Free path
Free HELM AI Kernel is open-source, with no-card Policy Lab evaluation.
MINDBURN LABS
Inspect surfacePricing
Simple pricing for governed AI actions.
Free HELM AI Kernel is open-source. Paid HELM AI is tiered as Developer, Team, Scale, and Enterprise, with checkout status listed separately for every plan.
Publication state
- Visibility
- Public pricing may be shown when every plan declares availability, CTA target, checkout status, and source-owned terms.
- Checkout
- Paid checkout stays inactive until backend self-serve mode, onboarding, support, legal, and Stripe smoke pass.
- Enterprise
- Paid HELM AI Enterprise remains reviewed access; legal hold, SSO/SCIM, long retention, private deployment, SLAs, and invoicing belong there.
Credit rule
One credit includes the boundary evaluation, verdict, and signed receipt.
Paid checkout
Paid checkout stays gated until self-serve billing and readiness checks pass.
Estimate first
Policy Lab and Paid HELM AI should show the estimated credit cost before a hosted action runs.
Credits
One credit covers the governed action.
One governed-action credit is one HELM boundary evaluation, including the ALLOW, DENY, or ESCALATE verdict and signed receipt. Receipt verification is free.
| Event | Meter | Use | Boundary |
|---|---|---|---|
| Action evaluation, including receipt | 1 credit / action | Boundary evaluation with ALLOW, DENY, or ESCALATE verdict and signed receipt. | Not a model token, seat, or generic task count. |
| Counterfactual simulation | 1 credit / case | Would-have-allowed or would-have-denied cases before enforcement. | Simulation is evidence, not authority to execute. |
| Approval ceremony record | 1 credit / approval | Human approval, escalation, or activation record bound to policy and receipt context. | A record does not bypass the Kernel boundary. |
| MCP tool call | 1 credit / action | MCP call routed through HELM quarantine and policy checks. | Ungoverned MCP calls are outside Paid HELM AI metering. |
| OpenAI-compatible tool request | 1 credit / action | Tool or action request routed through an OpenAI-compatible HELM boundary. | Model tokens remain billed by the model provider. |
| BYO coding-agent governed action | 1 credit / action | IDE or coding-agent side effect checked through HELM before dispatch. | Claude Code and Codex are examples until end-to-end contract and billing tests certify named support. |
| Receipt verification | 0 credits | Offline or hosted verification of signed receipts. | Verification stays free. |
| EvidencePack export | 5 credits | Export receipts, policy refs, approvals, and source evidence for review. | An EvidencePack is scoped evidence, not a compliance certificate. |
| Replay bundle | 10 credits | Replay bundle for a governed action or policy path. | Replay proves the recorded path; it does not alter history. |
| Template run | 1 credit | Run a public Policy Lab template against a proposed action or tool surface. | Templates are starting points, not legal advice. |
| Preset generation | 5 credits | Generate a policy preset from a selected tool or workflow shape. | Generated presets remain drafts until reviewed. |
| MCP scan | 10 credits | Scan an MCP tool surface and identify policy/quarantine posture. | Large scans may require review before execution. |
| OpenAPI policy generation | 10 credits | Generate a first-pass policy from an OpenAPI surface. | Generated policy remains draft until reviewed. |
| Repository scan | 10 base credits + surcharge | Inspect a repo surface for agent/tool policy generation. | Surcharge is ceil(max(file_count / 100, analyzed_bytes / 1,000,000)); show the estimate before scan. |
Try before signup
Policy Lab shows the path before you buy.
Choose a template
Start from a public Policy Lab template for MCP, OpenAPI, repo, or common agent-action flows.
Scan the tool surface
Inspect tools, routes, or actions and see what would need policy before execution.
Generate policy
Compile a least-authority draft that is explicit about allowed, denied, and escalated actions.
Simulate and verify
Run counterfactual cases, inspect the verdicts, and verify the receipt path before signup.
Plans
Free, Developer, Team, Scale, Enterprise.
Live
Try HELM locally and in Policy Lab - no card, no signup.
Free
$0 open-source Kernel plus Policy Lab trial
Free HELM AI Kernel is the open-source fail-closed execution boundary, with Policy Lab available for no-card evaluation.
- Included
- 500 hosted credits
- Limit
- 10 actions/min per anonymous session or IP
- Overage
- No overage
- Workspace
- 0 saved workspaces; session-local only
- Retention
- Session-local
- Rollover
- No rollover; monthly expiry
- HELM AI Kernel local execution boundary
- Local Console path where source-backed
- Offline receipt verification
- Templates and demo simulations
Preview
For solo builders shipping governed agents.
Developer
$49 per month
Hosted Paid HELM AI for saved projects, API keys, dev deployments, templates, and governed-action receipts.
- Included
- 30k credits / month
- Limit
- 120 actions/min per workspace across API keys
- Overage
- $12 / 10k credits, billed exact per credit
- Workspace
- 1 workspace
- Retention
- 30 days
- Rollover
- No rollover; monthly expiry
- Saved Policy Lab projects
- API keys
- Dev and preview deployments
- Email support
Preview Featured
For teams running agent workflows together.
Team
$149 per month
Shared Paid HELM AI workspace for approvals, policy libraries, receipt retention, and team review loops.
- Included
- 200k credits / month
- Limit
- 300 actions/min per workspace across API keys
- Overage
- $9 / 10k credits, billed exact per credit
- Workspace
- 1 shared workspace
- Retention
- 90 days
- Rollover
- No rollover; monthly expiry
- Shared workspace
- Approvals and policy library
- Receipt retention
- Email and Slack support
Preview
For high-volume agent teams and platforms.
Scale
$949 per month
Higher-volume Paid HELM AI with multiple workspaces, higher limits, extended retention, and advanced simulations.
- Included
- 2.5M credits / month
- Limit
- 1200 actions/min per workspace across API keys
- Overage
- $6 / 10k credits, billed exact per credit
- Workspace
- 5 workspaces
- Retention
- 365 days
- Rollover
- No rollover; monthly expiry
- Multiple workspaces
- Higher rate limits
- Extended retention
- Advanced simulations and priority support
Preview
For regulated teams governing autonomous work across company systems.
Enterprise
Custom annual contract
Custom Paid HELM AI Enterprise with SSO/SCIM, legal hold, long retention, reviewed connector workflows, private deployment, SLAs, invoicing, and compliance evidence.
- Included
- Custom volume
- Limit
- Custom limits
- Overage
- Contracted
- Workspace
- Contracted
- Retention
- Contracted
- Rollover
- Contracted
- SSO/SCIM
- Long retention and legal hold
- Reviewed connector workflows and private deployment
- SLAs, invoicing, and compliance workflows
Discounts and terms
Discounts apply to the base subscription, not overage.
50% startup, research, and nonprofit discount
Eligible startup, research, and nonprofit customers can receive 50% off the self-serve base subscription.
Applies to the base subscription only. Overage, taxes, and Enterprise terms are excluded unless contracted.Billing terms
USD pricing, taxes excluded. Upgrades prorate immediately; downgrades and cancellations take effect next billing period.
Credits expire monthly with no rollover. Overage is exact per credit at the published 10k-credit equivalent and rounded at invoice total.FAQ
What is free, metered, or reviewed.
What is free?
HELM AI Kernel is free to run locally, receipt verification is free, and HELM Policy Lab starts with 500 hosted credits without signup or a credit card.
Can I buy Developer, Team, or Scale today?
The public price anchors are source-owned, but paid checkout is gated until self-serve billing, onboarding, legal, support, and Stripe smoke checks pass. Access is reviewed through the contact flow.
What does one governed-action credit include?
One credit covers one HELM boundary evaluation, including the ALLOW, DENY, or ESCALATE verdict and the signed receipt. Receipt verification consumes zero credits.
Is a HELM DENY the same as a human denying an approval?
No. ALLOW, DENY, and ESCALATE are HELM verdicts from the execution boundary. Human approve or deny actions are approval ceremony records that become evidence for later governed actions.
Where does legal hold fit?
Legal hold belongs in Enterprise with long retention, SSO/SCIM, private deployment, invoicing, SLAs, and reviewed connector workflows. It is not included in Scale.
Why publish prices before checkout?
Price visibility, signup CTAs, and paid checkout are separate. The website can show source-owned terms while the backend stays gated.