Before we talk pedagogy, we need to talk plumbing. The teachers workbench, as it is configured for the workshop, does not see student data. That is a design posture, not a promise, and this post shows how to check it.
What the workbench is, and is not
The workbench is a set of small, inspectable tools a teacher uses to build gates, rubrics, and worked examples grounded in active inference (Class E, per Parr, Pezzulo, and Friston, 2022). It is a workshop artifact for teachers, not a classroom deployment aimed at learners. In the workshop configuration it runs against synthetic examples the teacher chooses, or against the teacher's own worked cases. There is no roster upload, no gradebook connector, no LMS integration, no learner identifier at rest (Class C, configuration in the workshop).
That last sentence is the load-bearing one, so let us break it down.
Four concrete data postures
- No roster upload. There is no field, endpoint, or import path in the workshop workbench for a class roster, an SIS export, or a student directory (Class C).
- No gradebook connector. The workbench does not read from or write to a gradebook system in the workshop configuration (Class C).
- No LMS integration. There is no OAuth handshake, LTI launch, or API token slot for Canvas, Schoology, Google Classroom, or similar, in the workshop build (Class C).
- No learner identifier at rest. The teacher works with prompts, gate designs, and worked examples. Names, IDs, and identifying details are the teacher's responsibility to keep out, and the workbench is not designed to receive them (Class C).
None of these are hidden features waiting behind a flag. They are absences. Absences are easier to audit than presences, which is part of the point.
How to check this yourself
The transparency posture we hold across the family is: do not take the claim on faith, test the build and inspect the gates. That applies here.
Concretely, at the workshop, a school leader or district technologist can:
- Read the config files with us at the table, and confirm the four absences above (Class C).
- Watch a live run of the workbench with synthetic examples, and confirm no network calls to an SIS, LMS, or grade store (Class B, code and inspection, when done live).
- Ask us to name every outbound endpoint the workbench can reach, and check that list against a packet capture on your own network (Class B, when done live).
If any of those checks fail, that is a finding, and findings are the point of a workshop. We would rather find a gap in the room with you than ship a claim past you.
Why this posture, and why now
There is a strong pull in the ed-tech market to hoover up student data and call it "personalization." That trade is easy to make and hard to reverse. The workbench takes a different posture: the teacher is the practitioner, the workshop is where craft is built, and the artifacts that leave the workshop (gates, rubrics, worked examples) are pedagogy, not learner profiles.
This is consistent with how we frame our larger work: UNI is a working hypothesis on an attainable path toward General Natural Intelligence, a natural, active-inference approach whose evidence is growing, evidence-classed, and tested in the open. The workbench is a teacher-facing slice of that same posture. Science in the open, receipts on the table, and student data kept out of the workshop by design.
If a classroom deployment is ever discussed
That is a separate conversation, with its own data protection review, its own consent posture, and its own gates. Nothing in this post authorizes that conversation. This post is about the workshop configuration of the workbench, and nothing beyond that.
Read next
- Our transparency posture, in one page
- The teachers workbench tour
- A school leader's checklist before any AI vendor
- Come to the workshop and inspect it with us
AI-authorship fence: this post was drafted with LLM assistance, reviewed and edited by a human author, and is published under human editorial responsibility. No claim is made that an LLM authored the reasoning. The framing is a working hypothesis on the attainable path toward General Natural Intelligence, natural not artificial, tested in the open.