AI assistants in Learnivo Beta
Learnivo includes seven AI assistants. They are Beta: functional and evolving, with known limitations. What sets them apart is governance — AI drafts and explains; humans approve publication and consequential decisions.
The seven assistants
| Assistant | What it does | Plans |
|---|---|---|
| Learner tutor Beta | Answers learner questions from your own course content, not from the open internet. | Pro+ |
| Course authoring copilot Beta | Drafts lesson content to your brief; authors edit, review and publish. | Every plan |
| Assessment assistant Beta | Helps build and review assessments and question banks. | Pro+ |
| Admin insights agent Beta | Summarises what's happening across your organisation and suggests where to look next. | Pro+ |
| Translation assistant Beta | Drafts content translations; a human review workflow approves them before publication. | Pro+ |
| Organisation setup wizard | Walks new administrators through configuring their organisation. | Every plan |
| Support guide Beta | Guides users through features; escalates complex queries to human support. | Pro+ |
What "human-governed" means here
- AI never publishes to learners or makes final assessment decisions by itself.
- Administrative actions — sending emails, enrolling people, changing data — are mediated by a human-approved proposal workflow on every plan.
- Every AI action is written to agent audit logs (Pro and above), so you can see what AI did and what human approved it.
- Each plan includes a monthly AI-credit allowance — 100 (Free), 10,000 (Pro), 50,000 (Max), unlimited (Enterprise) — with usage visible to administrators.
Honest limits, per capability
- AI evaluation of submissions Beta (Pro+) — advisory only; staff-only actions produce final grades.
- AI roleplay coaching Beta (Pro+) — advisory feedback against a fixed rubric; practice only, final grades are recorded by staff.
- Similarity review Beta (Pro+) — runs on a simulated provider adapter today; it makes no compliance claims.
- Tutor and insights — grounded in your organisation's data; outside their scope, they say so rather than guessing.
Plan availability
- Authoring copilot, setup wizard, human-governed approval workflow — every plan, including Free
- Learner tutor, admin insights, support guide, translation assistant, assessment assistant — Pro plans and above
- Agent audit logs — Pro plans and above
Questions people ask
Can AI contact our learners on its own?
No. Emails, enrolments and other administrative actions go through a human-approved proposal workflow — a person approves each action before it runs.
Does AI grade work?
AI evaluation is advisory only and labelled as such; final grades are always recorded by staff.
Where does the tutor get its answers?
From your own course content and organisation data. Out-of-scope questions get an honest "that's outside what I can help with", not an invented answer.
What happens when the Beta ends?
The assistants keep working and graduate out of Beta once their evaluation cases — a live suite that tests honest fallback, approval gating and tenant isolation — pass consistently. Status labels on the feature register are kept current.