Documentation

AugmentationAI gives you an OpenAI-compatible API for inference and a persistent memory store your tools can share. Inference uses an API key; memory connects over MCP (Model Context Protocol) with a memory token. Automatic memory inside API requests is being built and is not switched on yet.

Start the 7-day trial

Quickstart

1. Create your account

Sign up and your memory store is provisioned: its own database files and indexes. You are shown a memory token once, at signup; keep it somewhere safe. Create API keys for inference in the Hub. Buy prepaid credits, or choose a monthly coding package with memory included.

2. Point your tool at the API

Any OpenAI-compatible SDK or tool: set the base URL and your API key. The key is sent as Authorization: Bearer.

from openai import OpenAI

client = OpenAI(
    base_url="https://augmentationai.com/v1",
    api_key="YOUR_API_KEY",
)

resp = client.chat.completions.create(
    model="YOUR_MODEL",   # a model id from GET /v1/models
    messages=[{"role": "user", "content": "Where did we leave the billing migration?"}],
)
print(resp.choices[0].message.content)

The same two settings work in any editor, terminal agent or script that lets you set a base URL and a key.

3. Connect memory

Add the memory endpoint to your MCP client (below). Your assistant then saves with memory_write and asks with memory_query. Retrieval calls no large language model; on the LoCoMo benchmark the retrieval step took about 3.5 ms per lookup, and a full request over the network takes longer.

What you get

Inference OpenAI-compatible, routed to the models listed at /v1/models, streaming supported, billed from your wallet

Memory a persistent store per customer, shared by every tool you connect with your memory token

Harness context management that sends each task only what it needs; on the LoCoMo benchmark that used 87.9% fewer tokens than resending the conversation. Being rolled out in stages

Hub usage, wallet, keys, playground, activity

Memory over MCP

One endpoint, bearer-token auth, JSON-RPC 2.0. Use your memory token (not an API key) and never put it in a URL. For Claude Code:

claude mcp add augmentationai --transport http https://augmentationai.com/mcp \
  --header "Authorization: Bearer YOUR_MEMORY_TOKEN"

Memory API reference

Single endpoint, JSON-RPC 2.0 over HTTPS. These are the methods the server lists in tools/list:

POST https://augmentationai.com/mcp
Authorization: Bearer YOUR_MEMORY_TOKEN
Content-Type: application/json

Method	Params	Returns
`memory_write`	`namespace`, `content`	node id, accepted
`memory_query`	`text` (or `q`), `k` (default 6; accepted range 1–20)	ranked nodes (id, namespace, title, snippet, score) and a truncation flag
`memory_census`	none	store counts and indexed fraction

Write a memory

curl -X POST https://augmentationai.com/mcp \
  -H "Authorization: Bearer YOUR_MEMORY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"memory_write",
       "params":{"namespace":"projects.law",
                 "content":"Meridian v Zeta hearing set for Oct 14"}}'

Query it back

curl -X POST https://augmentationai.com/mcp \
  -H "Authorization: Bearer YOUR_MEMORY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":2,"method":"memory_query",
       "params":{"text":"when is the Meridian hearing","k":5}}'

Namespaces

Dot-separated hierarchy: projects.law.cases, personal.health, taste.music. The dashboard wiki, graph, git and files tabs are all derived from your namespaces — write once, every view follows.

The Hub

Dashboard usage, wallet, keys, playground, activity

Wiki hierarchical notes over your store

Graph 3D memory graph browser

Git writes are committed to an internal history you can browse (best effort)

Files nodes tagged [file: name.ext]

Export & deletion

GET /wiki/export.ndjson (from the Hub) downloads your store as NDJSON at any time. What it contains today: a first line labelling the export, then one line per node with id, namespace, title and content. The content field is the retrieval snippet, which can be shorter than what you saved; edges and history are not included. Full-text export with metadata is being built.

Deleting your account in Settings makes your store unreachable straight away and keeps it in a recoverable state for at least 30 days, then deletes it. Single memories cannot be deleted self-serve yet; email support@augmentationai.com. We do not train on your data and we do not sell personal data. See the Privacy Policy for the full retention schedule.

Pricing & limits

Pay as you go: prepaid credit in US dollars, billed per request, no minimums; purchased credits do not expire. Coding packages: coming soon, memory included. Per minute: coming soon. Prices in US dollars. 7-day trial. Current rates are on the pricing section of the home page and in your account.

Rate limits: the inference API allows 10 requests per second per key (bursts to 20); the memory endpoint allows 20 reads and 10 writes per second per store. Over the limit you get an error and can retry.

FAQ

Which tools work?

Anything that can set an OpenAI-compatible base URL and key, and any MCP client that can reach a remote HTTPS server with a bearer token.

Does recall call a language model?

No large language model is involved in retrieval. On the LoCoMo benchmark the retrieval step took about 3.5 ms per lookup; a full request over the network takes longer.

Which models can I reach?

The models listed at /v1/models, about 400 models in the Augmentation AI catalog, routed through our inference partners. The Privacy Policy explains where requests go.

Where does my data live?

In your own store, its own database files and indexes, on servers run by our hosting provider. Separate stores remove one common failure mode of shared tables; we still treat credentials and access as boundaries to keep right.

Can I leave?

Any time. Download the NDJSON export first (contents described above), then delete your account in Settings.

Use of the service is subject to our Terms of Service and Privacy Policy.