Microsoft: Reimagining AI for Specialized Educators
An AI-powered assistant built for special-education teachers.
Overview
Bridgit came out of a 16-week engagement at Microsoft's External Engagement Studio at Parsons — a four-person team (Alana, Azelia, Arpit, and me) working a single question: how might trustworthy AI support the specialized educators who hold underserved classrooms together, without eroding their judgment or trust?
I worked as UX researcher and interaction designer. I ran semi-structured interviews and affinity mapping to turn raw educator voices into the four opportunity areas, then designed the dashboard, session personalization, and communication flows that became the Bridgit prototype.
Our co-design process — turning educator sessions into the four opportunity areas and the Bridgit prototype.
The Teacher Workload Problem
Elementary teachers spend 15 hours a week outside class on planning — 12 of them unpaid. Turnover sits above 20%, and 40% of new teachers leave within five years. AI was arriving with no clear answer for how it should help.
We answered our question through three moves: research to map the strain, co-design to keep specialists in the loop, and a prototype to make the solution concrete.
How might we support elementary teachers with an AI tool that reduces burden and enhances collaboration — without compromising their agency or trust?
Our guiding research question
Led With Research, Not Technology
Rather than assume a technology solution, we started with the human rhythms of teaching. Seven interviews with parents, teachers, administrators, and specialists — plus three rounds of co-design and a literature review — built a systems-level picture before we designed anything.
The research pointed to four areas where teachers consistently lacked support: Productivity, Communication, Personalization, and Student Achievement.
Research Protocol & Methodology
Alongside a literature review and stakeholder interviews, we ran three rounds of co-design workshops. Each session set ground rules, opened up ideation, and had specialists co-create and prioritize directions with us — so the solution was built with teachers, not just for them.
Co-creation guidelines opened each session with a no-wrong-answers, think-out-loud space.
Ideation translated pain points into 'How might an assistant help?' need statements.
Specialists co-created against the core needs: productivity, personalization, play, and parent coordination.
Prioritization grouped and ranked ideas by real day-to-day impact.
Four Opportunity Areas
Across every interview, the gaps clustered into four areas where teachers consistently lacked support.
Productivity
Teachers need a way to handle admin and planning tasks more efficiently.
Communication
Teachers need a way to better communicate and collaborate.
Personalization
Teachers need a way to easily create personalized learning plans.
Student Achievement
Teachers need support so they can use their skills to address diverse learning needs and provide an equitable learning experience.
Where Existing Tools Fall Short
Three products already serve general classrooms well: Microsoft Copilot for Education (lesson planning, assessment), Magic School (80+ content tools), and Brisk (a secure Chrome AI extension).
None were built for specialized educators — ESL instructors, speech pathologists, behavioral therapists — who work across classrooms, write individualized plans, and juggle disruption-prone schedules. That gap was the opportunity.
Introducing Bridgit
Bridgit is named for its purpose: a bridge between primary teachers, specialists, parents, and students — the four groups whose coordination currently runs on text messages and hope.
The dashboard shows the specialist's day — sessions plus admin and parent meetings, tasks sorted by urgency. The AI assistant drafts emails, summarizes student profiles, and generates post-session reports.
The dashboard: personalized schedule, urgency-sorted tasks on the right, and the AI assistant ready to draft communications or pre-session summaries.
Session Personalization
Open any session to see its objectives, lesson plans, exercises, and the primary teacher's notes. Pick a task, and the AI reads the student's anonymized profile and returns specific tactics.
For a visual learner engaged only by her own interests, it suggested using her drawings as conversation starters and a visual agenda for structure — then synced the changes back to the task page automatically.
Drilling into a session surfaces objectives, lesson plans, exercises, and primary teacher notes — then the AI reads anonymized student data to suggest personalization tactics.
The AI reads the student's anonymized profile and returns specific tactics — not generic advice.
Notes arrive as a photo, audio recording, or document. Bridgit anonymizes and formats them for sharing automatically.
The AI drafts the note and routes it to Messages — Olivia's multi-channel problem addressed in one action.
Calendar view shows the specialist's own day plus other teachers' schedules and the school calendar — meetings can be set without any back-and-forth.
Adapting to Disruptions
When a student is absent, Bridgit flags the schedule impact and the missed progress, then auto-generates a makeup slot the specialist can adjust. The AI can fill the freed time with tasks, professional development, or a rest period.
A responsive mobile version keeps the same assistant on hand between classrooms.
An absent student triggers a notification, a schedule impact view, and an auto-generated makeup — the specialist adjusts or accepts.
Ask Bridgit to optimize the day and it slots tasks by urgency and time required — no manual juggling.
Research Reframed the Problem
Starting with human needs instead of technology capabilities changed what we built. The burden of AI adoption lands on teachers — so the opportunity wasn't a smarter classroom tool, it was infrastructure for the specialists holding the system together. That reframe drove every product decision in Bridgit.
Documenting our assumptions before fieldwork kept us honest when findings contradicted them. Next: test personalization across specialist types, keep co-designing with teachers and administrators, and pitch Bridgit to school leaders.