AMICUS

Duration

3 Months

Project Type

Generative AI

Web Design

My Role

UX Lead

Front End Dev

Target Users

Internal Staff

300+ Teachers

Contribution

User Reasearch

Flow Design

High Fidelity Design

Front End Dev

Outcome

Reduced Staff Leavers

Increase in staff support satisfaction

THE CHALLENGE

Inspired Group needed to address critical teacher retention—annual turnover reached 65 staff members, largely due to burnout from excessive administrative workload. Lesson preparation (creating PowerPoints, handouts, curriculum-aligned materials) consumed hours that should have been spent on student interaction.

The opportunity was to leverage generative AI to reduce this burden, but the challenge was multifaceted: teachers were skeptical of AI, worried about accuracy and reliability, and concerned that automation might undermine their professional expertise.

My role: Lead UX strategy and execution from discovery through launch, balancing teacher needs, technical constraints, and organizational goals while introducing AI into an environment that had never used it before.

UNDERSTANDING TEACHER NEEDS

Working with the teaching team, I conducted discovery to understand where time was being wasted and what would actually help:

What we learned from interviews and team discussions:

Teachers spent 2-3 hours per day on lesson prep, with much of it on formatting and finding resources
The biggest pain point wasn't creating content—it was tailoring existing resources to specific student needs and contexts
Teachers were concerned AI would produce "generic" or inaccurate content that didn't reflect our curriculum standards
There was significant anxiety about being replaced by automation
Teachers wanted to stay in control—they needed to review, edit, and adapt any AI-generated content

The core insight: Teachers didn't need AI to replace their expertise—they needed it to handle the repetitive, time-consuming work so they could focus on the pedagogical decisions only they could make. The tool needed to feel like an assistant, not an autonomous system.

KEY DESIGN DECISIONS

Stripping Back Complexity

Early wireframes included granular controls—temperature settings, token limits, model selection. This was based on assumptions about what "power users" might want.

What we learned: When we tested these interfaces with teachers, they were overwhelmed. Most had never used AI tools before and didn't know what "temperature" or "tokens" meant. The complexity was creating a barrier to adoption.

The pivot: We stripped it back to a simple three-step flow: Setup (who you are), Prompt (what you need), Creativity (how novel vs. standard). This meant losing some control, but it dramatically improved accessibility.

Renaming "Temperature" to "Creativity"

This seems small, but it was strategic. "Temperature" is AI jargon. "Creativity" is a concept teachers understand and use daily.

Why this mattered: It signaled that Amicus was built for teachers, not AI enthusiasts. It reduced cognitive load and made the tool feel familiar rather than technical.

Grounding in Internal Resources

One of the biggest concerns was accuracy—would Amicus produce content that contradicted our curriculum or used unreliable sources?

The solution: We worked with engineering to ensure all outputs were grounded in our internal educational resources and curriculum standards. This wasn't just a technical choice—it was a trust-building strategy.

The impact: Teachers felt confident that Amicus-generated content would align with what they were already teaching, reducing the review burden and increasing adoption.

Building "Amicus Together"

Post-launch feedback revealed that some teachers were getting much better results than others. Rather than just improving the tool, we saw an opportunity to leverage this.

The feature: Amicus Together allowed teachers to share effective prompts, setups, and generated materials with colleagues. This turned individual efficiency gains into collective learning.

Why this worked: It created a cultural shift—teachers who were initially skeptical saw their peers succeeding and learned from them. It also reduced the support burden on our team.

‹

›

THE SOLUTION

Simple Three-Step Flow

Step 1: Setup
Define your context (Year 11 English teacher, GCSE curriculum)

Step 2: Prompt
Describe what you need (Lesson plan on Macbeth Act 1, PowerPoint on photosynthesis)

Step 3: Creativity
Adjust output style (Standard curriculum-aligned vs. Creative exploratory)

Additional features:

Image upload capability (reference existing materials)
Amicus Together (share effective prompts and materials)
In-product feedback (continuous improvement)
Training materials and always-available support

Platform Considerations

Designed for both desktop (primary lesson planning context) and mobile (quick reviews and on-the-go use). Teachers could start planning on desktop and review materials on mobile during commutes.

IMPLEMENTATION & ITERATION

Working with engineering: This was Inspired's first generative AI tool. We navigated API rate limits, response time optimization, and content safety filtering together. I worked closely with the development team to specify edge cases and ensure the experience matched design intent.

Managing stakeholder expectations: Leadership wanted fast adoption. Teachers wanted reliability. Product wanted feature velocity. I facilitated alignment by establishing clear success metrics and running a phased rollout starting with early adopters.

Post-launch optimization: We continued iterating based on usage analytics, support tickets, and direct teacher feedback. Key improvements included image upload capability, improved prompt suggestions, and mobile optimization.

‹

›

Results & Impact

Business Impact:

Annual teacher turnover dropped from 65 → 25 (62% reduction)
Staff satisfaction with "technology and processes" increased 17% year-over-year
Steady adoption across multiple Inspired campuses
Teachers reported saving 2-3 hours per week on lesson prep

User Experience:

High engagement with Amicus Together (teachers actively sharing and collaborating)
Low support burden (simple interface reduced confusion)
Teachers reported saving 45 minutes per day on lesson preparation

What I learned

Simplicity over control: The decision to prioritize simplicity over advanced settings was validated—95% of teachers never requested more granular controls. Starting with the simplest possible interface and adding complexity only when needed is often the right approach.

Trust through transparency: Grounding content in internal resources built trust faster than any amount of UI polish could have. When teachers knew outputs were based on our curriculum, they adopted the tool with confidence.

Community accelerates adoption: Amicus Together (the collaborative feature) accelerated adoption more than we anticipated. We should have built it from day one rather than as a post-launch addition. Peer learning is more powerful than top-down training.

Cultural change matters: Beyond metrics, Amicus changed how teachers viewed AI—from a threat to a tool. This cultural shift opened the door for other AI initiatives within Inspired and demonstrated that thoughtful UX can bridge the gap between technological capability and human acceptance.