How OneRange Vero's AI works
Vero's training engine teaches in short explain-example-check cycles, adapts to each answer, and grades responses against a skill rubric rather than a multiple-choice key.
- Retrieval over your own documentation grounds every explanation
- Interactive checks, scenarios, labs, and role-plays instead of passive video
- Continuous assessment produces a proficiency level per skill
Frequently asked questions
How is Vero different from a general-purpose chatbot like ChatGPT?
A general chatbot answers whatever you ask. Vero is a training system: it works from a defined curriculum, teaches in short explain-example-check cycles, and will not skip ahead until a learner demonstrates understanding. Every explanation is grounded in your own documentation rather than the open internet, and every exchange is scored against a skill rubric so the outcome is a measured proficiency level, not a transcript.
Does Vero's AI hallucinate answers about our internal processes?
Vero retrieves passages from your own uploaded documents, GitHub repositories, and Notion pages before it answers, and it grounds explanations in that retrieved material. When the knowledge base does not cover a question, the tutor says so and steers back to the curriculum instead of inventing an answer. Course authors can see exactly which reference documents a course draws on.
How does Vero keep our company data private?
Every document, course, assessment, and result is scoped to your company at the database level with row-level security, so no other customer's Vero instance can read your content. Your material is used to ground your own training only — it is never used to train shared models.
How does Vero grade open-ended answers?
Each skill in a course carries a rubric describing what a novice, competent, and expert response looks like. The AI grades a learner's written or spoken answer against that rubric and reports the level reached, along with the specific gap that held the score back. That is why results are expressed as proficiency, not a percentage of clicks completed.
What keeps a learner from getting stuck or gaming the assessment?
The tutor adapts to each answer: a wrong response triggers a reframed explanation and an easier check rather than a repeat of the same question, and a strong response accelerates past material the learner has already proven. Because answers are open-response and graded against a rubric, there is no answer key to guess at.
Which content can Vero build training from?
Anything your team already has: PDFs, policy documents, slide decks, product specs, a connected GitHub repository, or a Notion workspace. Vero extracts and indexes that material, then generates a course outline, topics, interactive exercises, and an assessment from it, which an author reviews and edits before publishing.