How we use AI in our own teaching.

AI is used in the following parts of the Kindra experience, transparently:

• Explain Simpler. A per-lesson assistant that rephrases the current passage at a calmer level when the learner asks. The source lesson is never modified.
• AI Skills Snapshot. Deterministic scoring of a structured 8-question instrument. No free-form generation — the output is a score and a band, not an interpretation.
• AI Practice Labs. The Prompt, Workflow, and Decision labs are scenario-based exercises where the learner’s work is critiqued by an LLM acting as a calm reviewer. The critique is presented as a perspective, not as truth.
• Recommendations. The adaptive next-action surface — “Quietly recommended” — uses simple deterministic logic over the learner’s state. No LLM is in the loop here. We do not want a chatty dashboard.

• Lesson authorship is human-led. AI may be used as a drafting assistant, but every published lesson is reviewed and owned by a named human author. Wholesale AI-generated lessons are not published as Kindra material.
• Grading and credential issuance are deterministic. No LLM decides whether a learner has completed a course. Completion is a structural property of the data — not a judgment.
• Verification is not LLM-mediated. The public verification page reads from a relational record, not a generated answer.
• We do not use AI to write the institutional pages you are reading. They are written by humans because the standards they document are commitments — not predictions.

Kindra is model-agnostic by design. We currently use Claude Sonnet (Anthropic) as the default for learner-facing AI features because of its tonal restraint. We retain the right to change models without notice if a different model better serves the learner.

No learner data is used to train any third-party model. Lesson content is sent to providers only in the context of the active feature (e.g., Explain Simpler) and is not retained for training by the institution or, to our knowledge and per their terms, by the provider.

AI is a tool that produces fluent output. Fluency is not authority. The line we hold is that no AI output at Kindra is presented as the final word — every AI surface is framed as a perspective, a draft, or a starting point. The learner’s judgment is the authority, and the institution’s authority is in how we have structured the learning around it.