Ishika Soni
Resume

Accessibility • Inclusive Design

Deaf Customer Support Experience

During Microsoft’s Global Hackathon, I designed and prototyped the first ASL-integrated support workflows inside Dynamics 365, enabling Deaf support engineers to communicate, document, and collaborate without breaking eye contact or context. The solution introduced Copilot-assisted ASL-to-text, a unified communication tool inside the CRM — eliminating the need for multiple devices and reducing cognitive overload. The project won the #Hack4Inclusion award and is being considered for production.

Overview

Role

UX Designer

Timeline

2 Weeks

Platform

Dynamics 365 (CRM)

Focus

Accessibility, ASL-first workflows, Copilot integration

1. The User

Primary User:

👉 Deaf Support Engineers who rely on American Sign Language (ASL) as their primary mode of communication.
User context visual

Secondary Users:

  • Deaf customers receiving enterprise support.
  • Hearing teammates collaborating with Deaf engineers.

2. The Real Problem

Deaf support engineers face challenges that traditional enterprise tools don't acccount for:
  • ASL requires continuous visual attention
  • Most CRM workflows assume users can:
    • Listen while typing
    • Read while talking
    • Switch contexts seamlessly
User context visual
In reality, Deaf engineers are forced to:
  • Switch between 3-4 devices (videophone, computer, transcription tool)
  • Break eye contact to document cases
  • Put customers on hold to research issues
  • Reconstruct ASL conversations into text after calls
This leads to:
  • Longer handle times (30-35+ minutes per case)
  • Loss of nuance in documentation
  • Cognitive overload and communication fatigue

3. Why Existing Solutions Failed

From interviews, journey mapping, and standards review:
  • Live transcription is delayed and unreliable
  • CRM tools are text-first, not visual-first
  • No system supports ASL + documentation + collaboration in one place
  • Engineers must choose between:
    • Engaging with the customer
    • Doing their job tasks
This creates a Divided Attention Crisis.
User context visual

4. The Breakthrough Idea

What if ASL wasn’t an accommodation — but a first-class input modality inside the CRM?
User context visual
Instead of bolting on captions, a system was designed where:
  • ASL video is embedded directly into workflows
  • Copilot assists with real-time translation and rewriting
  • Documentation happens without breaking eye contact
User context visual

5. What I Designed

1. Copilot-Assisted ASL -> Text

  • ASL transcription after recording
  • Copilot refines and rewrites transcripts into professional case notes
  • Engineers can edit without re-recording

2. ASL-Enabled Email Drafting

  • Engineers draft emails using ASL
  • Copilot converts signing into polished written responses
  • Reduces after-call documentation time

3. Unified, In-CRM Experience

  • No external videophones
  • No device switching
  • No context loss

Live prototype of using ASL to write case notes.

User context visual
User context visual
User context visual
User context visual
User context visual
All designs were prototyped in Figma and validated against Dynamics 365 feasibility constraints.

6. My Role & Decisions

I was the UX Designer responsible for end-to-end experience and systems design, including:
  • Conducted interviews with Deaf participants and support engineers (US & Ireland)
  • Synthesized research into core problem themes
  • Defined ASL-first interaction principles
  • Designed end-to-end CRM workflows
  • Prototyped low- to high-fidelity solutions in Figma
  • Collaborated with engineers to validate feasibility
  • Iterated through usability testing with Deaf and hearing users

7. Impact & Outcomes

User context visual

🚀 Reduction of overall call time by 30+ mins.

User context visual

🚀 ASL recordings being sent to Copilot for transcription also further fine-tune the SLRT model.

  • 🏆 Won Hack4Inclusion Award sponsored by the Disability ERG
  • 🚀 Identified as a candidate for production development
  • 🌍 Raised internal awareness of Deaf engineers' invisible labor
  • 🧠 Demonstrated how Copilot can support non-text primary users

8. What I’d Do Next

With more time, I would:
  • Validate transcription accuracy across ASL dialects
  • Define trust & confidence indicators for Copilot translations
  • Measure real handle-time reduction in pilot teams
  • Explore privacy controls for video-based documentation
User context visual