged to learn, ask, explore, and investigate. They should not feel as though every productive interaction is being counted against them.
The best AI experience is one people don’t have to ration
As the market matures, companies will learn that enterprise AI adoption is not just a model question of even a pricing question. It’s a behavior question. Do users feel comfortable making AI part of how they work, or do they feel compelled to hold back? Are they encouraged to get the full value of the system, or reminded that every interaction has a cost attached to it?
At Selector, our answer is clear. We want our users to get the most out of their AI experience. We want them to use AI assistance freely and confidently. That’s why we don’t charge for tokens, and why we believe the best enterprise AI products will be the ones that support deeper usage instead of quietly discouraging it. AI should help teams do better work, but it shouldn’t come with a meter running.
Stay Connected
Selector is helping organizations move beyond legacy complexity toward clarity, intelligence, and control. Stay ahead of what’s next in observability and AI for network operations:
- Subscribe to our newsletter for the latest insights, product updates, and industry perspectives.
- Follow us on YouTube for demos, expert discussions, and event recaps.
- Connect with us on LinkedIn for thought leadership and community updates.
- Join the conversation on X for real-time commentary and product news.
