Give your AI a real memory

One API gives your AI durable memory. Ingest text and files, we encrypt at the edge and return relevant context in milliseconds.

Global edge data plane
Managed control plane
Encryption-first design

Developer-first memory infrastructure

Everything you need to add persistent, secure memory to your AI apps—without the ops overhead.

Edge-native ingest & search

Ingest and search happen at the edge with built-in rate limiting—no latency, no infra to manage.

Encryption-first by design

Every ownerId gets its own encryption key. Data is encrypted before storage and only decrypted on authorized reads.

Opinionated memory model

Simple RESTful endpoints: /ingest, /search, /attachments. Control retention, tags, pinning, and locality with clear parameters.

Operable out of the box

We handle org management, API keys, metrics, health checks, and SDKs. You ship features, not ops.

How it works

One layer owns keys, orgs, and limits. One layer ingests and searches your encrypted data at the edge.

Control layer

API keys · orgs · admin

Edge layer

ingest + semantic search

Storage

metadata · encrypted content · control records

Fast ingest and search at the edge
Bearer-key auth from the control layer
Optional scheduled backups

Built for real-world AI agents

Ingest text and search it back with a single API — this is the happy path most teams start with.

  • One /ingest and one /search endpoint
  • Bearer key from the control plane
  • Owner-scoped, encrypted storage by default
const API_BASE = "https://api.memorynode.ai";
// Use a server-side env var in production:
const API_KEY = process.env.MEMORYNODE_API_KEY;

async function ingestMemory() {
  const res = await fetch(`${API_BASE}/ingest`, {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "Authorization": `Bearer ${API_KEY}`,
    },
    body: JSON.stringify({
      ownerId: "customer_123",
      text: "Customer asked about upgrading their plan to Growth.",
      tags: ["support", "upgrade"],
    }),
  });
  return res.json();
}

async function searchMemory() {
  const res = await fetch(`${API_BASE}/search`, {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "Authorization": `Bearer ${API_KEY}`,
    },
    body: JSON.stringify({
      ownerId: "customer_123",
      query: "Did they ask about Growth pricing before?",
    }),
  });
  return res.json();
}                

Real-world use cases

What teams actually build with MemoryNode

Copilot memory

Keep every customer's history in one place so your copilot remembers past tickets, chats, and files. Encrypted per customer, with longer retention and attachments on Growth+.

Agent context

Let agents search their own notes and past conversations instead of stuffing prompts. Semantic search with filters and pagination; payload size and retention follow your tier.

Product assistants

Make product and feature feedback searchable for your team. Tag what you ingest, store it encrypted, and query it from the edge — console + SDKs included.

SaaS knowledge overlays

Add semantic search over your app's docs and tickets. Drop in the JS/Python SDKs, use built-in limits, and manage API keys from the admin console.

Simple, predictable pricing

Scale from side projects to SaaS without guessing your bill.

Experimental

Monthly Price (INR)₹0
Total API Requests10,000
Ingest Requests2,000
Search Requests8,000
Embedding Tokens50,000
Storage Limit1 GB
Free testing tier

Starter

Monthly Price (INR)₹2,999
Total API Requests100,000
Ingest Requests20,000
Search Requests80,000
Embedding Tokens500,000
Storage Limit10 GB
Solo builders & prototypes

Growth

Monthly Price (INR)₹12,999
Total API Requests1,000,000
Ingest Requests200,000
Search Requests800,000
Embedding Tokens5,000,000
Storage Limit100 GB
AI startups & SaaS teams

Security that ships with your code

Built-in security primitives

Per-ownerId encryption keys
Keys stored in secure vault
Storage scoped to ownerId
Retention enforced by tier
Optional backups & observability

What we don't do (yet)

Currently single-region deployment
Primary embedding provider only
DSAR/export/delete not yet implemented
Attachments only on Growth+ tiers
Focused surface area—this isn't a general-purpose vector DB

AI without durable memory is just a demo.

Ship agents and apps with long-term memory in minutes instead of building vector infra for weeks.