What is Alma in one sentence?

A persistent memory layer that gives any AI the ability to remember facts, preferences, decisions and behavioural patterns across every conversation, model and tool.

How is Alma different from ChatGPT Memory or Claude Memory?

Alma is independent from any specific platform. The same memory works in the Alma web app, in Claude Desktop / Cursor / Windsurf via MCP, in VSCode and via the REST API. ChatGPT Memory and Claude Memory are locked to their respective platforms.

Do I need to be a developer to use Alma?

No. Alma is a full product with web app, chat, settings UI and Soul Engine editor. Developers can additionally use the SDK, MCP server and REST API to embed Alma in their own tools.

What is Alma?

Alma is a persistent memory layer for AI. It extracts and stores structured memories from your conversations — facts, preferences, decisions and behavioural patterns — and re-injects the relevant context into every future interaction. The architecture is three layers (memories, episodes, procedures) plus a Soul Engine for AI identity, all model-independent.

The Problem

Conventional AI is noisy and scattered — your context, preferences and history end up spread across ten tabs and locked inside whichever product you used. Alma keeps it all in one calm, private place that is genuinely yours.

The Solution

Alma provides a 3-layer memory system: Memories (facts and preferences), Episodes (conversation summaries), and Procedures (learned behavioral patterns). The Soul Engine assembles the most relevant context for every interaction.

See plans