Build AI that remembers the work.

A useful assistant should not need the same introduction every session. Give your application a durable place for user context, product knowledge, and the sources behind its answers.

Illustrative example

Continuity belongs in the product.

Your model handles the conversation. Aether handles stored memory and retrieval. Keep that boundary clear so a new session can start with relevant context, without treating a growing chat transcript as your database.

Remember the person.

Store useful preferences and facts against an entity, then recall the relevant context when that person returns.

Ground the answer.

Retrieve product documents alongside memory so your application can show the source material that informed a response.

Keep your model choices.

Add memory to the agent or application you already run. Aether is the memory layer, not a replacement for your model or execution environment.

One place for remembered facts and source material.

Aether combines persistent memory with document retrieval. Start with a small, well-defined context set and test what your assistant recalls before expanding it.

A closer look at Aether
AetherMemory

Entity-scoped recall for returning users

Document-backed context for product questions

SDKs for TypeScript, Python, Go, and .NET

A support assistant that picks up the thread.

In this example, an assistant helps a developer resume an SDK setup. The remembered environment narrows the answer; the setup guide supplies the evidence.

  1. Store the developer’s language and current setup context.

  2. Recall those facts when the developer returns.

  3. Retrieve the relevant guide and let the application compose a sourced response.

A few useful answers.

Does Aether run an AI agent?

No. Your application or agent environment runs the model and tools. Aether stores and retrieves the context you supply; your application decides how to use it.

How does Scoped fit a software team?

Scoped organizes projects, issues, and progress. Compatible external agents can read work and post updates through authorized MCP connections, with the person and connection source recorded.

Are the three products one integrated platform?

They are separate products with distinct responsibilities. Start with the capability you need. These examples do not assume a live, automatic handoff between Aether, Scoped, and Charter.

Give your next AI feature a memory worth returning to.

Talk through your workflow