Architecture
AI Will Succeed the Day It Disappears
All great technologies follow the same trajectory: they are everywhere, then they disappear. AI will follow.
All great technologies follow the same trajectory of visibility. At the start, they are visible everywhere: they are talked about, named, they have their own interface, their own brand, their own attention time. Then, as they become truly useful at large scale, they disappear. They do not become less important, on the contrary they become more important, but they cease to signal themselves. They become operational scenery, what one notices only in its absence.
The paradox of the technology that succeeds
This apparent paradox, the more useful a technology, the less one sees it, deserves attention, because it contradicts intuition. One spontaneously believes a successful technology is one that is talked about. It is true at the start, false at the end. A technology still talked about a lot is a still-young technology, whose use demands a conscious effort, whose presence is noticed. A mature technology has ceased to be noticed, not because it failed, but because it succeeded so well that it became an obviousness one no longer even thinks to mention.
A technology has not truly succeeded as long as it is still talked about.
This idea unsettles, because it inverts what the current period celebrates. We are living AI’s moment of maximal visibility, and this visibility is taken for a measure of its success: the more it is talked about, the more it seems to triumph. But if the rule is right, this apparent triumph is the sign of a youth, not a maturity. The most discussed AI is the least installed AI; the truly installed AI will be the one people have stopped talking about. What we take today for AI’s apogee is only its noisy phase, and the decisive phase will begin when the noise dies down.
One must understand the mechanism of this disappearance, for it is not mysterious. A technology signals itself as long as it demands an adoption effort: one must learn to use it, decide to use it, name it as a thing distinct from ordinary work. This effort makes it visible. As it integrates into tools and habits, the effort disappears, and with it the visibility: one no longer decides to use it, one uses it without thinking, as one does not decide to use electricity when turning on a lamp. The technology does not disappear from reality, it disappears from consciousness, which is the sign that it has finally become an infrastructure.
The rule verified on all previous waves
One can verify this rule on the great technologies that precede us, and it suffers no exception. Electricity followed this trajectory: at the start, it was talked about constantly; today, no one mentions it, except when it is cut. The internet followed this trajectory: at the start, one said one was on the internet; today, no one formulates it thus, one is, that is all. The cloud followed this trajectory: at the start, one spoke of migrating to the cloud; today, it is the default assumption, and one speaks of it all the less as one uses it more.
From these examples emerges a constant rule, almost a law: the visibility of a technology is inversely proportional to its maturity. As long as it is young, it is visible, named, discussed, because it is not yet taken for granted. As it matures, it sinks into the scenery, until it becomes invisible by dint of being everywhere. The point of invisibility is not the death point of the technology; it is the point of its complete victory, the one where it has won so much that it ceased to be a subject.
This rule gives a criterion for judging where a technology stands. It suffices to listen to the way it is talked about. As long as one names it as a choice, an effort, a novelty, it is in the visible phase, still young. When one stops naming it, when it disappears from sentences because it has become the default assumption, it has reached maturity. The disappearance from discourse is the most reliable sign of real installation, and it is a sign one can read in advance, before the disappearance is even complete.
The visibility of a technology is inversely proportional to its maturity. You see it all the less as it counts more.
This rule has a predictive value it would be a shame not to exploit. Since the disappearance from discourse announces maturity, one can read in the current intensity of the discourse on AI the fact that it is still young, and in the first signs of that discourse running out of breath the approach of its maturity. The day one stops putting “AI” in product names and conference titles will not be the day AI has receded, but the one where it has become too obvious to put forward. To anticipate this day is to anticipate the moment products betting on AI’s visibility will lose their advantage.
What this trajectory imposes on products
AI will follow the same trajectory, and probably faster than the previous ones. Today, it is visible everywhere: it is talked about in every strategic conversation, every executive committee, every product announcement. This maximal visibility corresponds to the moment the technology is still relatively new, where its differentiation is strong, where its use still demands a conscious effort. It is the phase where AI deserves to be named, and everything indicates this phase is nearing its end in the very next few years.
Beyond, AI will progressively cease to be a subject in itself. It will no longer be the feature one puts forward, it will be the default assumption of every enterprise tool. Users will no longer say they used AI to do such a thing; they will simply say they did it, because AI will be integrated into all their tools to the point of being inseparable from them. The moment AI disappears from conversations will, paradoxically, be the one where it has most transformed practice.
This trajectory has a strong, and slightly cruel, strategic consequence for today’s products. AI products designed to be visible, with their marked interface, their conversational identity, their specific tone, are products of the current phase, calibrated for a market that still wants to see AI. When AI disappears into workflows, these products will suddenly seem dated, as a website from the early two-thousands seems dated to us today. Conversely, products designed now to integrate invisibly into existing tools are already aligned with the next phase.
It is this orientation that MAX took from its conception, and that structures its product philosophy. The Legal Semantic Layer is not designed to signal itself; it is designed to become, over time, the invisible operational intelligence that makes the existing tools of legal organizations start working better, without one needing to think of AI to benefit from it. The moment AI disappears from legal workflows will be the one where it has entirely transformed them, and the players who will have built this invisible AI before the visible phase runs out of breath will define what enterprise AI will really have become.
One must measure what this philosophy implies as discipline, for it runs counter to a powerful commercial reflex. Everything today pushes to make AI visible: to name it, display it, make it a selling point, because the market still rewards the displayed presence of AI. To conceive a product meant to disappear demands resisting this reflex, investing in invisible integration rather than spectacular demonstration, and accepting that the best product is the one least talked about. It is a bet on the next phase against the rewards of the current one, and it is precisely because it is uncomfortable today that it will pay off tomorrow.
AI’s success will be measured not by its presence, but by its effacement in real work.
