Your Shop Brain: what the assistant remembers

Open the Memory tab in your Assistant and you are looking at everything it knows about your shop that is not read straight from Etsy. Goals you have set, constraints you have stated, decisions you have made, patterns it has noticed.

It exists so you do not have to re-explain yourself in every conversation. Tell it once that you do not do print-on-demand, and you should not have to say it again next week.

It genuinely changes the answers

This is not a log you can safely ignore. These notes are handed to the assistant at the start of every conversation, so a wrong entry produces wrong advice, quietly and repeatedly.

Which is the reason the tab is editable rather than read-only. If something in here is out of date, correcting it is the fastest way to fix the assistant’s behaviour.

Where each note came from

Every entry carries a badge, because “you told me this” and “I worked this out” deserve different levels of trust:

• You – you typed it. Treated as fact.
• Learned – picked up from something you said in a chat.
• Auto – read from your shop data by a background job, such as your goal progress or how your last approved changes performed.

If a Learned or Auto note is wrong, it is not a bug so much as a bad inference, and deleting or rewriting it is the correct fix.

Pinning, and why it matters

Your brain can hold far more than travels with any one request. Only about 1,800 characters of memory are sent to the assistant each time it answers, and notes are packed into that budget in a fixed order: pinned first, then most recently seen.

So a long brain is fine. What decides whether a particular note is in the room when the assistant answers is whether it is pinned, and how recently it came up.

Pin anything that must never be dropped: a standing constraint, a hard rule about your brand, the goal you are working to all year. Anything you write yourself is pinned by default.

Deleting is permanent

There is no undo, so the delete button asks twice: the first click arms it, the second removes the note. Once gone it is gone, and the assistant stops accounting for it from the next message onward.

That is usually what you want. “Forget that I was focused on Halloween” is a real instruction, not a mistake.

It is per shop

Each shop has its own brain. Switching shops in the workspace switcher switches the memory with it, and nothing crosses between them.

That is deliberate: constraints that are right for one shop are often wrong for another, and a shared memory would quietly blend two brands together.

What to put in it

The useful test is: would you have to repeat this in every conversation? If yes, it belongs here.

Good entries are specific and durable:

• “Only digital downloads, no print-on-demand”
• “Target is $2,000 a month by December”
• “Paused the Halloween line, it underperformed”
• “Buyers keep asking for custom pet portraits”

Less useful: anything the assistant can already read from Etsy, like your listing count or last month’s views. It has those.

FAQ

Does editing memory cost credits? No. Reading and editing the brain is free. Credits are spent on creating products.

Will it re-learn something I deleted? It can. If you delete a note and then say the same thing again in a chat, it may be picked up again as a Learned note. Pinning the corrected version is the way to make a correction stick.

Can I turn the learning off? Not currently. You can delete anything it has learned, and pin what you want kept.

Why is my brain empty? A new shop has nothing to remember yet. It fills as you talk to the assistant and as background jobs read your shop. You can also write the first entries yourself, which is the fastest way to make early conversations useful.