Mnexium Integrations feel like one of the most important parts of the platform because they solve a different problem than memory. It also outlines the completion of the feature-set for the platform. I don't think any more features will offer any more utility.
Memory helps an assistant remember durable user context over time. Integrations let it work with live operational data from external systems right when a response is being generated.
We just published a new case study on Cartly, an iOS app that uses Mnexium to power a full receipt-tracking AI workflow. We really wanted to see what it would take to get a demo like this up and running.
In the post, we walk through how Cartly uses:
Memory for user preferences and continuity
Records for structured receipts and receipt_items storage
A single mnx runtime object to control identity, history, recall, and record sync
Request trace packets for auditability and debugging in production
Most automation workflows can call a model, but still need substantial glue code for memory, personalization, and structured data. The Mnexium connector makes those capabilities native in n8n.
As out platform continues to grow and captures more of an AI workload. There will always be new features & improvements we can make. This is one of those, we've always had and seen a need in the platform to direct and instruct our memory generation layer. This is what memory polices offers - the ability to guide Mnexium's memory layer.
Why Memory Policies?
Not every app wants to memorize everything. Some teams need strict extraction rules for compliance, quality, or cost. Others need per-workflow behavior, like high-signal extraction in support chats and minimal extraction in casual chats.
Mnexium memories are great for capturing facts, preferences, and context from conversations. But many AI applications also need to manage structured business data events on a calendar, deals in a pipeline, contacts in a CRM, tasks on a board, inventory items, support tickets.
Until now, you had two choices: build a separate database and API layer for your structured data, or try to shoehorn everything into unstructured memories. Neither is ideal.
