No explainer videos, no multiple choice: every Curonaut course is run by a learning AI built specifically for corporate training — working with each person individually, on their real tasks, in their working world. Here is how it works. In full.
Like a good human trainer, the learning AI switches roles — depending on where the learner currently stands.
In the placement interview (12–15 min.), it first captures the workplace profile: role, degree of autonomy, communication load. Then it covertly and without exam pressure probes what is already mastered — through open narrative questions, never quizzes. Whatever is demonstrably mastered leaves the learning path.
In every building block it practises with the person on their real tasks: it asks, challenges and gives feedback — but the thinking stays with the learner. Role-plays are played out in character, from the sympathetic colleague to the stubborn customer.
Proofs of practice and module assessments are graded against fixed, predefined criteria — the same for everyone. Every verdict is justified per criterion and traceable: passing means applied, not memorised.
The AI follows fixed didactic rules we sharpened across hundreds of test conversations. The most important one: it never takes the learning moment away from anyone.
AI grading is only fair if it is traceable and repeatable. That is why we put more work into the grading engine than into any other component.
Every assessment has predefined criteria with clear pass conditions — created by subject-matter experts, identical for all learners. The AI invents no additional requirements; it checks exactly what the rubric demands.
A criterion may only be marked as not met if the AI can name the concrete gap. For every verdict it must cite the part of the answer it relies on — gut feeling is not a justification.
Every submission is graded by three independent evaluation runs; the result per criterion is the majority verdict. Ties are resolved in the learner’s favour.
"Not passed yet" means: concrete feedback per criterion and a new attempt — the existing answer is preserved and revised, not retyped. There are no grades, no rankings, no comparison with colleagues.
Trust comes from boundaries that hold — these are anchored technically, not just promised.
Learners know at all times that they are talking to an AI. Every assessment is labelled as AI-graded and justified per criterion. Grading logs are retained and professionally audited on a sample basis — findings flow back into the engine as rule updates.
Curonaut is deliberately built to be co-determination-friendly: no individual monitoring, no behavioural or performance control at content level, clear data separation. We gladly provide the technical documentation for your committees.
The fastest way to understand the learning AI is a conversation with it — in the pilot programme with your own team.
Looking for continuous AI coaching for your workforce — independent of courses? That is our second product: The Curonaut Coach →