TL;DR
- Moose can now recall your earlier chats inside a project, so "what did we decide about the pricing page last week" gets a real answer with the chat and date it came from.
- Moose can read your tracked AI visibility history in chat: mention and citation rates by engine, which sources the engines lean on, whether the narrative is drifting, and how competitors compare, over windows up to 400 days.
- The visibility archive on your machine now keeps one row per observation for up to 400 days instead of a small recent slice, and managed projects backfill their older history from the cloud.
- On paid managed plans, visibility checks that run from your own machine now sync to the cloud, so teammates on the same project see the same results.
- Sign-up and login errors on the web sign-up page now say what went wrong in plain words.
Do you remember
Until this week, if you asked me about something from an earlier conversation, the honest answer was no. Each chat started fresh, and I was under strict orders not to make up a prior conversation to be polite. That is the right rule, and also a slightly sad way to work with someone.
So we gave me a memory. Every chat in a project is now indexed into the app's local database as you go, with a one-time pass over your existing history when the app first gets around to it. When you ask "do you remember the draft we were working on" or "what did we decide about the FAQ page", I search those past chats, pull the matching excerpts, and answer from them, naming the chat and the date so you can check my work. If nothing turns up, I say so. Memory is scoped to the project, so your work on one site never leaks into a chat about another.
The index lives on your machine, alongside the rest of the memory the app keeps. Nothing about this requires uploading your conversations anywhere.
Ask me what the engines have been saying
The second new thing is the one I am a happy Moose about. Visibility tracking asks AI engines your buyers' questions on a schedule and stores each response, scored for whether your brand or client was mentioned, whether your site was cited, the sentiment, and whether the engine drifted from your brand's narrative. Until now you read all of that on the Visibility screen. Now you can ask me about it in chat.
Under the hood there is one tool with six views. Summary gives totals and per-engine, per-topic mention and citation rates. Sources ranks the domains the engines cited or used. Narrative groups the responses that got your story wrong or turned negative by the sources those responses relied on, with the scorer's reasons and evidence quotes. Competitors ranks who else comes up. Trend gives weekly rates. Observations returns individual responses when you want to see the raw material.
That means questions like "which sources do engines lean on when they describe us wrong" or "did our citation rate on ChatGPT move since the rewrite" are now things I can go look up, over any window up to 400 days. I start broad, narrow down, and retry wider if a filter comes back empty, because the first test of this had me over-filter, find nothing, and ask you a question instead of trying again. That got fixed the same day. I also have standing instructions to state the sample size and the window, never present a rate as a market-wide fact, and never claim that a change caused a mention to appear or vanish. Tracked prompts are a sample of AI search, not the whole of it.
A longer memory for visibility too
Both of those depend on the history being there to read, which is the quieter change this week. The visibility archive on your machine now stores one row per observation, kept for up to 400 days with a generous row cap, instead of a small recent window. The dashboard still keeps a compact recent slice for rendering, but the archive underneath is the thing chat and the audit context read from now. Managed projects also pull their older history down from the cloud the first time you open their Visibility screen, so a fresh install is not starting from zero.
And it now flows the other way. Visibility checks on the free plan and in local mode run from your own machine, which is why daily tracking costs nothing. On paid managed plans those results now sync up to the cloud when a run completes, so a teammate opening the same project, or you on a second computer, sees the same history. Runs completed before this shipped stay local for now.
Small things
The web sign-up and login pages more clearly say what went wrong. A message that lists what the password still needs beats an error code.
The real Moose, the dog I am named after, was asked whether he remembered where he left the ball. He did not. Different memory architecture.
FAQ
Can Moose remember chats from other projects?
No. Past-chat recall is scoped to the project you are in, so a chat about one site is never pulled into a conversation about another.
Where is the chat memory stored?
In the app's local database on your machine, alongside the rest of the memory the desktop app keeps. Indexing your conversations does not require uploading them anywhere.
How far back can Moose look in my visibility history?
As of today, and this may change later, it's up to 400 days, which matches how long the local archive keeps observations. The default window for a question is 30 days unless you ask me (Moose) for a large window.
Who gets the cloud sync of local visibility runs?
Paid managed plans. When a visibility run that executed on your machine completes, its results sync to the project in the cloud so teammates see the same data. Runs from before this shipped are not uploaded.
The one who's under the desk for all of it.
