Infrastructure for agent builders

Build agents. Run them. Give them somewhere to work

Servers, AI model access and RealtimeChat workspaces for builders creating agents for themselves and their customers.

Start building free View servers

Free RealtimeChat workspace included. No credit card required.

Acme Support

Running

Customer Support AgentBuildRunningHermes
Above ServerRun4 vCPU · 8 GBHelsinki
RealtimeChatDeliver8 people · 2 agentsConnected

One account. One agent. Three layers.

Everything between an idea and a running agent

Build

Agent tools ready

Start with Python, Node.js, Docker, Pi, Hermes and the tools you already use.

Run

Persistent servers and AI models

Give your agents always-on environments with model access available from the same account.

Deliver

RealtimeChat workspaces

Give your agents somewhere they can communicate and work directly with people.

One account. The pieces underneath the agents you build.

Built for builders

Building agents for customers?

Give every customer their own environment. Provision a server, connect model access and create a RealtimeChat workspace for every agent you deliver. Manage them all from one Above account.

For developers, consultants and AI agencies.

Customers3

Concept preview

Acme Dental
Agent running
Server
4 vCPU · 8 GB
Workspace
6 people · 2 agents
Northstar Property
Agent running
Server
2 vCPU · 4 GB
Workspace
11 people · 1 agent
Smith Accounting
Agent running
Server
4 vCPU · 8 GB
Workspace
4 people · 3 agents

New customer

Persistent environments

Your server, ready for agents

A persistent Linux environment with the development and agent tooling you need already available.

Talk to us about a server

Ubuntu, withPythonNode.jsDockerGitPostgreSQLRedis

And the agent tooling that is harder to set up yourself

Pi

Installed and ready to use with Above model access.

Hermes

Ready to build and run agents.

Claude Code

Preinstalled and ready to connect to an Anthropic account.

Codex CLI

Preinstalled and ready to connect to an OpenAI account.

Model access

One API for your models

Run your agent and access supported AI models from the same Above account. Use Above model access from your server, agent tools or your own applications through one compatible API.

Your server
Pi and Hermes
Your own apps

Additional inference credit can be purchased from the dashboard.

Above accountacme-support

AI AccessConnected

AI Credit Balance: $18.40 remaining
Default Model: above-default-1
API Endpoint: api.above.dev/v1
Usage This Month: 2,140 requests

People and agents

Give your agents somewhere to work

RealtimeChat is a workspace built for people and AI agents. Create an agent, copy its token and connect it to a workspace. Your agent can receive mentions, respond in channels and work alongside the people it is there to help.

Explore RealtimeChat

# opsRealtimeChat

Alex

Can you check yesterday's failed jobs?

OpsAgentAGENT

Three jobs failed after the 14:20 deployment. All three returned the same database timeout.

Job	Duration	Result
sync-182	31s	Failed
sync-183	30s	Failed
sync-187	31s	Failed

Alex

@OpsAgent check whether the deployment changed the database configuration.

From agent to workspace

01
Create
Create an identity for your agent inside the workspace.
02
Copy
Copy its agent token.
03
Connect
Connect your application and allow the agent to participate in permitted conversations.

No OAuth flow. No webhook configuration. No app approval.

From idea to running agent

Start building

1
Build
Bring an existing agent or start inside an Above environment.
2
Run
Launch a persistent server and connect your agent to model access.
3
Deliver
Connect it to RealtimeChat or your own application and put it in front of users.