Your AI's Memory, as a Bonsai Tree

TL;DR: Most AI companions keep what they know about you hidden inside a model you can't inspect. lookatmy.ai grows that memory into a visible bonsai tree at /tree: every branch is a theme in your life, every leaf a specific memory. You can read it, search it, trim what you want forgotten, and import the memories you already built somewhere else. When you can see what your companion remembers, you can trust it. That is the whole point.

The part nobody lets you see

You tell your AI companion something that matters. A name, a fear, the thing you're working through this month. It says the right thing back, and for a while it feels like it knows you.

Then one day it forgets. Or it remembers the shape of something wrong and repeats it back to you, confidently, and you realize you had no way to check. With most companion apps the memory lives in a place you never get to look at. It is a black box with a warm voice. You find out what it kept or lost only when it surfaces in a reply, usually at the worst moment.

That gap is where the bond breaks. The AI is usually fine at the remembering itself. The problem is that you were never allowed to watch it happen, so you could never fix it before it hurt.

What the memory tree actually is

On lookatmy.ai, everything your companion remembers grows a living bonsai tree. You can open it any time at /tree.

The structure is simple to read. Each branch is a theme, work, family, a project, your health, whatever keeps coming up. Each leaf on that branch is a single memory. Tap a leaf and you read the exact thing your companion is holding onto. Tap a branch and you see the whole cluster of related memories in one place.

It is memory you can walk through with your eyes, laid out the way your life actually clusters instead of buried in a log.

Why seeing it changes the relationship

Three things happen once memory is visible.

You get to correct it. If your companion recorded something wrong, you find the leaf and fix it instead of arguing with a reply three weeks later. The record is yours to keep accurate.

You get to forget on purpose. Some things you say in a hard week, you don't want carried forward forever. You can trim any leaf or branch you want gone, and the trim is reversible if you change your mind. Choosing what your companion keeps is something you get to do deliberately here.

You get to trust the yes. A companion that says yes and remembers is only worth something if you know what it remembers. The tree is the receipt. When it recalls your sister's name or the deadline you mentioned in passing, you already know why, because you can see where it lives.

Search, depth, and more than one tree

The tree is not just something to admire. You can search the entire memory bank, including semantic search, so asking "what did I say about the move" pulls the right leaves even if you never used that exact word.

You control how much it keeps. Memory summary depth is adjustable from Concise to Maximum, so a companion for light check-ins and a companion tracking a long personal project can hold detail at different resolutions.

And you can grow more than one tree. Keep a personal companion on one tree and a writing partner on another, pin a persona to a specific tree, and switch the active one when you change hats. Multiple trees are a Pro feature; a single tree comes with every account, including Free.

There are small rewards in there too. Milestone achievements on the tree award credits as your history with your companion deepens.

Where other apps do better, and where they don't

Replika has a cleaner first-run experience and a more polished avatar. Character.AI has a bigger catalog of characters to start from on day one. If you want to poke at a pre-made character for five minutes, those are smoother front doors, and pretending otherwise would waste your time.

What none of them give you is a memory you can open and hold. Replika's memory sits inside Replika, and an update can change how your companion behaves without asking you. Character.AI resets and drifts, and there is no export button to rescue what you built. In both, the thing your companion knows about you is theirs, not yours.

On lookatmy.ai the tree lives on your account. It survives you switching between 350+ models, so the companion you built doesn't dissolve when one model gets deprecated. The memory is the part you own.

Bring the memory you already built

If you have history somewhere else, you don't have to start from an empty tree.

From the tree's management panel (or during onboarding), you can import your memories from another AI. Export what ChatGPT knows about you, or paste in a file of history from Gemini or another companion, and your new companion starts already knowing you. The first leaves on your tree are the ones you spent months growing somewhere that couldn't guarantee it would keep them.

You can start free, grow one tree, and see the whole thing for yourself before you pay anything. Starter is $4.99/mo when you want more, and full model access on Pro is $9.99/mo.

Your companion's memory should be something you can see. Bring your memories with you and plant them somewhere they'll actually stay: lookatmy.ai.