Docs / AI Pilot / Auto Memory and Learning Controls

Auto Memory and Learning Controls

Published · Last updated: May 2026 · 5 min read

Who this is for

Anyone who wants to understand what AI Pilot remembers between sessions, how to review its stored memories, and how to control what the AI learns from your conversations.

What you will complete

Understand the AI memory system, review and delete stored memories, toggle Auto Memory on or off, and opt out of the learning corpus if desired.

Before you begin

  • Owner or Admin role required to manage memory settings.
  • Go to Settings → AI Agent → AI Memory.

What is Auto Memory?

When Auto Memory is enabled, AI Pilot automatically saves notes about your infrastructure preferences and setup as you chat. These notes persist between sessions, so the AI remembers context without you having to re-explain it each time.

Examples of what the AI might remember:

  • "This organization prefers nginx over apache for web servers"
  • "Production servers are all in the eu-west-1 region"
  • "The team uses Composer for PHP dependency management"

Memories are grouped into three categories:

  • Preferences — how your team likes to do things (tool choices, naming conventions, workflow preferences)
  • Topology — infrastructure facts (server roles, environment layout, provider configuration)
  • Facts — specific one-time information the AI noted (a past fix that worked, a known limitation)

How to review stored memories

  1. Go to Settings → AI Agent.
  2. Scroll down to the AI Memory section.
  3. You will see memories grouped by category: Preferences, Topology, and Facts.
  4. Each memory shows a key (what was noted) and a value (what was remembered).
  5. Hover over any memory entry to see the delete button.

How to delete a specific memory

  1. In Settings → AI Agent → AI Memory, find the memory you want to remove.
  2. Hover over the memory row — a trash icon appears on the right.
  3. Click the trash icon.
  4. The memory is deleted immediately. The AI will no longer use that fact in future conversations.

How to clear all memories

  1. In Settings → AI Agent → AI Memory, click Clear All in the top right of the memory section.
  2. A confirmation dialog appears. Click Clear to confirm.
  3. All memories are deleted. The AI starts fresh with no stored context about your organization.

When to clear all: If the AI has learned incorrect information (for example, a server was renamed but the AI still uses the old name), clearing all memories and letting it relearn is the fastest fix.

How to disable Auto Memory

  1. Go to Settings → AI Agent → Agent Controls.
  2. Find the Auto Memory toggle.
  3. Click it to disable.
  4. Auto Memory is now off. The AI will continue to use existing stored memories but will not add new ones during conversations.

Disabling Auto Memory does not delete existing memories. To also remove stored memories, use Clear All (see above).

The learning corpus — what it is and how to opt out

By default, after each conversation, CloudAIPilot extracts scrubbed lessons from your session — for example: "when port 8080 was occupied, switching to 8081 resolved the conflict." These lessons are stripped of all personal data, secrets, IP addresses, hostnames, and code before storage. They help improve AI Pilot for all users over time.

To opt your organization out:

  1. Go to Settings → AI Agent → Agent Controls.
  2. Find Opt out of AI Pilot learning corpus.
  3. Enable the toggle.
  4. No lessons from your conversations will be extracted or stored going forward.

What opting out means:

  • Your conversations are not used to improve the shared knowledge base.
  • Your organization's private memories (stored under AI Memory) are not affected by this setting.
  • This setting applies to future conversations only — it does not retroactively remove already-extracted lessons.

What success looks like

  • After chatting with AI Pilot, new memories appear under Settings → AI Agent → AI Memory within a few minutes.
  • After deleting a memory, the AI no longer references that fact in new conversations.
  • After enabling the opt-out toggle, no new lessons appear in the system-level learning database from your organization.

Common errors and fixes

"No memories are appearing even though Auto Memory is on" Cause: Memories are saved based on notable preferences and facts — short or purely conversational exchanges may not produce any. Fix: Have a more detailed conversation about your infrastructure setup. Ask the AI to remember something specifically: "Please remember that we always use Ubuntu 22.04 for new servers."

"The AI keeps using an old server name I deleted" Cause: The old name is stored in a topology memory. Fix: Find and delete the specific memory, or use Clear All to reset.

"I deleted a memory but the AI still references it" Cause: The AI's current conversation context may include the deleted memory from earlier in the session. Fix: Start a new conversation. The deleted memory will not be recalled in fresh sessions.

"I want to manually add a memory" The AI Memory section is read-only for management. To add a preference or fact, tell AI Pilot directly in chat: "Please remember that..." The AI will note it automatically (when Auto Memory is on).

Safety notes

  • Stored memories are organization-wide — all AI Pilot conversations in your organization share the same memory pool.
  • Memories never contain secrets, environment variable values, or credential information. The AI is designed to avoid storing sensitive values.
  • Clearing all memories is irreversible. The AI cannot recover deleted memories.

Related articles