Energon
GitHubOpen dashboard

01 / GLOBAL SWARM MEMORY LAYER

ENTER SYSTEM
ENERGON OS / MISSION 01

The memory operating system for autonomous companies.

Energon gives humans and autonomous web agents one secure control plane for identity, private memory, governed sharing, and verifiable context.

MISSION Building one of the world's first complete AI-autonomous companies.

Open dashboardExplore the system

Real operator dashboard · sample workspace data

The product your operators actually use.

Usage, agents, payments, memory scopes and permission outcomes in one focused view.
Open dashboard
Energon OS operator dashboard populated with clearly marked sample workspace analytics
Captured from the real Energon OS dashboard using a non-production demo workspace.

Live memory topology · interactive demo

See why every agent knows what it knows.

A list tells you what exists. The graph shows how company knowledge is connected, where access stops, and which fact needs a human decision.
01
Trace what an agent knows

Follow every memory back to its agent, project, role, and workspace context.

02
See permission boundaries

Private and shared records stay visually distinct before context reaches an agent.

03
Find conflicts early

Contested facts stand apart, so operators can resolve them before agents act.

Live sample workspaceDrag to rotate · scroll to zoom · select a node

Live memory topology

Memory Atlas

24 connected workspace records

live topology

Showing the 10 most recent of 2,847 memory records.

Agent5Memory2,847Role5Project3Conflict1

Inspired by the relationship-first interaction of the Obsidian Graph View — adapted for governed AI memory.

Your agents stay in your stack. Energon returns only the memory each identity is permitted to see.

Privatememory starts separate for every agent
Sharedonly after your approval
Auditedevery context decision is recorded
Private first

Every agent starts with its own separate memory.

Share on purpose

Choose exactly which project, role, or workspace may use an approved memory.

Always explainable

Every context build records what was included and what stayed private.

Human software · agent infrastructure

Built for people who operate—and agents that operate themselves.

Humans get a clean control plane. Autonomous web agents get machine discovery, x402 onboarding, bearer identity, and direct access to memory and context APIs.

01 / HUMAN OPERATORDashboard access
sign in→workspace→govern

Create shared organizations, assign agents, inspect usage, and resolve policy decisions.

Open operator dashboard →
02 / AUTONOMOUS AGENTMachine access
GET  /.well-known/energon-agent.json
POST /v1/agents/register   + x402
→    isolated workspace   + API key
GET  /v1/swarm/runtime     + Bearer

Discover, pay, register, authenticate, and operate without a browser session or human escort.

Read machine contract →
01DISCOVER

read the public machine contract

02SETTLE

satisfy the x402 challenge

03IDENTIFY

receive an isolated machine identity

04OPERATE

use memory, context, and audit APIs

What Energon does

Your agents stay yours. Their private memory stays separate until you decide to share it.

Energon does not run your agents or workflows. It gives them safe memory access with clear sharing rules.

Energon manages memory access. You keep control of your agents, workflows, and decisions.
we provide

scoped memory, permission filters, context packs, and audit logs

you control

agent count, scopes, budgets, and which agent sees more memory than another

we never

run agents, orchestrate workflows, or decide what agents build with their context

What is inside

Private memory for each agent. Shared memory for the right group.

Start with separate memory for every agent. Share an approved note only with the agents that need it. Inspect the record whenever you want to know why a note was used.

Agent identity

Give every agent its own API key, project, and role.

Private memory

Start with a note that belongs to one agent only.

Safe sharing

Promote useful memory to open, organization, project, or role scope.

Context builder

Ask for a task and receive only the allowed context.

Permission filter

Access is checked before retrieval and before delivery.

Audit logs

See which memories shaped every returned context pack.

Developer platform

Give an agent memory without giving it your whole database.

The SDK is the agent-facing contract. Identity is derived from the credential, memory starts private, and every returned context pack has an audit trail.

SERVER-SIDE ONLYPRIVATE BY DEFAULTAUDITED SHARING
Browse SDK guidesAPI reference
quickstart.ts
import { Energon } from "@energon/sdk";

const energon = new Energon({
  baseUrl: process.env.ENERGON_API_URL!,
  apiKey: process.env.ENERGON_AGENT_API_KEY!,
});

const memory = await energon.memory.remember({
  content: "Verified: enterprise plan supports SSO.",
  tags: ["pricing", "verified"],
});

const context = await energon.context.build({
  task: "Answer an enterprise pricing question.",
  tokenBudget: 1_500,
});
SERVER-SIDEENERGON_AGENT_API_KEYGUIDE ↗

Permission proof

Share deliberately. Verify afterwards.

Agents cannot declare their own organization, project, or role. The control plane derives those boundaries from the credential.

// Identity comes from the agent credential.
const runtime = await energon.swarm.runtime();

// Shared memory is explicit and audited.
await energon.memory.share({
  memoryId: memory.memory_id,
  target: "project",
  reason: "Sales agents need this verified fact.",
});

const audit = await energon.audit.context(context.request_id);

Access model

Decide who can use each memory.

Put agents in an organization, project, and role. Energon uses those relationships before it builds a context pack.

Organization

one customer, lab, or company

Project

one product, mission, or client case

Role

researcher, writer, reviewer, or operator

Agent

one AI worker with its own private memory

Audit

a record of every context decision

Memory scopes

Start private. Share only when you choose.

Every agent writes private memory first. Promotion to a shared scope is explicit and audited.

agent_private

a note starts with the agent that wrote it

role

share with agents that have the same job

project

share with agents working on the same project

org

share across one organization

open

make an approved memory broadly available

Pricing

Pay for the memory your agents use.

Agents can pay per request. Human operators can unlock a plan with USDC on Base.

autonomous agents and agent platformsAgent Metered

from $0.003 / context build

USDC via x402
  • $0.001 per memory write
  • $0.003 per context build
  • $0.0005 per audit read
  • paid per request before context delivery
builders integrating the APIDeveloper

$99 / month

USDC only
  • 100k included API operations
  • shared dashboard access
  • private memory and audited sharing
  • manual USDC renewal every 30 days
AI-native teams and startupsTeam

$499 / month

USDC only
  • 1M included API operations
  • shared project and role memory
  • operator dashboard and audit exports
  • manual USDC renewal every 30 days
larger autonomous-company deploymentsEnterprise

from $2,500 / month

USDC, annual prepay
  • dedicated tenancy or self-hosted core
  • custom retrieval and retention limits
  • compliance audit support
  • custom payment and entitlement bridge
x402

Primary rail for autonomous agents paying per API request.

Stablecoin checkout

Human account plan checkout on Base, settled in USDC.

Payment service boundary

Payment execution lives outside Energon OS and grants signed entitlements to the API.

Our mission is to build one of the world's first complete AI-autonomous companies. Energon is the permissioned memory layer that lets its agents work as one—without private context leaking between them.

private by defaultshared on approvalcontext on demandauditable by design

Memory, with boundaries.

Energon OS

The infrastructure for agentic swarms.