@Human forPublic sector
Governed AI for public services your citizens can trust.
Build citizen portals, case tools and assistants that meet EU AI Act duties from day one. Fundamental rights assessed before you build, people deciding every outcome, and everything hosted in-country.
- EU AI Act Article 27 FRIA
- In-country hosting
- Works without AI
What's slowing AI down in the public sector
The pressure is real, and so are the rules. Here's what teams like yours are up against.
AI that decides access to benefits is high-risk
The EU AI Act lists AI used to evaluate eligibility for essential public services and benefits as high-risk (Annex III), with strict duties on risk, oversight and logging.
Public bodies must assess fundamental rights
Before deploying high-risk AI, public bodies must carry out a fundamental rights impact assessment. Most build tools give you nowhere to record one.
Citizen data can't leave your control
Procurement rules and public trust demand data stays in-country, with no dependency on one AI vendor that could change terms or switch off.
Applications you could build with @Human
Beautiful, everyday tools for your teams and the people you serve, each one a governed project from the first line.
CityConnect
Residents report problems and track requests on a live map. Rules route each request to the right team instantly, with no AI tokens spent on routing.
Benefits Case Review
An assistant summarises each case for the caseworker, hides special-category data, and checks the decision against EU Charter rights. The caseworker decides.
CityConnect mobile
A friendly assistant answers questions from council records, books reminders and hands anything about benefits or housing to a person.
Example apps with illustrative data, showing what can be built on @Human.
Governance built for public services
Every @Human app moves through six stages, from setup to compliance evidence. These are the parts that matter most to you.
Fundamental rights impact assessment
Risk scenarios default to the EU Charter of Fundamental Rights, covering harms, benefits and vulnerable groups, ready for Article 27.
Human decision on every outcome
Approval gates and role permissions make sure AI never approves, refuses or publishes on its own.
Reasons recorded
Every decision, change and release is logged for audits, appeals and freedom-of-information requests.
Data register to the prompt
Track where citizen data lands: fields, files, prompts, assistant memory and outputs.
Field-level access
Caseworkers see what their role needs, down to individual fields and their own cases.
Procurement-ready
100% open source, self-hosted in-country, any AI provider or none. No lock-in written into the contract.
AI can propose. People decide. In @Human, AI can prepare work but can never approve or publish a release.
Cheaper to run. Impossible to lock in.
@Human favours tested, programmatic automation wherever the choice exists, so your apps cost less, behave the same every time, and keep working without AI.
Lower running costs
Routing, reminders and status updates run as tested automation. Tokens are spent only where an assistant adds real value.
Services stay open
If the AI provider changes terms or goes down, requests, cases and payments keep flowing.
Stable releases
Changes are tried in a draft and signed off before residents ever see them.
Reuse across departments
Policies, risk frameworks and data models from one service become the starting point for the next.
Is @Human suitable for high-risk AI under the EU AI Act?
@Human is built to support the evidence high-risk systems need: risk management, data governance, human oversight, logging and records. Conformity remains the responsibility of the deploying body.
Can we host @Human in our own country?
Yes. @Human is software you run, not SaaS. Deploy it in your own data centre or with a cloud provider in your region.
Does it support fundamental rights impact assessments?
Yes. Risk scenarios default to the EU Charter of Fundamental Rights and record impacted groups, rights, harms, benefits and mitigations.
What happens if we switch off AI?
Forms, cases, routing, approvals and reports keep running as programmatic automation. AI features pause until reconnected.