Skip to main content

Memories Overview

Memories are markdown documents stored in folders that your workflows can create, update, query, and retrieve. Unlike files attached to prompts, memories are dynamic—workflows can modify them over time to build up persistent context.

The Problem Memories Solves​

Most AI interactions are amnesia by design. You run a workflow, it processes data, it returns results, and then it forgets everything.

Your customer support workflow analyzes a ticket. Next time the same customer contacts you, the workflow has no idea they've had three billing issues this month. Or you're doing competitive analysis weekly, but each run starts from zero—there's nowhere for that data to accumulate.

Memories fixes this by letting your AI workflows build up context over time within Aisle.

What This Enables​

Customer context accumulation - Each interaction with a customer updates their memory. Over time, your workflows have institutional memory of past issues, preferences, and resolutions.

Research aggregation - Weekly research workflows create or update memories with findings. A separate workflow queries all memories to generate monthly reports. The data builds up instead of disappearing.

Workflow state management - Scheduled workflows track what's been processed. Query memories to see what was already handled, process new items, update memories with the latest state.

Cross-workflow data sharing - Multiple workflows work with the same persistent data store. Workflow A creates memories, Workflow B queries them, Workflow C updates them—all without external integrations.

Memories vs. Files vs. Databases​

Use attached files when:

  • The content is static
  • You're providing reference documentation that doesn't change
  • You want simple, permanent context for a prompt

Use memories when:

  • Workflows need to create or update data
  • You're building up context over time
  • Multiple workflows need to share data
  • You need queryable, structured storage within Aisle

Use external databases when:

  • You need complex relational queries
  • You're storing massive amounts of data
  • You need transactional guarantees
  • Other systems outside Aisle need direct access

Memories are the middle ground—more dynamic than files, simpler than databases, perfectly suited for AI workflow state and context.

Accessing Memories​

Navigate to Tools → Memories from the main navigation.

Folder Organization​

Memories are organized in folders you create. Think of folders like databases or collections:

  • "Customer Profiles" folder for customer information
  • "Meeting Notes" folder for AI-summarized meetings
  • "API Responses" folder for cached external data
  • "Research" folder for competitive intelligence

Create folders for different projects, departments, or data types. Once created, folders remain available across all your workflows.

Creating Memories Manually​

You can create memories directly in Aisle using the built-in markdown editor.

When to create manually:

  • Setting up initial context before running workflows
  • Manually documenting information for AI to reference later
  • Creating reference materials that workflows can query

Click "New Memory" in any folder. Give it a name, write the markdown content, and save.

Metafields​

Add key-value pairs like status: active, category: billing, priority: high to make memories searchable.

Workflows can query memories based on these metafields. For example, find all memories where status = active or priority = high.

Think of metafields like tags or filters. The more consistent you are with naming (always use priority not sometimes priority and sometimes importance), the easier it is to query later.

Version History​

Every memory maintains complete version history. Whether you create a memory manually or a workflow updates it, every change is tracked.

You can view previous versions, see exactly what changed and when, and restore earlier versions if needed. If a workflow accidentally overwrites a memory with bad data, you can roll back.

Sharing​

By default, memories are private to the folder owner.

To share, click the Sharing button on any folder:

  • Share with entire organization
  • Share with specific team members
  • Grant View permission (read-only) or Edit permission (full access)

If your team is collaborating on workflows that use shared memories, everyone needs access to the relevant folders. Share folders, not individual memories—it's easier to manage permissions at the folder level.

Memory Operations​

Create - Add new memory (folder, name, content, optional meta-fields)

Update - Modify existing memory (replace entirely or append new info)

Find - Retrieve specific memory by ID

Query - Search memories based on meta-field values

Retrieved data becomes available as variables for subsequent workflow nodes.

For workflow integration and memory operations, see Workflow Memory Nodes.