Tessarion

System · Implemented

Product overview

What Tessarion does, how the learner loop works, and where its boundaries are.

What Tessarion does

Tessarion turns source material into an evidence-linked concept model. A learner explains one concept in their own words; the system compares the explanation with source evidence, concept dependencies, prior mastery signals, and tutoring history before selecting the next action.

The product does not reduce understanding to a single score. It records covered ideas, omissions, misconceptions, unsupported claims, evidence references, and the route selected from those findings.

Evidence-linked learning loopA learner explanation creates evidence. The system chooses the next action from that evidence.

Learner loop

1

Add a source document or paste focused study material.

2

Create source-linked chunks and extract concepts and relationships.

3

Choose a concept and teach it back without copying the source.

4

Classify coverage, omissions, misconceptions, unsupported claims, and weak prerequisites.

5

Route to another teach-back, Socratic tutoring, or review.

6

Record new evidence and repeat.

System boundaries

  • Postgres owns canonical user, workspace, source, concept, mastery, review, tutoring, and checkpoint records.
  • Vector and graph systems are derived and rebuildable.
  • Generated output is validated before it can update learner state.
  • Tutoring guides recovery but does not mark a concept as understood.
  • The public demo is deterministic and writes no account data.