Managed AI Platform
Your own on-demand AI runtime - run agents, workers, and APIs that scale to zero when idle. Bounded, metered, isolated, BYOK. Never oversold, never a surprise bill.
The quotas, in plain English
Every plan is bounded on four things. Think of it like a car - here's what each one means.
Burst (vCPU)
How powerful your app can be in any single moment - its top speed. Bigger burst = handles heavier spikes.
Compute (vCPU-hours)
How much total running time you get each month. Idle apps sip almost nothing because they scale to zero.
Storage (GB)
How much you keep at rest - your files, data, and models. It stays put whether the app is running or not.
Egress (GB/mo)
Data leaving to the internet each month. Your included allowance covers real workloads; beyond it is metered.
Plans
Bounded compute, storage, and egress - listed plainly. Order online today; during our soft launch a real person completes your environment setup and confirms by email (guided onboarding), so it is not an instant automated deploy yet. Monthly is billed as shown, annual is indicative, and you're only ever billed for your plan plus metered usage beyond it.
Non-USD amounts are indicative. You're billed in your selected currency at checkout at the current rate.
Agent Starter
Run your first AI agent or worker
$190 billed annually - indicative (~2 months free)
Monthly billed as shown; annual indicative. Local currency at checkout.
- 75 vCPU-hr/mo compute included
- Burst up to 0.5 vCPU / 1 GB RAM
- 5 GB persistent storage
- 40 GB/mo egress, metered beyond
- 1 concurrent agent, 2 scheduled jobs
- Scale-to-zero - idle time costs you nothing
- BYOK - bring your own AI key, zero markup
- Hardened isolation, 3-day log retention
- Community / email support
Agent Business
Production agents with real headroom
$490 billed annually - indicative (~2 months free)
Monthly billed as shown; annual indicative. Local currency at checkout.
- 350 vCPU-hr/mo compute included
- Burst up to 1 vCPU / 2 GB RAM
- 15 GB storage, 120 GB/mo egress, metered beyond
- 3 concurrent agents, 10 scheduled jobs
- 1 always-warm service + scale-to-zero
- BYOK + managed AI gateway option
- Hardened container isolation, 7-day logs
- Email support
Agent Pro
+ meteredScale a team of agents, metered beyond plan
$990 billed annually - indicative (~2 months free)
Monthly billed as shown; annual indicative. Local currency at checkout.
- 750 vCPU-hr/mo included + metered overage
- Burst up to 2 vCPU / 4 GB RAM
- 40 GB storage, 250 GB/mo egress + metered
- 6 concurrent agents, 30 scheduled jobs
- 2 always-warm services + scale-to-zero
- BYOK + managed gateway + metered credits
- Hardened container isolation, 14-day log retention
- Priority support
Enterprise & Custom
Dedicated capacity, private networking, compliance, and a custom SLA - scoped to your workload.
- • Dedicated / private infra tier - never shares capacity
- • Custom compute, egress, and storage envelopes
- • GPU add-ons (metered)
- • SSO, audit logging, and compliance support
- • Custom SLA + quarterly reviews
- • Migration help from your current stack
Beyond your plan = metered, never a surprise bill
Your included allowance is a real ceiling, not a trap. As you approach it we warn you, then protect you - so you decide what happens next, not a runaway invoice.
You're notified at 80% and 100% of your included compute, egress, or AI-token allowance - no silent creep.
At the ceiling we throttle rather than bill blindly, giving you time to top up credits or upgrade.
Past that it's a clean stop, or transparent metered overage you opted into - egress overage is metered on every plan; Agent Pro also meters compute and AI credits. No "unlimited", no surprise charge.
Compare plans
Every quota and feature, side by side.
| Feature | Agent Starter | Agent Business | Agent Pro |
|---|---|---|---|
| Monthly compute included | 75 vCPU-hr | 350 vCPU-hr | 750 vCPU-hr + metered |
| Burst (max at once) | 0.5 vCPU / 1 GB | 1 vCPU / 2 GB | 2 vCPU / 4 GB |
| Persistent storage | 5 GB | 15 GB | 40 GB |
| Egress / mo | 40 GB + metered | 120 GB + metered | 250 GB + metered |
| Concurrent agents | 1 | 3 | 6 |
| Concurrent jobs | 1 | 3 | 8 |
| Scheduled jobs | 2 | 10 | 30 |
| Always-warm services | Scale-to-zero | 1 + scale-to-zero | 2 + scale-to-zero |
| Rate limit | 10 req/s | 50 req/s | 200 req/s |
| Custom domains | 1 | 3 | 5 |
| AI access | BYOK | BYOK + gateway | BYOK + gateway + credits |
| Log retention | 3 days | 7 days | 14 days |
| Support | Community / email | Priority |
Pricing indicative; local currency is shown at checkout. Overage beyond the included allowance is metered.
Questions
How is this different from the AI-Ready cPanel plans?
Tier 1 (AI-Ready cPanel) is a slice of a shared server for AI-enabled websites and PHP/Node/Python apps. Tier 2 is your OWN on-demand runtime for agents, workers, and APIs - it scales to zero when idle, is isolated per tenant, and is metered by real usage rather than a fixed cPanel envelope.
What does "scale-to-zero" mean for my bill?
When your agent isn't handling a request or job, it consumes almost no compute - so idle time barely touches your included monthly allowance. You pay for a base plan plus what you actually run, not for a server sitting idle 24/7.
What happens if I go past my included allowance?
You're warned at 80% and 100%, then throttled, then blocked before you can run into a surprise bill. Beyond your plan is metered overage or prepaid credits (Agent Pro) - never an uncapped, silent charge, and never "unlimited".
Is my workload isolated from other customers?
Yes. Every tier runs your agent in a hardened container - non-root, read-only filesystem, every Linux capability dropped, hard CPU/memory/process caps, no shared kernel access to other tenants' processes. Enterprise gets a dedicated, private infra tier on top of that. We never run paying customers' workloads on the same boxes as our own marketing sites.
Not sure which tier fits?
Compare the full Tier 1 and Tier 2 ladder, or tell us what you're building and we'll recommend a plan.