Aether Documentation

Agent memory, documented like a product.

SDK-first docs for remembering facts, recalling context, and using source-backed retrieval when your agent needs it.

Get startedView on GitHub
from aether import AetherClient


client = AetherClient()


# Insert a document
doc = client.insert_text("To be or not to be...")


# Run a similarity search
results = client.search("Hamlet quotes", k=3)


# Check status
status = client.status()
Python

Give your AI agents durable memory.

Python quickstart

Store and recall per-user memories with the Memory facade.

TypeScript quickstart

Add persistent agent memory to a Node or web app.

Go quickstart

Use the first-party Go SDK for memory and retrieval.

.NET quickstart

Use the first-party .NET SDK for memory-backed apps.

Aether is the memory layer for AI agents. Create one Memory per user, customer, patient, agent, or session; call remember when something should persist; call recall when the agent needs context later. Every operation is scoped to that entity, so one user's memory never leaks into another's.

text
Agent learns a fact → Memory remembers it for one entity
                                ↓
Later request → Memory recalls the relevant facts
                                ↓
Facts + user message → LLM → personalized answer

The raw document and search APIs are still here when you need source-grounded retrieval over PDFs, docs, or a knowledge base. Treat them as the advanced layer under memory: useful for RAG, BYOE, and full document lifecycle control.

Quick start

Install an SDK, set one environment variable, and store your first memory.

Install the SDK

pip install aether-ai

For .NET, install the NuGet package AetherDb.Sdk; the namespace you use in code is still Aether.Sdk.

Create an API key in the Aether Dashboard, then load it from an environment variable, .env file, or your secrets manager. The SDKs call the production API at https://api.aetherdb.ai by default; no endpoint flag is required for the normal hosted service.

Bash
export AETHER_API_KEY="your-api-key"

Remember and recall

Each SDK exposes the same memory shape in its language's idioms: create a memory scoped to one entity, write facts with remember, then recall by meaning.

import os

from aether import Memory

memory = Memory("user-42", api_key=os.environ["AETHER_API_KEY"])
memory.remember("Prefers concise weekly summaries")

for item in memory.recall("communication style", k=3):
    print(item.score, item.text)

The IDs and scores vary by account and corpus, but successful output has this shape:

text
0.74 Prefers concise weekly summaries

Advanced API

Use AetherClient directly when you need lower-level document and search control: upload files, retrieve full passages for RAG, list or delete documents, provide your own embeddings, or debug the REST contract. Start with Documents, Search & Retrieval, or the 5-minute RAG guide.

Next Steps