Faculty designing a Blackboard or Canvas course want an agent that drafts learning objectives, aligns assessments, and proposes module structure. They do not want a chat log where decisions disappear. They want each draft objective attributed to the agent that wrote it, reviewed by a named faculty member, and tied back to the course shell in the LMS. Dock is where that record lives.
Blackboard Learn and Canvas stay the system of record for the raw data. Dock is the system of record for what the agent interprets. Each Dock row carries a pointer back to the platform record, agent identity, decision, reviewer, and timestamp. The agent re-fetches platform data via fresh API reads when it needs current state. This is the same pattern we describe in the broader education rail and in the research workspace pieces.
Surface: Course Design Review
| Module | LMS course ID | Draft objective | Bloom level | Assessment alignment | Agent | Faculty reviewer | Status |
|---|---|---|---|---|---|---|---|
| Wk 3: Regression | bb_course_PSYC241_S26 | Students will distinguish predictor from outcome variables in a multiple regression model | Analyze | Quiz 3 Q4-Q7 | ada-curriculum | Prof. Lin | Approved |
| Wk 5: ANOVA | bb_course_PSYC241_S26 | Students will compute a one-way ANOVA by hand and verify with R | Apply | Lab 5 rubric | ada-curriculum | Prof. Lin | Changes requested |
| Wk 7: Power | canvas_course_PSYC241_HONORS | Students will calculate required sample size for a given effect size and alpha | Apply | Problem set 7 | ada-curriculum | Prof. Okafor | Pending review |
Each row links back to the LMS course shell. The agent does not store the syllabus inside Dock. When Prof. Lin opens the Wk 5 row, the agent re-fetches the current Blackboard module from the Anthology API and shows the live state next to its proposed change.
Workflow: drafting and approving a week's objectives
The faculty member opens the Wk 5 row and asks the agent to revise the objective so the verb matches the lab activity. The agent reads the current module from Blackboard, drafts a new objective using the action-verb conventions from the Carnegie Mellon Eberly Center's course design framework, and writes the draft into the row. The reviewer field stays empty until Prof. Lin approves. On approval, Dock writes a signed event with the agent ID, the reviewer ID, the timestamp, and a hash of the approved text. Only then does the agent push the change to Blackboard via the Anthology API. The push event lands back in Dock as a confirmation row. If Prof. Lin reverts the change in Blackboard directly, the next fetch surfaces the divergence in the row's status field. This is the attribution pattern documented in our audit and compliance piece.
Why it matters
Blackboard publishes more than 450 user-requested features and ships native generative AI for course building (Anthology). The Eberly Center documents that effective course design aligns objectives, assessments, and instructional strategies, and that vague verbs like "understand" should be replaced with measurable ones (CMU Eberly). Generative drafts move fast. Faculty review does not. Dock holds the gap. Every objective the agent proposes carries a named author and a named reviewer, which matters for accreditation reviews, departmental curriculum committees, and the faculty member's own peace of mind. The agent has its own identity, separate from any human, and its proposals are reviewed before they reach the LMS, as outlined in the collaboration primer.
Get started
Spin up a Course Design Review surface in Dock, connect your Blackboard or Canvas instance, and give the curriculum agent an identity of its own.
FAQ
Does Dock replace Blackboard or Canvas? No. The LMS remains the system of record for course shells, enrollments, grades, and student submissions. Dock holds the agent's interpretive layer: drafts, decisions, reviewers, and timestamps.
What happens if a faculty member edits the objective directly in Blackboard? The agent re-fetches the LMS state on the next read. The Dock row shows the divergence in its status field, and the agent flags the row for re-review rather than overwriting the human edit.
Can two agents work on the same course? Yes. Each agent has a distinct identity, and each row records which agent drafted which field. A co-design agent and a rubric-alignment agent can write to different columns of the same row without losing attribution.
Is the approval signature usable for accreditation review? The signed event includes agent ID, reviewer ID, decision, timestamp, and a content hash. Accreditors who accept digital audit trails can verify the chain. Institutions with stricter requirements export the log as a signed PDF.
