Orchid: are we delegating our relationships to AI? • Bliss

On 28 July 2026, a startup called Orchid posted the launch video for its AI personal assistant on X.

So far, you might say, nothing special, right? By now, dozens of artificial intelligence tools come out every day. Not to mention that the product itself is not revolutionary: a text-based assistant that integrates with your calendar, email and other apps, accessible via iMessage as if it were a friend. In practice, one of the most immediate forms of AI agent: a system that does not merely respond, but can use tools and carry out actions on the user’s behalf.

So what? What ignited the debate was the story they chose to present it with.

A man forgets his anniversary. His girlfriend, frustrated, writes to the Orchid assistant. The assistant notifies the man, who in turn uses Orchid to book a dinner and order a bouquet. Tagline: “The first assistant that truly understands what you need. It remembers what you forget. Down to the flower.”

Three point two million views. Over 600 replies. All outraged. Are we perhaps crossing the boundary of what is acceptable for an AI?

What triggered the reaction

The comments split in two directions. The first is outright disgust, with comments referencing Black Mirror and declaring themselves outraged. The second is more precise: “adult babies uninterested in the world they live in”. “An AI to perpetuate a dead relationship they both secretly hate”, and so on.

And this is an important point, because yes: the process is clear. We are gradually relieving ourselves of responsibility, placing ever more weight on machines. First that weight was work, then it was cognitive. Now it risks being emotional.

Because, think about it: a calendar reminds us of anniversaries and birthdays; a little notebook reminds us of things we would otherwise have forgotten (like our girlfriend’s favourite flowers). Nobody is scandalised by that.

And yet the idea of delegating the care of one’s relationship to a digital assistant, that really is dystopian. And worrying.

Because the line being crossed is not technological, but human.

And yet the problem is not Orchid

One theory circulating in the comments on the video is worth taking seriously: “I think they knew exactly what they were doing and were hoping the ragebait would work. Unfortunately, it did.”

It is plausible. Three point two million views on a launch video from an unknown startup do not happen by chance.

If so, Orchid deliberately used polarisation as a tool for visibility. It worked in the most immediate sense: everyone is talking about it. But that does not change the question the video raised, which exists independently of a startup’s launch strategy.

Are we building tools that make it easier not to pay attention, or tools that are supposed to support us?

The point, in fact, is not whether these tools are dangerous. They are, but because of us, since an object is always neutral: at most, it exploits the hidden corners and insecurities of our minds. What we lose when we use them: that is the point.

What there is really to fear

The absence of AI Governance and of a genuine ethical discipline applied to AI is starting to reveal how critical it is. Orchid is frightening because it is technologically trivial and socially plausible.

This is not about an AI developing consciousness or slipping out of human control. It is about a tool that lowers the cost of inattention. Before, forgetting an anniversary had a consequence: the conflict, the argument, the moment someone looked you in the eye and said “you don’t think about me any more”.

For better or worse, we needed that too.

The more autonomy an assistant gains, the more important it becomes to establish where human oversight must remain: which actions it can take on its own, which require confirmation and which decisions should never be delegated to it. Integrating AI agents into our daily lives, however, flattens every gesture of care we make, cancelling out all feedback until we are prevented from learning anything at all.

And what if one day we found ourselves no longer learning anything from a relationship? Every relationship is a mirror that shows us the truth about the other person, yes, but also, and above all, about ourselves. Losing this process (essential above all in adolescence, but in adulthood too) would mean no longer growing emotionally. No longer growing as people.

Am I exaggerating? No, not really. If anything, I may be underestimating how dangerous this is.

Not because I believe Orchid will be a hit (personally, I think it will turn out to be a comet like so many others). But its trajectory has shown a direction: and that prospect is now clear and vivid in everyone’s mind. Sooner or later, an AI like the one in the video will arrive. And it will be a problem.

And when that happens, without those uncomfortable, pointless conversations that make us better, what kind of human beings will we be?
I am afraid of the answer.

Domande frequenti

Is Orchid really any different from a phone reminder for anniversaries?

The function may look similar, but the process leading to the notification changes. A calendar reminder stems from a prior choice: a person decides that something matters and entrusts technology with reminding them. In the Orchid video, by contrast, it is the girlfriend who tells the assistant that her partner has forgotten. The AI interprets that discomfort and steps in directly, notifying the other person. Between the expressed need and the resulting action, the step of individual intention disappears. It is this shift that provoked the public's reaction. An assistant capable of entering personal dynamics, interpreting requests and steering behaviour requires far clearer boundaries than those needed for a simple calendar.

Is the public's outrage justified or excessive?

The most visceral part of the reaction seems out of proportion to a single promotional video. The substantive criticism, however, raises a real question: to what extent should an assistant make up for a lack of attention within a relationship? Orchid responded mainly in terms of provocation and communication. The product, however, called for a more thorough explanation of its rules: which requests it can accept, when it can intervene, what information it can share and which actions must remain under people’s direct control. The suspicion that the video was engineered to generate outrage adds a further responsibility. When a company brings to market tools capable of influencing behaviour, communication cannot stop at capturing attention. It must also make the system’s limits understandable.

Is there a legitimate use for Orchid or similar tools in relationships?

Yes. Shared reminders, bookings, deadlines and the management of day-to-day logistics can lighten a couple’s organisational load. The decisive issue is the level of autonomy granted to the assistant. Supporting a decision already made means aiding memory. Interpreting discomfort, choosing how to intervene and prompting a person means taking part in the decision. The more this autonomy grows, the more explicit criteria become necessary on consent, data access, accountability, human oversight and the ability to halt the action. This is where the Orchid case moves beyond the subject of relationships and meets that of AI Governance. Serious governance work serves to establish what a system can do, on whose behalf, with what information and to what extent. These boundaries must be designed before the product enters people’s lives, because relying solely on the user’s common sense or the company’s intentions leaves precisely the most delicate situations unanswered.

Fonti e riferimenti

Scritto il 1 August 2026 · Aggiornato il 12 October 2026