Save working memory

Save or update a reusable memory so future agents follow preferences, avoid repeated mistakes, and remember useful procedures.

Created by Chris Moen • Version 15 • 14 steps

Build a smarter knowledge base

This automation helps your AI agents get smarter over time by capturing and storing critical information. Instead of treating every interaction as a fresh start, this flow allows you to save or update personal preferences, specific procedures, and lessons learned. It ensures that your agents remember how you like things done, which reduces redundant questions and prevents repeated mistakes. By maintaining a living memory bank, you create a more persona-driven experience where the AI adapts to your unique working style.

How the memory storage works

The workflow uses a structured process to manage your information. It starts by checking the new memory you want to save and cross-referencing it with your existing memory bank settings. The system then searches your storage to see if a similar entry already exists. If it finds a match, it updates the record. If not, it creates a new one.

Once the data is processed, the flow builds a searchable copy of the memory. This step is vital because it allows your agents to find and retrieve this information instantly in future conversations. Finally, the system links the searchable data to the primary record, saves everything securely, and provides a clear summary of what was stored.

Key benefits for your team

Implementing this template provides a consistent framework for agent behaviour and knowledge retention. It acts as a set of guardrails that keep your automated tasks on track and aligned with your standards.

  • Avoid repetition: Stop giving the same instructions to your agents for every new task.
  • Improve accuracy: Reduce errors by documenting specific procedures that the AI must follow.
  • Customise output: Save style guides and tone of voice preferences to ensure brand consistency.
  • Scale expertise: Share successful workflows across different agents so your whole system learns from every interaction.
  • Better continuity: Maintain a permanent record of decisions and logic for future reference.

Integrated storage

This flow integrates directly with your memory file storage and internal databases. It uses key-value storage to remember where your memory bank is located, making the retrieval process fast and reliable. By using simple functions to build and prepare data, it keeps your knowledge base organised without requiring manual maintenance.

Steps

  1. Check the memory to save (function)
  2. Find saved memory bank settings (kv)
  3. Read saved memory bank settings (function)
  4. Find the saved memory bank (search)
  5. Select the saved memory bank (function)
  6. Check for an existing memory (table)
  7. Confirm whether this memory exists (function)
  8. Build the updated memory (function)
  9. Prepare memory for future search (function)
  10. Save searchable memory copy (function)
  11. Link memory to its search copy (function)
  12. Save the memory record (function)
  13. Remember where the memory bank is (kv)
  14. Return save summary (function)

FAQ

What does the "Save working memory" workflow do?

This flow captures and updates reusable information, preferences, and procedures so your future AI agents can follow specific guardrails and avoid repeating past mistakes.

How does the memory update process work?

The process reads your existing memory bank, checks for matching records, and builds an updated version to save back into your file storage. It also creates a searchable copy of the memory to ensure your agents can find the right information when they need it next.

Which services does this workflow integrate with?

This workflow integrates with your memory file storage and uses Key-Value (KV) storage to maintain your specific settings and preferences. It uses these services to keep your memory bank organised and accessible across different sessions.

How do I set up and customise my memory bank?

You'll need to configure your memory bank settings in the Key-Value storage and connect your file storage. Once connected, you can customise how the workflow builds and links searchable memory copies to fit your specific procedures and requirements.