By Javier Cuervo
For three years we have lived peacefully. If someone asked which artificial intelligence "won", the answer came out almost automatically: ChatGPT wins. End of discussion. The rest were "interesting" alternatives, but secondary. And November 2025 arrives, OpenAI launches GPT-5.1, Anthropic makes a move with Claude Sonnet 4.5, and six days later Google responds with Gemini 3. Three big moves in a matter of weeks. At that moment, when I read the analyses of the major media and consulting firms, I had the feeling that something had changed: not only were we facing more powerful models but we were witnessing the end of an era. The era of Unique model winner It's over. And this has consequences for the technology and people strategy of any company, especially if your digital home is called Google and your day-to-day goes through Gmail, Docs and company.
First it was the collective infatuation with ChatGPT. Managers, marketing managers, operations teams: everyone tested the model and was left with the same idea in their heads: This writes better than I do , in many organizations it was decided, explicitly or not, that the AI is ChatGPT and that everything else would be nuances. Then came the comparisons, the benchmarks, the "battle tests". Bard, Gemini 1, Gemini 1.5, Claude, Perplexity... But the conclusion remained similar: ChatGPT as the standard of experience, the rest in chase mode. It was comfortable to think like this: Reducing complexity to a proper name gave us a false sense of control.
Gemini 3 breaks exactly that.
The analysis that has accompanied its launch tells us that it's no longer a question of who has the best AI , but who is better at each specific area . It is not a league with a single champion, it is more like a team where each player stands out in a different position. The advantage is still ChatGPT's, but in complex, mathematical, scientific and strategic reasoning, the conversation has shifted to Gemini 3, instead for search with bibliography and updated context, Perplexity has established itself as the best AI, on the other hand, in working with code, Claude has gained very serious respect among developers. And in the midst of all this, strategic consulting firms repeat (we repeat) the same idea: the future is not about an omnipotent model, it is about well-orchestrated multi-model architectures.
Imagine a steering committee on any Tuesday morning. The company has been working with Gmail, Calendar, and Google Drive for ten years. The CIO is tired of hearing the same phrase: If we want to take advantage of real AI, we'll have to go to Microsoft, right? Until very recently, the honest answer was uncomfortable: if what you wanted was to exploit ChatGPT as much as possible in your day-to-day life, indeed the most direct path was through the Microsoft world. But now OpenAI, and therefore, ChatGPT has emancipated itself from Microsoft , and he is no longer the sole winner of the artificial intelligence race.
What Google brings is not just a model that scores very well on reasoning tests. It brings something more important: the real possibility that a "Google-first" company—those that live in Gmail, Docs, Sheets, Meet, Chrome—can deploy top-notch AI without lifting its entire stack, without re-educating the entire organization, and without forcing traumatic migrations. Suddenly, your inbox is no longer just a graveyard of unread emails; It's a space where an agent can prioritize what's important, summarize long conversations, and set the context for meetings. Your documents aren't loose files in folders, they're living materials that are rewritten, structured, and adapted to specific goals with the help of AI. Your calendar isn't a collection of colored squares, it's a system that can be rearranged to maximize focus and reduce noise.
It's perfect? Not at all. The same media that celebrates Gemini 3's capabilities document its mistakes, its moments of confusion, and its strange behaviors. But the qualitative leap is there. For many companies with Google as their backbone, this is the first "big yes" to the question of whether they can play in the first division of AI without giving up their stack.
As Google accelerates, OpenAI and Microsoft are redrawing their relationship.
What was once perceived as a kind of almost indissoluble marriage is being transformed into something more mature: Microsoft is still a key partner, but OpenAI is regaining degrees of freedom to work with other providers, close parallel agreements and operate with less dependency. It is not a rupture, it is a calculated emancipation.
The strategic reading for any company is simple and, at the same time, uncomfortable: if even giants feel the need to move away from the single-supplier model, why should a medium-sized company tie its hands and feet to a single player? It makes no sense to say "I'm a Microsoft company" or "I'm a Google company" as if we were talking about a football club. The question is no longer one of fidelity, it is of architecture.
And this is where the idea most repeated by serious analyses in 2025 fits in: the winning strategy is not about choosing A model , but in learning to orchestrate Various models according to the task. ChatGPT for text, narrative, argumentation. Gemini for complex reasoning, scenarios, math, science, deep constraint analysis. Perplexity when the critical thing is to find, contrast and cite sources. Claude when you're working on large codebases and need order and context.
What's interesting is that this is no longer a pretty theory. The studies being published compare "single-model" implementations versus multi-model strategies and find significant differences in return on investment. Organizations that stop looking for an outright winner and organize by "winners by area" are capturing more value, faster.
All this is not only about technology but about people
Just look at what happens inside the strategic consulting firms themselves, those that fill presentations talking about the new era of AI that has compressed deadlines and triggered our expectations. Where before three days were given for an analysis, now a deliverable is expected" for tomorrow morning ", because "well, you already have AI".
That same pattern is repeated in companies across all sectors. AI doesn't come alone; It comes with a quiet pressure: produce more, faster, with less room to think. In many places, artificial intelligence is being used to squeeze more out of people, not to unleash their creative capacity.
That's why I think the important conversation is no longer "what model do we use", but "what kind of work do we want to design with these models". Fewer prompt engineering courses and more serious conversations about processes, criteria, and limits. Less dazzling ourselves with spectacular demos and more agreeing on which tasks we are never going to delegate completely to a statistical system, no matter how brilliant it may be.
The companies that are going to be strengthened will not be those that have "more AI", but those that achieve something much more difficult: building a system in which people and models collaborate in an intelligent, sustainable and clear way. Where teams know when to ask ChatGPT to write them a first draft, when to ask Gemini for a scenario analysis, when to turn to Perplexity to find the original source of a piece of data, and when to simply say, "this is up to us to decide."
The opportunity is clear: you can finally imagine a powerful deployment of AI without having to give up your natural stack. Without changing emails, without migrating all documents to another cloud, without introducing another platform into the already saturated lives of your people. If you do things minimally well, Gemini 3 allows you to start where it really hurts: inbox, documentation, meetings, agenda.
The wake-up call is less comfortable: the argument of "it's still early" or "we'll see when this matures" is no longer valid. The pace of releases, the speed at which the distances between models are shortening, and the way giants are reorganizing their alliances indicate that this game is being played now, not five years from now.
That doesn't mean you have to jump crazy into "putting AI in everything." It means something more serene and more demanding: you have to sit down and decide which parts of your business depend mostly on good text, which depend on deep reasoning, which need quick research with sources, which live off code, and calmly draw which model best fits in each area. You have to accept that you will have several providers, several interfaces and several parallel learnings. And you have to give your people time and permission to learn, experiment and make mistakes without fear.
When I look at everything that has happened in these last few weeks, I am left with a simple idea that I try to repeat to Proportione customers: the question is no longer "which AI wins", because that competition no longer exists. The question is another, much more interesting: How you're going to organize your own internal league of artificial intelligences and people . Gemini 3 is not coming to dethrone ChatGPT, just as ChatGPT can no longer aspire to be "the only one". What Gemini does is consolidate a new map in which different models shine in different things. And on that map, the real competitive advantage isn't in the model you choose, but in the system you design to make everyone work in your favor.
That is where, I think, we are playing for the next few years.