Remember once, pick up later

By the end, you will have told Claude something once and had it answer from memory in a new conversation, and come back to a piece of work and picked it up where you left it.

You need: Neuronz.ai connected to your agent.

Tell it onceIt saves what you said to your memory.

Start freshA new conversation knows nothing — except your memory.

AskIt finds what you saved and answers from it.

1. Tell it something once

1. Say something Claude could not know on its own — a fact about your work that you would otherwise repeat every time:

   remember that our client Dupont & Fils closes its books on 31 March, not 31 December, and files VAT every quarter

   Claude saves it. In a chat, it asks your permission first: approve fact_add, or choose Always allow so it stops asking.

2. Open a new conversation. In Claude Code or Oh-My-Pi, start a new session in the same folder. In a chat, open a new chat — on another device, if you like.

3. Ask:

   when is the Dupont year-end?

   In Claude Code or Oh-My-Pi, that is all: before every message, Claude is told what it already knows that matches, so it answers 31 March without being asked to look.

   In a chat, Claude only checks your memory when you ask, or when it decides to. If it answers from general knowledge, say so:

   check Neuronz.ai: when is the Dupont year-end?

   You should see it look the fact up and answer from it:

   

What it saved is on the Facts page of the dashboard, where you can correct it or delete it:

Working on code?

The same works for anything about a project the code does not say: "remember that staging is reset every Monday night".

2. Come back later and ask "where were we?"

1. Work on something real for a while — a month-end close, a proposal, a bug.

2. Make sure the session was saved:

   • In Claude Code or Oh-My-Pi, it saves on its own. At natural stopping points Claude takes one extra turn to save what was learned, decided and left open.

   • In the Claude desktop app's Code tab, nothing is saved after you close a session, so before closing one you care about, type /neuronzai:capture.

   • In a chat, nothing is saved unless you ask. Before you leave, say:

     capture this conversation to Neuronz.ai

3. The next day, open a new conversation (same folder, or any chat) and ask:

   where were we?

   You should see Claude list what was done, what was decided and what was left, instead of starting from zero. In Claude Code and Oh-My-Pi, a question about past work widens the search, and Claude is told to check the record before it says something was not done. In a chat, add "check Neuronz.ai" if it does not look on its own.

Longer pieces of work can be grouped into topics, which you can follow on the dashboard's board:

What just happened

Each thing worth keeping became one small entry in your memory, in the profile you were working in. Any later conversation in that profile can find it — in Claude Code and Oh-My-Pi automatically, in a chat when you ask. Your memory is not tied to a model: switch models and it is still there.

More depth: persistent memory, recall before every message, saving a session on its own and topics.

Next: Keep a piece of work in one topic.