Your employees are already using AI. The question is: how many tools?

TL;DR

The problem: Most enterprises are still thinking about AI as a platform decision, while employees have already moved on and are using multiple tools at once. 

Key insights: 

• The AI market is expanding, not consolidating. ChatGPT, Copilot, Gemini, and Claude are all growing in parallel, meaning organizations are accumulating AI tools rather than standardizing around one 

• Deployment and adoption aren’t the same thing. Copilot reaches 70.9% of organizations but ChatGPT leads on engagement at 80.3%, showing a clear gap between what gets rolled out and what employees actually use 

• Multi-tool AI usage is already a cost problem, with overlapping subscriptions accumulating through locally justified decisions and no consolidated view of total spend 

Your next step: Stop asking which AI tool to approve. Start mapping which tools are actually in use, for which purposes, and whether the overlap is justified. 

You’re asking the wrong question

The enterprise AI conversation tends to start as a platform decision. Should we standardize ChatGPT? Roll out Copilot? Approve Claude? Restrict everything else? Those are reasonable questions. They’re also the wrong starting point. 

What’s actually happening in most organizations is messier and more interesting. Employees aren’t waiting for a policy decision. They’re testing tools, comparing outputs, and quietly building AI into their workflows wherever it proves useful. AI is clearly being used. The more important issue is how many tools, for which tasks, and whether anyone has visibility into any of it. 

For CIOs, IT leaders, and SAM teams, this is where the pressure is landing right now. Innovation and experimentation are no longer the right frame. Governance, cost, and operating model are. 

The AI market is expanding faster than it is consolidating 

In most software categories, rapid early growth eventually gives way to consolidation. A few vendors pull ahead, a default standard emerges, and enterprises align around it. AI assistants aren’t following that pattern, at least not yet. 

According to Xensam’s State of SaaS Q1 2026 report, based on observed usage from more than one million professional users on corporate devices, growth is happening across all major AI platforms at the same time. ChatGPT and Microsoft Copilot remain the two clear leaders. But Google Gemini is gaining ground, Microsoft 365 Copilot continues to expand, and Anthropic Claude has gone from marginal to increasingly central in enterprise discussions. 

The market is widening. For enterprises, that means AI adoption isn’t a transition from one tool to another. Different assistants are finding their footing in different workflows, and the total number of tools in use is growing. There’s no single emerging standard to align around. There’s a whole class of fast-moving tools, and more are coming.

Deployment isn’t the same as adoption

One of the clearest signals in the Q1 2026 data is the gap between reach and engagement. Microsoft Copilot reaches 70.9% of organizations surveyed, reflecting strong formal deployment across enterprises. ChatGPT leads on engagement at 80.3%, meaning that once employees have access to both, they spend more time with ChatGPT. 

One platform is winning distribution. Another is winning attention. And in the long run, the tool that becomes habitual is the one that shapes behavior, drives spend, and resists displacement. 

For CIOs and IT leaders, this is a familiar dynamic in an unfamiliar category. Software has always been deployable without becoming indispensable. Procurement decisions don’t automatically produce consistent behavior. When employees keep returning to one tool over another, it usually means something real: the output is better, the interface is faster, the results feel more trustworthy. That signal deserves more weight than rollout data alone. 

This is why AI governance cannot be built around rollout data alone. Leaders need to understand not only which tools are officially present, but which tools are becoming habit-forming in real work. That is where the long-term strategic implications often begin.

Different tools are increasingly being used for different kinds of work

Part of what keeps the market fragmented is that users have developed genuine preferences, and those preferences are task-specific. One assistant works better for ideation. Another for writing. Another for summarization or spreadsheet work or staying inside an existing productivity environment. Users aren’t being irrational. They’re going to whatever produces the best result for the work in front of them. 

For the enterprise, that creates a more complicated picture. Multi-tool usage means overlapping subscriptions, inconsistent governance, and real difficulty deciding where to standardize and where to allow flexibility. In a more mature category, redundancy is usually identifiable and addressable. In AI, overlap doesn’t necessarily mean duplication. It may reflect real functional differences that aren’t yet well understood or formally documented. 

‘Which AI tool should we approve?’ is the wrong question. Start instead with: which capabilities do we actually need, and how many tools are currently filling those roles? 

The AI market is still fluid, and that makes governance harder

Claude is a useful illustration of how quickly things move. It didn’t appear at sufficient scale to be included in Xensam’s 2025 SaaS report. By Q1 2026, it had already reached approximately 9% engagement, making it one of the fastest-rising entrants in the category. A tool that wasn’t part of most enterprise conversations a year ago is now increasingly part of them. 

Claude is one example, but the dynamic is consistent across the category. Capabilities improve fast. User sentiment shifts fast. A tool that feels secondary today can be a serious consideration by the next planning cycle. 

Traditional governance models aren’t built for that pace. In established categories, organizations can take time: assess, decide, negotiate, standardize. AI compresses that timeline. The answer is a governance model designed for conditions that keep changing, not a faster version of the same process. More frequent review cycles. Better visibility into actual usage. A realistic acceptance that some overlap will persist while the category matures.

The governance issue is also becoming a cost issue 

The Q1 2026 data shows AI adoption growing across all major platforms simultaneously, which means many organizations are already running multiple AI subscriptions in parallel, often without a consolidated view of what they’re spending. 

One subscription per user is manageable. Several, adopted at different times by different teams, often outside central ownership, is a different situation entirely. The problem compounds when pricing models vary across flat tiers, premium features, credits, and usage-based billing. Costs accumulate through individually justified decisions, and no one ends up with a clear view of the total. 

AI spending is still treated as experimental in many organizations. It hasn’t been brought under the same budget discipline that applies to more mature SaaS categories. That gap will close eventually. It will close by design or by surprise. That choice is still available. 

What a more practical approach looks like

Both extremes create problems. Full decentralization creates cost and governance exposure. Premature consolidation locks the enterprise into choices that may not reflect where the market ends up. The more practical approach is to treat the near term as a period of controlled plurality. 

That means accepting that multiple tools will coexist for a while, while building the visibility to understand why. Some of that overlap is noise and experimentation. Some reflects genuine business need. The leadership task is to tell the difference early enough to make deliberate choices about the future state. 

That requires a different operating model. Instead of asking which tools are approved, start by understanding which tools are actually in use, how deeply, for which purposes, and whether the overlap is justified. Build governance that can adapt as the market shifts, rather than assuming a single early decision will hold. 

AI adoption in the enterprise isn’t a future state. It’s the current one. Your employees are already using AI. The real issue is how many tools they’re relying on, and whether your organization has a coherent strategy for what comes next. 

Want the full picture?

The data referenced in this post comes from Xensam’s State of SaaS Q1 2026 report, based on observed usage from more than one million professional users on corporate devices.