Reflection
What it meant to design a curriculum authoring studio
A first-person account of building enterprise software for curriculum teams — why curriculum authoring, not lesson generation, is the real instructional-design challenge, and what it taught me about turning expertise into a product.
Designing enterprise software for curriculum teams
I set out to design for an organization, not an individual. A lesson generator serves one teacher in one moment; a curriculum studio serves a team of directors, authors, subject-matter experts, reviewers, and publishers who hand work back and forth over weeks. That shift changed everything. I had to think about shared sources of truth, roles and permissions, review gates, version history, and audit trails — the unglamorous scaffolding that makes professional software trustworthy. The hardest design decisions were not about features but about the seams between people, and about making a complex, multi-week process feel coherent inside a single environment.
Why curriculum authoring is the real instructional-design challenge
It is tempting to treat a good lesson as the unit of instructional design, but a great lesson inside an incoherent program does not move learning. The real challenge is the architecture: aligning standards, sequencing a year, ensuring vertical and horizontal coherence, balancing cognitive rigor across a unit, and keeping assessments tied to outcomes. That is the work that takes months and the work no chatbot addresses. Building this studio forced me to keep my attention on the whole program — the level at which curriculum either holds together or quietly falls apart.
Human-in-the-loop and pedagogical integrity
The more capable the generation became, the more important the review workflow felt. I designed the studio so AI can never publish on its own — every artifact is a labeled draft that an expert must approve. This is not caution for its own sake; it is how pedagogical integrity survives automation. The platform can draft a unit in minutes, but a person still has to decide whether it is right for these learners, this context, and this community. Keeping that decision visibly human is the feature I am most committed to.
UX for expert authors
Designing for experts is different from designing for novices. Curriculum authors do not want hand-holding or a single magic button; they want speed, control, and the ability to override anything. So I favored a consistent input-and-draft pattern, a keyboard-first command palette, version history, and a dark mode for long sessions — efficiency tools, not training wheels. The interface had to respect that the author knows more about pedagogy than the model does, and stay out of the way once it has delivered a strong starting point.
Integrating AI into instructional-design workflows
The breakthrough was deciding that AI should live inside the workflow rather than beside it. Instead of asking authors to leave their work, copy a prompt into a separate tool, and paste results back, I embedded generation, alignment, and transformation at each stage of the pipeline. AI drafts in the builder, checks alignment in the standards tool, re-levels content in publishing — always handing a structured, editable artifact back into the same environment. Integration, not novelty, is what makes the assistance actually save time.
Translating instructional-design expertise into commercially viable edtech
This project is where my instructional-design practice meets product thinking. The studio encodes the same rigor I bring to any curriculum — backward design, standards alignment, UDL, Bloom's and DOK balance — but packages it as software a ministry, district, or publisher could adopt and scale. Building it taught me that domain expertise is a moat: the templates, the review gates, and the alignment logic are only as good as the pedagogy behind them. That is the difference between a generic AI wrapper and a product that curriculum professionals would trust with their work.
What it demonstrates
This build demonstrates that I can take deep instructional-design expertise and translate it into the architecture of a commercially viable platform — defining the workflows, governance, and human-in-the-loop safeguards that let AI accelerate curriculum work without compromising pedagogical integrity. It is evidence of leading AI innovation in education at the level of systems and products, not just individual lessons.
What I'd build next
If I carried this forward, I would deepen the alignment engine to map continuously across multiple standards frameworks, add automated accessibility and bias review on every artifact, and build the translation and multi-language publishing that ministries need to release equitable content at scale. I would also invest in the enterprise layer — single sign-on, roster sync, and API connections to LMS, SIS, and publisher pipelines — so the studio becomes the authoring hub at the center of an institution's content, not an island beside it.