Praxis: AI-assisted classroom walkthrough platform

Praxis helps school leaders capture classroom observations, deliver meaningful feedback to teachers, and track growth over time. It runs across web, iPad, and iPhone, with an AI layer that synthesizes raw walkthrough notes into structured feedback the moment an observation ends.

The product is built for districts: Microsoft and Google sign-in, scoped administration across schools, audit trails, configurable rating scales, transactional notifications, and the kind of operational discipline that K-12 IT teams expect before they sign anything. The mockups in this case study are the artifact we delivered, and the spec we built against.

Most enterprise builds at this scope are 12 to 24 month undertakings. Praxis went from a blank scoping doc to a production deployment in roughly three. The work happened in discrete phases, each with a clear artifact at the end of it.

Sign-in is backed by Firebase, with both Google and Microsoft as identity providers. A JWT-only middleware on the API keeps the surface area small. The split-panel layout doubles as a brand moment for the first impression a district gets of the product.

The dashboard is built around a prioritized, AI-generated feed of what's actually worth a leader's attention this morning. Attention alerts, trends, bright spots, and anomalies are surfaced as cards, each anchored to evidence and a one-click action. The product reads the data so the user doesn't have to navigate charts to find the story.

System administrators can operate inside any district or school from a single header dropdown. Each override is validated server-side and written to an audit log. The dashboard behind the switcher is dimmed to make the cross-scope state explicit, so an admin always knows when they're viewing outside their home scope.

Each district picks the rating scale used across all walkthroughs and look-fors: binary, three-point, five-point, or fully custom, with an optional N/A. The AI feedback prompt adapts to whatever scale is active, and each rating row writes a snapshot of its scale at capture time so historical reports stay legible after a change.

At the close of a walkthrough, Praxis suggests three framed next steps: focused, stretch, and reinforcement, each anchored to evidence from the observation and the teacher's growth trajectory. A 'write my own' card sits underneath. The interaction is built around 'pick the one that fits,' not 'accept the AI's answer.'

School leaders do this work on their feet. The phone-native flow is designed for a summer 2026 release: quick-tap scoring, voice-to-text evidence notes, and the AI-drafted feedback step inline. One-handed thumb reach across three screens, with progress and timer always visible.