Per-user credentials
The moment a second user shows up, one shared API key stops being acceptable. Now you're building a credential vault and an OAuth flow.
The Operating Layer for AI-Native Applications
Describe the agent you need. Kodeus drafts it, gives it an identity, credentials and memory, then keeps it running under the policy you set.
Any model
Any tool
Your infrastructure
Execution Layer
Runtime Management
Creates, executes, evicts and terminates AI workloads inside isolated per-tenant
environments.
45
guardrails built in, on by policy
24/7
Warren runs live markets on Kodeus
AES-256
GCM encryption, per-user keys
Your VPC
self-host the runtime, or let us host it
Why agents stall before production
Not because the model is hard. Because everything around it is.
Guardrails and approvals
A tool call that touches money or customers needs a policy in front of it, and a human behind it. That's a system, not a prompt.
Traces and audit history
Two weeks in, someone asks what happened on Tuesday. Without traces and execution history you have no answer.
How it works
Describe what you want. Kodeus drafts it, provisions it and keeps it alive. Here is what that looks like when the CLI opens up.
Describe it
Say what the agent should do, in plain language or in a spec file. Kodeus drafts the model choice, suggests MCP servers and scaffolds the skills.
Deploy it
You get an isolated runtime with its own identity, credentials, memory, policies and limits, on your infrastructure or ours.
Operate it
It keeps running: background routines, retries, approvals, traces and metering, with policy active the whole time.
$ pip install kodeus $ kodeus init support-agent $ kodeus run --dev
What you get out of the box
The parts you'd otherwise spend two quarters building, already running.
Runtime that survives restarts
Long-running work resumes, failures recover, and turns are scheduled per tenant so one busy user can't stall everyone else.
Memory that stays affordable
Context compaction, short- and long-term memory, and eviction, so a six-month-old agent doesn't cost more each week.
Any model, swapped in a line
Per-agent model selection with token and prompt-cache optimization. The model market changes; your application doesn't.
Every user's own identity
Per-user credentials, encrypted with per-user keys. One employee's connected accounts are unreachable from another's session.
45 guardrails, on by policy
Policy intercepts a turn at five points and can monitor, redact, block, escalate or hold for human approval.
Traces you can actually read
Structured logs, OpenTelemetry traces, per-call latency, token and cache usage, and retrievable execution history.
Live on Kodeus
Warren: a customer running on Kodeus today
A private AI fund manager built on Kodeus. Users set the capital and the risk limits; Warren executes, monitors and manages the portfolio inside them, against live venues, around the clock.
Read how it's builtMarkets don't forgive a runtime that drops work, an identity model that leaks, or a trace you can't read. It's the least forgiving place to prove a platform, which is why we built there first.
Security and governance
Run the runtime inside your own VPC against your own database. Every request carries an identity derived from a verified credential, never the request body. Every action is attributable.
- Tenant isolation across orgs, users, databases and secrets
- Encrypted vaults with API-key issuance, rotation and revocation
- Human approval gates on the actions you name
- Structured tracing and retrievable execution histories
- Self-hosted or managed, with the same core either way
Frequently asked questions
How do you deploy AI agents in production?
Deploying AI agents in production means more than hosting a model loop. The agent needs an isolated place to run, an authenticated identity, credentials scoped per user, memory, policy that can refuse an action, and a trace of what it did. Kodeus supplies that operating layer, so you describe the outcome and it handles deployment, whether self-hosted in your VPC or managed by us.
Does my data leave my infrastructure?
No. The runtime executes against your own database, under your own network policy. Runtime conversations, user data and credentials stay inside the environment you deploy into.
Am I locked into one model provider?
No. Models are selected per agent and swapped in a line of config. Cost-, latency- and policy-based routing with provider fallback is maturing on top of that.
Is this a framework, or the infrastructure under one?
Both, and that is the point. A framework helps you compose prompts and tools, then leaves you to build the runtime, identity, policy and observability yourself. On Kodeus you author the agent as a spec, the platform drafts it, and the same platform runs it: lifecycle, per-tenant concurrency, credentials, policy, traces and metering. It is where the agent gets built, not only where it gets hosted.
Can I use my own models and tools?
Yes. Models are chosen per agent and stay interchangeable, and any MCP server can be attached with one command. Your agent can also be exposed as an MCP server itself, or talk to other agents over agent-to-agent transports. If you already have a product, the CLI and SDK let you embed Kodeus operating services inside it.
What will it cost to run an always-on agent?
On the planned rates, roughly $58 a month in runtime before model tokens, and nothing at all if you self-host. Usage is metered at the execution level with model and tool consumption captured per agent, so you can see what each one costs before the invoice does. See pricing for the full breakdown.
Can I try it today?
Not yet. Kodeus is in private preview while we finish the SDK and the demo app. Join the early-access list and we will get you in touch with a build as soon as there is one, starting with teams whose use case we can support properly.
Stop building the platform. Start shipping the product.
Early access opens in stages. Tell us what you want to build and we will put you near the front of it.