Dock for education is the workspace where an AI agent's pedagogical reasoning lives: the lesson plan it drafted from a learning objective, the assessment items it generated, the per-student grading rationale, and the teacher's review on each. The LMS keeps the course and the gradebook. Dock keeps the trail of how every artifact was made, by which agent, on what evidence, and which educator signed off before it reached students.
Canvas LMS, Google Classroom, Schoology, PowerSchool, and Blackboard stay the system of record for raw course data: roster, assignments, submissions, official gradebook entries. Dock is the system of record for what the agent interprets from that data. The drafted lesson plan, the assessment rationale, the reviewer's sign-off, the audit log. Each Dock row carries a pointer back to the platform record (canvas_course_id, google_classroom_course_id, schoology_section_id), agent identity, decision, reviewer, and timestamp. The agent re-fetches roster and submission state via fresh API reads when it needs current data. Dock holds the persistent interpretive layer that survives across terms, teacher handoffs, and accreditation reviews.
The lesson-and-assessment table
A single Dock table holds the unit of educator work, one row per drafted artifact.
| artifact_id | type | course (canvas_course_id) |
learning_objective | agent_draft | educator_review | status | sign_off_at |
|---|---|---|---|---|---|---|---|
| ART-4411 | lesson_plan | BIO-201 (cv_88123) | LO-3.2: cellular respiration stages | 45-min lab plan, 3 stations, exit ticket | ms.alvarez: reordered station 2 before 3 | approved | 2026-05-22 14:11 |
| ART-4412 | assessment | BIO-201 (cv_88123) | LO-3.2 summative, 12 items | 8 MCQ + 4 short-answer, rubric attached | ms.alvarez: replaced item 7 (ambiguous) | approved | 2026-05-22 14:34 |
| ART-4419 | grading_rationale | BIO-201 (cv_88123) | Student S-204, item 9 short-answer | 7/10: partial credit, missed Krebs cycle output | ms.alvarez: agreed, added comment for student | published | 2026-05-24 09:02 |
An agent identity column (omitted above for width) names which agent drafted each row, citing the agent identity model that lets districts attribute work to a specific service principal, not a shared login.
One workflow, end to end
A biology teacher opens a new unit on cellular respiration in Canvas. The lesson-planning agent reads the course's learning objectives via the Canvas API, drafts a 45-minute lab plan plus a 12-item summative, and writes both to Dock as rows ART-4411 and ART-4412. Each row links to canvas_course_id: cv_88123. The teacher reviews in Dock, reorders one lab station, replaces one ambiguous assessment item, and approves. Only after sign-off does the agent push the approved assessment back to Canvas as a published quiz. When students submit, a grading agent drafts per-item rationale rows, the teacher approves, and the rationale becomes the comment students see in the Canvas gradebook.
Same shape as Dock for research and Dock for design: the platform records the artifact, Dock records the reasoning.
Why it matters
Lesson plans and assessment rationale are the most expensive intellectual work in a school, and the most often lost. They live in a teacher's head, a shared drive, or a comment thread that ages out. When an agent drafts them, the loss compounds. Dock makes the trail durable.
Accreditation bodies and district auditors ask for evidence of objective alignment and human review. A Dock row with learning_objective, agent_identity, reviewer, and sign_off_at answers that directly, in the shape an agent audit and compliance review needs.
It also makes the teacher's edit the canonical artifact. The agent's first draft is preserved, the teacher's revision is preserved, and the diff between them records the educator's professional judgment, exactly what an agent collaboration primer treats as the institutional asset.
Start a Dock workspace for your education team.
FAQ
Does Dock replace Canvas or Google Classroom? No. Canvas, Google Classroom, Schoology, PowerSchool, and Blackboard stay the system of record for rosters, assignments, and gradebook. Dock holds the agent-drafted lesson plan, the assessment rationale, and the educator's sign-off, with a pointer back to the LMS course id.
How does Dock track which agent drafted a lesson plan?
Every Dock row carries an agent_identity field tied to the provisioned service principal that ran the draft. See agent identity lifecycle for how identities are issued, rotated, and retired across a school year.
Can students see the agent's reasoning? Only what the educator approves. The agent writes a draft to Dock, the teacher signs off, and only the approved rationale is published back to the LMS as a student-visible comment. The full draft history stays in Dock.
Is this compliant with FERPA? Dock holds interpretive metadata, not student PII beyond what the educator pulls in for review. Roster and submission records stay in the LMS. Per-row agent identity and reviewer sign-off support the audit posture FERPA reviews ask for. Districts should run counsel review against the OECD Education Working Papers (oecd.org), the EDUCAUSE Horizon Report (educause.edu), and the Carnegie Mellon Eberly Center assessment guidance (cmu.edu).
