Private AI content workflow
Content Writing Agent
Krystian designed and shipped an internal AI writing workflow with channel briefs, approved sources, writing guidance, saved examples, and human review.
Exact role
Senior Web Developer (AI workflow design and technical lead)
University of Arizona
Current status
Shipped internally
2024 - Present
Problem
What needed to change
Content writers needed help producing branded, audience-specific copy because generic AI output sounded too obviously AI-generated and did not reliably follow university or department guidance.
Users
Who the system serves
Content writers, content team leadership, and division teams that need enrollment- and brand-aligned copy.
Team context
Where I fit
Built with input from content writers and content leadership, from problem framing through interface and internal production workflow.
Important technical decision
Krystian put channel, audience, approved sources, writing guidance, saved examples, and writer review around the generation step.
Tradeoff or limitation
Writers remain responsible for reviewing generated copy; the workflow does not publish it automatically.
Implementation
What I personally owned
- Owned project inception, product design, AI workflow design, implementation, coordination, prototype, and production version.
- Worked with content writers and content leadership from the initial quality complaint through the internal workflow.
- Designed the interface around channel, audience, voice, style, writing goals, source guidance, and writer review.
Constraints
Hard parts that shaped the work
- The interface had to capture channel, voice, style, audience, and writing goals without reducing the task to one generic prompt.
- Research had to remain within approved university sources and university and department writing guidance.
- Generated copy still required a writer's review before use.
Shipped result
What exists now
- Prototype and internal production workflow.
- Channel-specific briefs for voice, style, and audience.
- Saved examples of good outputs for future improvement.
- Research mode using approved university-specific sources.
- Guideline cross-reference against university and department writing standards.
Result
Verified result and current state
- The internal workflow shipped with channel briefs, approved-source research, saved examples, and writing-guideline checks.
- Writers remain responsible for reviewing generated content before use.
Technical scope