MeetPlanet Crust · @Human: The Sovereign AI Enterprise System
The AI governance platform for ISO 42001 & EU AI Act compliance.
Build, run and prove your AI agents and automations. Full event logging, risk management, incident tracking and real-time alerting, every AI system classified against the EU AI Act and ISO 42001. Sovereign, self-hosted, under your control.
- EU AI Act · ISO 42001 · GDPR
- Self-hosted
- Any AI model
Trusted by organisations building secure enterprise systems.
@Human For EnterpriseCore systems, governed by design
Run your core enterprise systems on governed AI
Finance, insurance and CRM are where AI pays off fastest, and where one bad decision costs the most. @Human builds these systems with policies, approvals, guardrails and evidence from the first line, on infrastructure you control.
Finance operations
Payables, approvals, spend and month-end close, with invoices matched by rules and anomalies explained.
- Three-way matching runs as automation, with no tokens
- Payments over limit need a named approver
- Every anomaly is explained before anyone pays
Claims and complaints
Claims intake, explainable fraud signals, settlement and Consumer Duty reporting in one governed system.
- Explainable signals, never a black-box score
- Vulnerable customers routed to a named handler
- Complaints and outcomes ready for the board pack
Accounts and pipeline
Pipeline, accounts, contacts and forecasts, with account briefs drafted for the owner and consent checked on every send.
- Consent checked by rule before every email
- Account briefs drafted by AI, owned by people
- Renewals and stale deals chased automatically
Plus HR and workforce, service desks, procurement and any system your teams need, each one a governed project with signed approval gates, runtime guardrails and EU AI Act and ISO/IEC 42001 evidence.
Beautiful apps, with the governance built in
These screens come from projects built on @Human: maps, dashboards, rosters and portals your teams and the public actually use.
Enterprise-grade software, not prototypes
Full applications with data models, pages, workflows, agents and dashboards, each one a governed project from its first line.
Example applications with illustrative data. Dashboards are native @Human charts.
Built for teams and communities
Apps for the people you serve, with protections that come from the governance record itself.
Biodiversity Watch
Citizens report poaching, pollution and illegal logging on a live map. Rangers verify reports and plan action.
Programme Impact
Monthly returns become board-ready charts and insights, every figure traceable to a reviewed, locked return.
Ward Rostering
Shifts, attendance, leave and corrections for hospital wards, with decisions surfaced to the right manager.
Every app is a governed project
Service desks, accounting, classrooms, content calendars and risk checkers live side by side, each with its own policies, risks and approvals.
Dashboards your board will actually look at
Every chart here is a native @Human chart, built from reports over a single module of 48 orders in our Fireworks showcase. No custom code, no extra BI licence: heatmaps, Sankey flows, sunbursts, treemaps, ThemeRivers, calendars and every classic, all governed like the rest of the app.
Your AI is moving faster than your governance
Teams buy one tool to build AI and another to document it. Developers ship on Monday, the risk register catches up next quarter, and the gap between them is exactly where audits fail and projects get blocked.
The top EU AI Act penalty
Of worldwide annual turnover, whichever is higher, for prohibited AI practices. Most other breaches can reach €15m or 3%.
One gap in the middle
Governance suites document AI built elsewhere. App builders leave governance to admin settings. Nobody keeps the two in step.
Is a chance to drift
A new field, prompt or permission can quietly change your risk. If the record isn't updated, your evidence is already out of date.
Penalty figures from Article 99 of Regulation (EU) 2024/1689, the EU AI Act.
Three promises built into every @Human app
Most platforms promise speed. We promise control: over risk, over cost, and over who you depend on.
Nothing goes live ungoverned.
Policies, the sensitive data register and rights-based risk scenarios come before the build. Named people sign every release, and AI can never approve or publish on its own.
You never pay a model to repeat itself.
AI helps you build. The app then runs on tested, programmatic automation wherever the choice exists, so routine work costs nothing in tokens and behaves the same every time.
You can always unplug.
No lock-in to any AI provider. Switch models without rebuilding, or disconnect AI entirely, and your forms, rules, records and reports keep running.
Responsible AI that also makes business sense
@Human builds the app and its governance record in the same project, and is designed to keep both your running costs and your risk low.
- ✓Policies written and owned
- ✓Sensitive data register complete
- ✓Risk scenarios for harms and benefits
- ✓Governance owner signed
- ✓Data protection officer signed
- …Executive authority to sign
Governance first
Policies, a sensitive data register and rights-based risk scenarios are written before anything is built. Only people can approve and publish, and every later change is re-checked automatically.
Lower token costs
Routine work runs as tested automation, so you only pay for AI where it adds judgement.
Data sovereignty
Software you run, not SaaS. Your data stays in your infrastructure, with any AI provider, or none.
Reusable codebase
Data models, pages and automations you can see, test and reuse, on an open-source core.
Stable from day one
Pre-testable in a draft, predictable in production, with every version kept on record.
From AI idea to compliant, live system in three steps
The same discipline every time: propose, confirm, approve. AI can help with the work; people make the decisions.
Classify and set the rules
Describe the AI system and its purpose. @Human classifies it against the EU AI Act and ISO/IEC 42001, then you set policies, the sensitive data register and risk scenarios.
Governance exists before the buildBuild it governed
Create the data model, pages, automations and AI assistants in the same project. Every change raises an impact review against your rules.
Programmatic first, agents where neededApprove, run and prove
Named people sign the approval gates. Once live, event logging, incident tracking and alerting keep the evidence current for your auditors.
Evidence that stays trueBring one AI system to a 20-minute walkthrough and see it classified, logged and reported inside @Human.
Governance isn't a review at the end. It's how the app gets built.
Every @Human application is a governed project that moves through six stages. Your risk register and policies exist before the first screen is designed, and your compliance evidence builds itself as you go.
- SetupPeople, roles, approval gates and providers
- Policies and riskWritten before anything is built
- BuildData, pages, automations, assistants, apps
- GovernRisk re-checked against what was built
- ManageIncidents, requests, reviews and reports
- ComplyEU AI Act, ISO/IEC 42001 and GDPR evidence
- Rights-based risk assessmentHarms and benefits assessed against the EU Charter of Fundamental Rights (FRIA), plus any sector rules you answer to.
- A data map that follows data into AIEvery place personal data lands, including prompts, assistant memory and AI outputs.
- Versioned policies and privacy noticesOwners, clause history, and end users re-accept when it changes.
- Field-level permissionsDecide who sees each field, down to "only their own records".
AI can propose. People decide. AI can prepare work in @Human, but it can never approve or publish a release.
Change something. Governance follows.
Governance records usually drift from what is actually running. In @Human they can't. Edit a form, an automation, an assistant or a permission, and an impact review opens on its own.
- New personal data becomes a register entryWith its purpose, retention and who can see it.
- New purposes call for new risk scenariosAssessed before release, not discovered in an audit.
- The privacy notice is re-versionedUsers re-accept it where the change affects them.
Spend AI on thinking, not on repeating
Many AI platforms run your business process through an agent that calls a model every single time. @Human uses AI to help you build, then favours programmatic automation wherever the choice exists. Lower LLM token costs, and behaviour you can test before it reaches anyone.
Agent-first automation Common elsewhere
A model decides what to do, every time it runs.
- Tokens billed on every run, so costs grow with use
- Can answer the same input differently
- Hard to test fully before going live
- Stops when the model or provider is unavailable
Programmatic first @Human
AI helps design the automation. The automation then runs on its own.
- No tokens spent on routine runs
- Same input, same result, every time
- Pre-testable in a draft before release
- AI agents kept for tasks that need judgement
Where the token bill comes from
With an agent, every run is billed. With programmatic automation, AI is used while you build and routine runs cost nothing in tokens.
Illustrative only. Actual costs depend on your model, provider and volumes.Pay for AI where it adds judgement, not for routine steps.
Automations and permissions are tried in a draft version first.
Predictable behaviour your team and auditors can rely on.
Where both would work, @Human picks the one you can test.
Your AI, your choice. And your app keeps running without it.
Sovereignty means no lock-in to any AI. Choose any model or provider, change your mind later, or unplug AI completely. Your forms, rules, records and reports carry on.
- No AI vendor lock-inEU-hosted, hyperscaler or open models, switched without rebuilding the app.
- Operational if unplugged from AICore processes run as programmatic automation, so an outage or policy change doesn't stop your service.
- Software you run, not SaaSInside your own infrastructure and security controls. Planet Crust doesn't host or process your production data.
Free 20-minute governance walkthrough
Bring one AI system you are unsure about, and leave knowing exactly where it stands.
- How it would be classified under the EU AI Act
- Which ISO/IEC 42001 controls it touches
- How it would be logged, monitored and reported
- Where you could cut token costs
Build once. Reuse with confidence.
An @Human app isn't a pile of prompts. It is a clear set of parts you can read, test, version and carry into the next project.
Parts you can see
Data models, pages, charts, automations, assistants and custom web apps, on the open-source Corteza low-code platform.
Every version kept
Work happens on a draft. Approved versions go live, and earlier versions stay on record for audit and roll-back.
Governance you can reuse
Policies, risk frameworks and data register patterns from one project become the starting point for the next.
One platform instead of two purchases
AI governance tools document AI built somewhere else. AI app builders leave governance to admin settings. @Human closes the gap.
| Capability | AI governance tools | AI app builders |  |
|---|---|---|---|
| Builds the app: data, pages, automations, assistants | No | Yes | Yes |
| Policies and risk written before the build | Separately | No | Yes |
| Rights-based risk assessment (FRIA) | Yes | Rarely | Yes |
| Data register down to prompts and AI memory | Partly | Partly | Yes |
| Re-review triggered by every change | No | No | Yes |
| Human approval gates before go-live | Partly | Partly | Yes |
| Runs without AI connected | Not applicable | Varies | Yes |
| Open source, self-hosted, any AI model | Rarely | Varies | Yes |
@Human supports your compliance evidence; conformity remains the responsibility of the organisation deploying the AI.
Made for organisations that answer to someone
Regulators, boards, funders, patients, citizens. If someone will ask how your AI was governed, @Human gives you the answer.
Public sector
EU AI Act high-risk duties, fundamental rights impact assessments and in-country hosting.
Case triage, citizen servicesHealthcare and social care
Special-category data, patient safety and DPO scrutiny, with the register following data into AI.
Rostering, referral triageFinancial services
Model risk, consumer duty and auditability through approval gates and versioned policies.
Claims intake, complaintsEducation
Minors' data, safeguarding and fairness, assessed with vulnerable groups in mind.
Classrooms, learner supportNGOs and charities
Funder and donor rules as custom risk frameworks, at a cost that fits the budget.
Programme reporting, beneficiariesManaged service providers
Governed AI for every customer, with many projects on one instance.
Service desks, knowledge assistantsHas your legal, risk or data protection team blocked or delayed an AI project this year?
Pay what you can. Seriously.
@Human is open source and the whole platform is yours, whatever you pay. Most people start at a few euros a month.
a month
If money is tight. Students, researchers, anyone learning the platform.
Start at €1a month
What we suggest if @Human is doing real work for you. Enough of these and the project pays for itself.
Start at €10a month
If it saves you more than it costs and you want the roadmap to keep moving.
Start at €25Any amount above €1 works. Change it or stop at the end of any month, and nothing in your install changes either way.
The same platform at every amount
- Unlimited users, agents and automations
- Every module, nothing held back
- Full activity log and audit export
- Risk classification against the EU AI Act
- Self-hosted, or hosted in an EU region
- Any LLM you like, Mistral included
Need more than the forum? Support plans add our time, not more features.
Prices exclude VAT. See the full comparison
Free 20-minute governance walkthrough
Bring one AI system you are unsure about, and leave knowing exactly where it stands.
- How it would be classified under the EU AI Act
- Which ISO/IEC 42001 controls it touches
- How it would be logged, monitored and reported
- Where you could cut token costs
