# Memory Manager: The Librarian for Your Whole AI Team

> The Memory Manager is the agent for AI memory management in the Ai1 platform by MyZone AI: it connects your website, tools and documents to your Ai1 server, checks they stay fresh, scores knowledge quality, finds gaps and lets you search what your agents know.

Canonical page: https://myzone.ai/pages/agents/ai-memory-agent
Part of Ai1, by MyZone AI. Book a Memory Manager walkthrough: https://calendly.com/d/ct6h-tcy-8qf/ai1-demo?a1=Memory%20Manager&utm_source=myzone.ai&utm_medium=agent-page&utm_content=ai-memory-agent-final
Last updated: 2026-10-05. Reviewed by the MyZone AI team.

The librarian for your AI team, keeping what it knows fresh.

## At a glance

- **Sources:** Websites, documents and third-party tools, connected, tested and kept syncing
- **Health:** One memory health score, with counts of what is stored and indexed
- **Freshness:** Weekly pipeline checks and a monthly audit of every source
- **Quality:** A test-question score you can compare over time
- **Routine upkeep:** Runs as planned and is recorded
- **You decide first:** Logins, spending, changes that cannot be undone and client releases

## What the Memory Manager does

- **Connects new data sources:** Want your agents to know about your website, a tool you use or a set of documents? It sets up the connection, tests it and keeps it syncing on a schedule.
- **Shows memory health at a glance:** A dashboard gives one memory health score with counts of what is stored and indexed, plus tabs to look closer at each part of your knowledge.
- **Checks each agent's knowledge:** A read-only view for any agent shows how much it knows, when it last updated and whether it is in sync, so drift is spotted early.
- **Runs weekly checks and monthly audits:** Each week it confirms the knowledge pipeline is running. Each month it checks every source, and anything not updated in over a week is flagged with a suggested action.
- **Scores knowledge quality:** It runs a set of test questions against what your server knows and gives you a quality score you can compare over time.
- **Diagnoses and routes problems:** When agents give outdated answers or a source stops updating, it investigates, tells you what it found, then fixes it or passes it to the agent that owns it.

## How the Memory Manager differs from a document search tool

A document search tool finds files but cannot tell you whether they are still current or whether your agents can see them. The Memory Manager connects and tests each source, keeps it syncing, checks every week that the pipeline runs and audits every source each month. It scores knowledge quality with test questions, fixes routine problems as planned and asks you first on logins, spending, permanent changes or client releases.

How this differs from the Brain Manager (https://myzone.ai/pages/agents/ai-company-brain-agent): the Brain Manager runs the Brain itself on servers where it is installed, while the Memory Manager looks after everything your Ai1 server and its agents know.

## How it works

1. **You say what you need** (Whenever you ask): Add a source, search for a topic, check health or look into a problem. Just ask.
2. **It investigates** (Before any change): It gathers the details, checks the current state and drafts what it plans to do.
3. **Plan and check** (you approve) (Before the work starts): It checks the plan is allowed. You decide when the change involves a login, spending, something that cannot be undone or a client release.
4. **It does the work** (As planned or approved): It connects the source, runs the fix, rebuilds the wiki or schedules the sync, as planned or approved.
5. **You get a clear result** (As soon as the work ends): A short summary of what is connected, what is healthy, what improved and any next steps.

## When to use it

- **Your agents need to know about a new tool or document set:** It connects the source, tests it and keeps it syncing so your agents work from it.
- **You want to know what your Ai1 server already knows about a topic:** It searches every layer and tells you what it found, where it came from and how current it is.
- **Agents seem to be working from old information:** It finds what is stale or broken and fixes routine problems, bringing anything sensitive to you first.
- **You want a regular view of knowledge health:** Weekly health checks and monthly quality snapshots keep the picture current.
- **A source needs a connector that does not exist yet:** It works with the Skill Manager, which builds the capability, then wires the source in.

## What you get

- A memory health dashboard with one score and coverage across every layer
- Per-agent views of memory size, last update and sync status
- Weekly health checks and monthly freshness audits with suggested actions
- Knowledge quality scores from test questions
- A clear summary after each piece of work

## Example: Monthly memory audit with a new source

Example with a fictional company. Names, people and figures are invented to show the agent's output. Any resemblance to a real company or person is unintended.

What it was asked: Run this month's memory check: our agents keep quoting old opening hours, and please add the collections catalogue so they can answer questions about our objects and loans.

The sample follows Memory Manager's real monthly steps from its guide: the memory dashboard, a staleness audit that flags any source more than 7 days old, connecting and testing a new source on a weekly sync, a search across documents, wiki and knowledge graph, a 20-question quality score taken before and after, and a read-only check of each agent's own notes. Every company, figure and result is invented and no real accounts were read. A person reconnected the expired drive login, the curator shared the catalogue folder, the operations manager approved hiding the old website, and permanent deletion is left waiting for her decision. Run date: 1 October 2026.

### What it found

- Two of six sources had gone stale: the website crawl broke when the site moved address 12 days ago, and the shared drive stopped syncing 9 days ago after a password change.
- The collections catalogue is now connected and syncs weekly, adding 2,318 object records, so a question about the lace-making collection and its loans now gets a sourced answer.
- Memory health rose from 71 to 89 out of 100 and the test-question score from 6.1 to 8.0; permanently deleting an outdated 2019 website still waits for the operations manager.

## Guardrails

- Routine, reversible upkeep, such as fixing a broken index or refreshing stale data, runs as planned and is recorded.
- You decide first on anything involving a login or credential, spending, a change that cannot be undone or a client release.
- Agent health views are read-only.
- Every change is planned and recorded, so you can see what was done and why.

## Frequently asked questions

### How do you connect data sources to AI agents?

Set up each connection, test it and keep it syncing on a schedule. In Ai1 by MyZone AI, the Memory Manager does this for websites, documents and third-party tools such as an accounting system, a CRM or a helpdesk. If a connector does not exist yet, the Skill Manager builds one first.

### How often should AI knowledge be checked for staleness?

Often enough to catch a broken connection before agents give old answers. The Memory Manager confirms the knowledge pipeline is running every week and audits every source every month. Anything more than seven days out of date is flagged with a suggested next step.

### Can AI search company knowledge and say where an answer came from?

Yes. Ask the Memory Manager about a topic and it searches your indexed documents, wiki pages and knowledge connections, then tells you what it found, where it came from and how current it is. That is AI memory management in practice: one agent owns what your Ai1 server remembers. It connects sources, checks they stay current, scores knowledge quality and fixes or routes problems, so your other agents work from fresh information.

### Does the Memory Manager need my go-ahead for every change, and how is it different from the Brain Manager?

No. Routine upkeep, like fixing a broken index or refreshing stale data, runs as planned and is recorded. You are asked first when a change involves a login or credential, spending, something that cannot be undone or a client release. The Memory Manager looks after everything your Ai1 server and its agents know: sources, freshness, gaps and quality. The Brain Manager runs the Brain itself on servers where it is installed.

## More on this topic

- [The CEO Communication Trap](https://myzone.ai/pages/blog/ai-auto-learn-from-comms): Every CEO spends hours answering the same questions. An AI knowledge brain learns from every conversation and gives you back 12+ hours every week.

## About Ai1

Ai1 is the AI operations platform by MyZone AI, where each client runs on its own private server. The Memory Manager is not sold on its own: every Ai1 agent, including this one, is included on Developer Pro and every Fully Managed option, with no per-agent charge. Developer Core includes the development agents.

Pricing: https://myzone.ai/pages/services/ai1-pricing. Security: https://myzone.ai/pages/security.
