AI in course production
Course production, rebuilt around AI agents
Building a course took 12 weeks and 90 hours of subject experts’ time. I designed and built TRICIA 2.0, a set of topic-specific AI agents that draft the work inside a review process people control.
TRICIA 2.0. United Way Worldwide, 2026. 11 courses. I built the agents, the review gates, the rubrics and the acceptance criteria.
Workflow
How a course moves
- Brief. The course starts from our frameworks, learner personas and design standards.
- Agents draft. Topic-specific agents draft the outline, the modules and the job aids.
- Expert review. A subject expert reviews every draft and can reject it or send it back.
- Learning QA. Each course is checked against a rubric and acceptance criteria before it ships.
- Publish. 11 courses were built this way in 2026.
Try it
Review an AI draft
You are the subject expert. An agent drafted this page of a course on sharing data with partners. Mark every line you would not publish, check the source if you need to, then approve the page or send it back.
Agent draft
Module 2. Sharing data with partners
Click a line to mark it. Click again to clear it.
The framework the agent drafted from
- Share reports only with the people they are for.
- Personal data, including names and contact details, is shared only through the approved tool, with a named owner.
- Set shared links to expire.
- Remove names from anything published.
- When unsure, ask the data lead.
An example page, written for this site. In TRICIA 2.0 a subject expert reviews every draft against its source, and can reject it or send it back. A checkpoint only counts if the reviewer can say no.
Results
12 weeks to 4
Why it works
What made it work
Reviewers can say no
Real review needs knowledge, the source, time and the authority to reject. Both gates can send a draft back, and nothing publishes without them.
The review was redesigned too
Faster drafting is worthless if review becomes the bottleneck, so I redesigned the review along with the drafting: the review gates, quality rubrics, evaluation guidelines and acceptance criteria the experts work from.