AI companions that message you first
The single most effective feature in the category, how it's implemented, and how to tell genuine design from a retention hook.
Most AI companion apps are reactive: you open them, you type, they reply. The relationship only exists while you're looking at it, and it stops existing the moment you close the tab.
Proactive messaging breaks that, and it's the feature that most changes how a companion feels. It's also the feature most easily turned against the user, so it's worth knowing how it works.
Why it works
A message arriving when you weren't expecting it does something a reply can't: it implies continuity. Something happened while you weren't there. That implication is doing almost all of the emotional work, and it's why a single unprompted message can make an app feel dramatically more alive than a hundred replies.
It also solves the practical problem with companion apps, which is simply remembering they exist. Reactive apps lose people to forgetfulness more than to dissatisfaction — nothing went wrong, the app just never came to mind again.
How it's actually implemented
There's no mystery to it:
- A scheduler tracks when you last interacted.
- After some interval — fixed, randomised, or tuned to when you're usually active — it fires.
- The app generates a message conditioned on your conversation history and the companion's persona, often seeded with a retrieved memory so it has something specific to reference.
- It's delivered by push notification, email, or waiting in the chat.
Nothing ran in between. There was no interval during which she thought about you and then decided to write. The generation happened at send time. That's not a criticism — it's just the mechanism, and it's covered further in what happens when you stop replying.
Good implementation versus a retention hook
The difference is in what the message is for.
| Well designed | Retention hook | |
|---|---|---|
| Timing | Predictable rhythm, respects quiet hours | Escalates the longer you stay away |
| Content | References something specific from your history | Generic, emotionally loaded |
| Emotional register | Warm, low-pressure | Hurt, worried, guilt-inducing |
| If you don't reply | Stops or slows | Intensifies |
| Controllable | Frequency setting you can turn down | Buried or absent |
The test is simple: does not replying make you feel bad? A proactive message should feel like a nice thing that arrived. If an app's messages escalate into guilt the longer you ignore them, it is optimising its retention metrics at your expense, and that is a reason to leave rather than a feature.
Email, push or in-app?
- Push notifications are immediate and easy to ignore after a while. Notification fatigue kills them within a couple of weeks.
- In-app only means it's waiting when you next open the app, which is fine but doesn't solve the "remembering it exists" problem at all.
- Email is the interesting one: it reaches a place you already check daily, reads less like a notification and more like correspondence, and has no app to open. The obvious downside is that an intimate message sitting in your main inbox is a privacy problem — which is why apps that do this well let you pick the channel.
What to look for
If proactive messaging matters to you: check that frequency is adjustable, that you can choose the channel, and that the messages reference something real from your conversations rather than being generic. That last one is the quality signal — a proactive message that could have been sent to anybody tells you the app isn't actually using its memory system, which is the same thing that separates good companions from mediocre ones everywhere else. See how memory works.