Astro AI is the platform that turns your agents and people into a team with shared intelligence. It gives you a single place to manage every agent your team relies on, with built-in observability, cost visibility, guardrails, auto-scaling, and the tooling to deploy, tune, and scale agents safely.

How do I install Astro AI?

Install Astro AI with a single command: curl -fsSL https://astropods.com/install | sh

How do I deploy an agent with Astro AI?

Three commands: 'ast create my-agent' to scaffold your agent and spec, 'ast dev' to run and iterate locally with hot-reload, and 'ast push' to deploy to the Astro AI registry with tracing, scaling, and versioning.

Does Astro AI support self-hosting?

Yes. Astro AI can run in your own cloud, on-premises, or as a fully managed platform. Your agents, your data, your infrastructure, with no vendor lock-in.

What AI frameworks does Astro AI work with?

Astro AI is framework-agnostic. It works with CrewAI, LangChain, Mastra, OpenAI SDK, Vercel AI SDK, Google AI, and custom frameworks via a unified adapter.

Pricing | Astro AI

Pricing

Start free. Scale when you're ready.

Simple, clear plans for building your agentic teammates, from your first deploy to a full AI agent team running in your own cloud.

Pay as you go

From your first agent,
to your whole team

Free to startFree monthly credits, then pay for additional usage

Get started free

Create and run agents, and pay for the usage you add
Invite your team and build faster together
Track every credit spent by every agent and teammate
Private knowledge stores with AWS PrivateLink
Download agent training datasets
Community and email support

1 credit = $0.001How credits work

Enterprise

Agent fleets, in your
cloud, on your terms

CustomWe build a plan around your team and your goals

Everything in Pay as you go, plus:

Astro AI engineers help you build and onboard
Create and run fleets of AI agents
Invite as many teammates as you want
Dedicated or bring-your-own cluster
EU region availability
Data residency and sovereignty
Auto-curated training and evaluation datasets
Single sign-on (SSO)
Audit logs and per-team isolation
Custom limits and committed credits
Dedicated support and SLA

Enterprise

Agents out of the box

Ready-to-work agents, tuned for the way your org works.
On your team from day one.

Sales3 agents

Stagehand

Surfaces the day's key deal events and stage changes from your CRM.

Scout

Qualifies inbound leads so reps focus on the ones that convert.

Coach

Coaches your AEs on how to position, sell, and close deals faster.

Engineering3 agents

Compass

Turns issue-thread friction and feedback into a prioritized view of what PMs and EMs should fix next.

Sonar

Probes your live architecture graph and telemetry to answer how services are built and wired; read-only, and honest about what it can't verify.

Atlas

Tracks each project across Jira, GitHub, Slack, and more, and answers what shipped, what's blocked, and what's at risk.

Marketing & content2 agents

Quill

Researches, co-authors, and stages blog posts and other content.

Tally

Answers B2B marketing-funnel questions in plain English, with the metrics and charts to back them up.

All plans

Community blueprints

A growing library of proven blueprints for engineering, marketing, research, customer support, and sales, all ready to customize and deploy.

Browse the catalog →

Credits

How credits work

Astro AI gives you everything to build and run your agentic teammates, metered in one unit: the credit. Credits are spent on what your agents do (each chat, Slack, or API request spends credits for the compute it uses and the AI tokens it sends and receives) and on what they keep, like knowledge stores. More powerful models spend more credits. One balance, no separate infra and token bills to reconcile.

Per requestComputeAICredits

No AI or your own model202

Haiku agent235

Sonnet agent2810

Opus agent21315

ExampleA request with 1,000 input and 300 output AI tokens, on a small agent. More powerful models spend more credits.

FAQ

Common questions

What is an Astro AI credit?

One unit for everything your agents use. A request spends credits for the compute it runs on plus the AI tokens it sends and receives; knowledge stores spend a little per GB per day. One balance instead of separate infra and token bills.

What does a request cost?

A flat amount based on the agent's size (its CPU and memory), plus AI tokens at cost. A plain app with no AI or where you bring your own model only spends compute credits. A request on a more powerful model spends more.

How are AI tokens priced?

Pass-through: the same per-token price as the underlying model provider, re-expressed in credits. AI credits apply only to models you run through the Astro AI Gateway; bring your own model and you pay for compute only.

How are knowledge stores charged?

Storage is a standing footprint, so it draws a small number of credits per GB per day rather than per request. Each plan includes an allotment, and there's a cap so a large store can't drain your balance unexpectedly.

Can I download my agent's training and eval datasets?

You can download your agent's training and evaluation datasets. Enterprise adds automatic curation of them from your agents' activity. Larger exports will be metered in credits by size.

Can I run agents in my own cloud?

Yes. Enterprise supports bringing your own cluster and running inside your VPC for security and data sovereignty.