Lit Booster

Lit Booster was built inside Render, CZI Education's AI innovation studio, to help teachers address post-pandemic reading deficits.

Post-pandemic literacy gaps were real and severe. Teachers knew what interventions to use, but they had no way to generate the materials to actually run them. A reading passage calibrated to a 5th grader reading at a 2nd grade level, about Minecraft or Taylor Swift, doesn't exist off the shelf.

We decided AI could generate it. But that decision unlocked a harder problem.

How do you design a reliable UI on a non-deterministic system with outputs accurate enough for a struggling reader?

The prompt is the product. Every dropdown (resource type, grade, Lexile level, student interest) was a parameter injected into an OpenAI call. Designed the input architecture in direct collaboration with the data scientist. Wrong dropdown structure meant bad AI outputs, so the form fields had real stakes.

Designing for iteration without full regeneration. Built modifier buttons (Shorter / Longer / Less Challenging / More Challenging) as additive re-runs so teachers could refine an output without losing the version they'd already spent time on.

Pivoting the information architecture mid-product. Reframed the core IA from strategy-first to materials-first through rapid shipping and iteration -- research from two pilots made clear teachers wanted to generate materials directly, not navigate a strategy library first. Reframing the IA required PM and eng alignment on what the MVP of a standalone flow actually needed to ship.

Lit Booster shipped as a two-part tool: a library of 20+ evidence-based reading intervention strategies across 5 skill domains, and an AI materials generator that produced practice texts and word lists calibrated to a student's grade level, reading level, and interests.

Teachers could generate a passage about Fortnite at a 3rd grade reading level in under a minute, then refine it with one click. The final version decoupled the generator from the strategy library entirely, making it the primary entry point based on how teachers actually used the product.

The product was sunsetted, but the infrastructure we built to validate its AI outputs became a platform used across CZI.