When you sit across the table from enough companies working through AI adoption, patterns start to emerge. Not because the companies are alike, but because the environment they're operating in is the same. The questions they're asking come from the same pressures, the same news cycles, the same shifts in what the technology can do.
We've seen this many times at Gadoci Consulting. Early on, the convergence was around acceptable use policies and data privacy. Every client, within the same few weeks, moved past the broad question of "will these providers train on our data?" and into more specific conversations about security controls, what types of data were actually entering these systems, and what guardrails needed to exist at the workflow level. That wasn't coordinated. It was the natural result of organizations maturing past the same threshold at the same time.
We saw it again when Claude's capabilities accelerated rapidly and every client conversation shifted to what that meant for their stack. The timing wasn't a coincidence. It was the market responding to the same signal.
The most recent convergence is one worth writing about, because it carries real strategic implications.
Across multiple clients, in different industries, the same question has surfaced: "We have some people on Claude, some on ChatGPT, some on Copilot, and some on Gemini. We picked a horse early and put resources behind training on it. But now we're wondering, should we go all-in on one, or should we maintain relationships with several?"
It's a good question. And after working through it with enough companies, we've formed an opinion.
Most enterprises should be thinking in terms of three tools, not one.
Here's the reasoning. If you're a Microsoft shop, you already have Copilot. It comes with the ecosystem. If you're a Google shop, you already have Gemini. These aren't tools you chose for their AI capabilities. They're tools that came bundled with decisions you made years ago about email, documents, and collaboration. But they're there, and they work well enough in their native environments that ignoring them doesn't make sense. That's your first tool, and it's effectively free.
The real question is what sits alongside it. And that's where the ChatGPT and Claude conversation gets interesting.
Claude has pulled ahead in the enterprise space. The depth of reasoning, the ability to handle complex instructions, the quality of output on substantive business work. For most enterprise use cases, Claude is the right primary investment right now. If you're going to put serious training, workflow design, and organizational energy behind one platform, Claude is the one.
But dropping ChatGPT entirely isn't the right move either, for two reasons.
First, ChatGPT is still the most familiar AI tool for most people. It's where a lot of employees first experienced what these models can do. That familiarity makes it a strong on-ramp. For people who are still building confidence with AI, ChatGPT lowers the barrier to entry in a way that matters. You want those people using something, and ChatGPT's interface and brand recognition make it easier to get them started.
Second, this market is moving fast. Claude is ahead today. The competitive landscape six months from now is anyone's guess. Maintaining a working relationship with ChatGPT, even at a lower investment level, keeps optionality open. Locking into a single vendor in a market this volatile is a risk that doesn't need to be taken.
So the framework is three tools: the one you already have (Copilot or Gemini), the one you invest in heavily (Claude, for most organizations right now), and the one you keep in the mix for accessibility and optionality (ChatGPT).
But choosing three tools is only half the answer.
The harder work is understanding what each tool is best at and matching those capabilities to the problems your organization actually has. This is where most companies get stuck. They provision licenses without a clear picture of which problems belong to which tool, and the result is uneven adoption and wasted spend.
Not every problem needs the same level of sophistication. Some are individual productivity gains that any of the three tools can handle. Some are workflow-level automations that require deeper integration. Some are custom builds that demand a specific platform's strengths. When you document your use cases and level them by complexity, the licensing picture gets clearer, the training becomes more targeted, and the whole portfolio starts working together instead of competing for attention.
The companies that do this well won't just have better AI adoption numbers. They'll spend less on licenses they don't need, train their people more effectively, and be ready to move when the landscape shifts again.
Because it will.