Every merge approved. Every release traceable. Learners' data under your control. All inside your own boundary.
Your digital teams deliver services for schools, families and teachers faster without loosening a single control. AI agents do the routine work, and every step lands in a record your data protection, security and audit teams can read.
For digital, engineering, platform, security and data protection teams at ministries of education, exam and qualification bodies, student finance agencies and their suppliers, in the UK, Ireland, the EU and the US.
Where the hours and the risk go.
Your AI agents take the routine work, your people keep the decisions.
Sensitive records kept inside
Learners' data, safeguarding records and unpublished exam papers stay in your installation. When your AI agents run on your own models, none of it reaches an AI company.
No training on your data
JuhJuh trains no AI models of its own, and the open-weight models inside your boundary run under your control. Pupils' and teachers' work is never used to train a model by JuhJuh.
People decide about learners
AI agents never make a decision about a learner, a school or an exam result. They prepare the facts for the person who decides.
Ready for results days and deadlines
The SRE on-call AI agent opens the incident, finds the release behind it and has the rollback ready for one tap, so services stay up when families need them.
Evidence for GDPR, Children's Code, FERPA, COPPA
Access by role, retention per channel and a record of who did what support your GDPR and DPIA work, the Children's Code, FERPA and COPPA.
Every change reviewed, then approved
A reviewer AI agent on a different model checks every acceptance criterion, and a person decides the merge. Ticket, review, build and release sit in one record.
Builds from your approved libraries
Mirrored copies of the open-source packages you approve sit inside your boundary, and updates arrive as reviewed changes.
Incidents with a timeline
Every incident has its timeline written as it happens, so your incident manager starts a possible 72-hour breach report from the facts.
The same week, with JuhJuh.
Alerts arrive from separate tools, and the on-call engineer works out the cause across chat, phone and spreadsheets.
The SRE on-call AI agent opens the incident, finds the release that caused it and sends the on-call engineer a voice note. She approves the rollback with one tap, and the service is healthy by 2:31am.
Engineers spend days gathering approvals, builds and release notes from several tools by hand.
The compliance evidence AI agent assembles the record by 5pm, and the head of engineering signs it off.
Engineers go without AI help, or use unapproved tools nobody can see.
JuhJuh runs in the department's own cloud account with open-weight models beside it. Engineers get AI help inside the boundary, and nothing about the data rules changes.
Every role gets time back.
A person approves anything consequential.
What your AI agents never do on their own
- Make a decision about a learner, a school or an exam result.
- Merge a change. A person decides, after a reviewer AI agent on a different model has checked it.
- Deploy to production, roll back or change production settings, access or secrets without a person's approval.
- Send anything outside the organisation beyond the answers you've approved in advance.
- Use an AI model your rules haven't approved.
- Decide whether to report an incident to a regulator. Your incident manager decides.
Self-hosted in your own cloud account
JuhJuh runs as your own installation, usually in your own cloud account in the region you choose, or in your own data centre. Open-weight models from the Mistral, DeepSeek, GLM and Kimi families run inside it, so learners' data and your sensitive work stay in your boundary.
Run it connected, reaching only the outside services you choose and public model families such as Claude, GPT and Gemini only where your rules allow, or fully air-gapped with no connection to the public internet. Nothing leaves your boundary unless you connect it or a person approves it.
Bring AI to your digital services, safely.
Start with one or two teams, and add the rest when you're ready. That's the juju.
