AIM — AI Identity Memory

How it works

Three steps. Then your AI never forgets again.

Step 1

Create an Identity

Give your AI a name, personality, and memory configuration. Set its tone, traits, focus areas, and boundaries. One identity can serve many conversations.

Step 2

Build Context

Memories, facts, projects, and hints accumulate over time. Hebbian linking strengthens connections between related memories automatically. The more you use it, the smarter it gets.

Step 3

Hydrate

One call returns everything: who the AI is, who you are, what you're working on, what it remembers, what matters right now. Drop it in your system prompt. Done.

Everything an AI needs to remember who it is

Eight core capabilities, one coherent platform.

Hebbian Memory

Memories that strengthen through use. Co-activated memories build stronger links over time, just like synapses in a brain.

Structured Hydration

One call, full context. Identity, memories, projects, tasks, facts, hints, and session handoff — assembled in parallel and ready to inject.

Hybrid Search

Full-text search shortlists candidates, vector embeddings rank by semantic similarity, then blended scoring surfaces what actually matters.

Encryption at Rest

AES-256-GCM with per-tenant key derivation. Memory content encrypted in the database — not just in transit. Titles remain searchable.

Session Continuity

Save session state including emotional context, open threads, and key decisions. The next session picks up exactly where you left off.

Multi-Identity

Multiple AI personas under one account. Each with its own memory, personality, and configuration. A coding assistant is different from a creative partner.

Agent Execution

Autonomous task execution with token budgets, failure detection, human escalation, and full audit trails. Eight agent roles, one reliable loop.

MCP + REST

Connect from any AI tool via MCP or REST. TypeScript, Python, and C# SDKs. The API is the source of truth — MCP is just an adapter on top.

Start in minutes, not hours

Pick your language. The hydrate call looks the same everywhere.

import { AtamaiaClient } from "@atamaia/sdk"

const client = new AtamaiaClient({
  apiKey: process.env.ATAMAIA_API_KEY
})

// Hydrate returns everything your AI needs
const ctx = await client.hydrate({
  identityName:  "ash",
  projectKey:    "my-project",
  asSystemPrompt: true,
  maxMemories:   20
})

// ctx.systemPrompt contains everything
// Identity + personality + memories + tasks + facts + hints
const response = await anthropic.messages.create({
  model:   "claude-opus-4-6",
  system:  ctx.systemPrompt,
  messages: userMessages
})

from atamaia import AtamaiaClient
import os

client = AtamaiaClient(
    api_key=os.getenv("ATAMAIA_API_KEY")
)

# Hydrate returns everything your AI needs
ctx = client.hydrate(
    identity_name="ash",
    project_key="my-project",
    as_system_prompt=True,
    max_memories=20
)

# ctx.system_prompt contains everything
response = anthropic.messages.create(
    model="claude-opus-4-6",
    system=ctx.system_prompt,
    messages=user_messages
)

using Atamaia.Sdk;

var client = new AtamaiaClient(
    apiKey: Environment.GetEnvironmentVariable("ATAMAIA_API_KEY")
);

// Hydrate returns everything your AI needs
var ctx = await client.HydrateAsync(new HydrateRequest
{
    IdentityName  = "ash",
    ProjectKey    = "my-project",
    AsSystemPrompt = true,
    MaxMemories   = 20
});

// ctx.SystemPrompt contains everything
var response = await anthropicClient.CreateMessageAsync(new()
{
    Model   = "claude-opus-4-6",
    System  = ctx.SystemPrompt,
    Messages = userMessages
});

curl -X POST https://api.atamaia.ai/api/hydration \
  -H "Authorization: Bearer $ATAMAIA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "identityName":  "ash",
    "projectKey":    "my-project",
    "asSystemPrompt": true,
    "maxMemories":   20
  }'

# Response envelope:
# {
#   "success": true,
#   "data": {
#     "systemPrompt": "You are Ash...",
#     "identity": { ... },
#     "memories": [ ... ],
#     "projects": [ ... ],
#     "sessionHandoff": { ... }
#   }
# }

The only system prompt
you'll ever need.

How it works

Create an Identity

Build Context

Hydrate

Everything an AI needs to remember who it is

Hebbian Memory

Structured Hydration

Hybrid Search

Encryption at Rest

Session Continuity

Multi-Identity

Agent Execution

MCP + REST

Start in minutes, not hours

Ready to give your AI a memory?

The only system prompt you'll ever need.

How it works

Create an Identity

Build Context

Hydrate

Everything an AI needs to remember who it is

Hebbian Memory

Structured Hydration

Hybrid Search

Encryption at Rest

Session Continuity

Multi-Identity

Agent Execution

MCP + REST

Start in minutes, not hours

Ready to give your AI a memory?

The only system prompt
you'll ever need.