MIRA

Voice AICustomer ExperienceAI71

Enhancing customer experience through Voice AI

MIRA is an enterprise Voice-AI platform that answers, understands, and resolves customer calls and chats — deployed across UAE government, healthcare, and corporate support. I led the redesign that made it feel human, fast, and worth trusting.

Role: Lead Product Designer
Company: AI71
Timeline: 2024–25
Scope: Research · UX · Visual · Design QA
Sectors: Government · Healthcare · Enterprise

The problem

MIRA’s AI-driven approach was genuinely innovative — but early user feedback surfaced five gaps that were holding the experience back.

No context awareness

Responses were static. They didn’t adapt to who was calling or why, so every conversation felt impersonal.

Emotional disconnect

People calling support wanted reassurance; corporate clients wanted speed and precision. One flat tone served neither well.

Integration bottlenecks

Pulling real-time information from legacy systems was slow, and those delays showed up in the middle of live calls.

Low trust & adoption

Confidence in AI-led support was low — especially for high-stakes, emotionally charged conversations.

A “faceless” product

Voice AI is invisible by nature. Users never see a polished screen, so trust has to be earned through behaviour, not UI.

Process

Design approach

I ran a double-diamond process — diverging to explore the problem and the solution space, then converging on what to actually build and ship.

Discover

Problem exploration
Foundational research
User interviews
Competitor analysis

Define

Synthesising research
Product requirements
User stories

Develop

Concept sketches
Wireframes
Prototyping
User testing
Design reviews
Recommendations & rationale

Deliver

High-fidelity designs
Design reviews
Handoff
Design QA

Competitive analysis

MIRA plays in a crowded field of global and regional players. To position it well, I audited more than 100 companies — here are the leaders worth knowing.

Boost.ai

Enterprise conversational AI built for scale in banking, telecom, and government. Strong multilingual virtual agents and omnichannel support.

G42

A UAE AI leader with deep government partnerships — AI infrastructure, NLP, and speech analytics.

Emaratech

Focused on government digital transformation: secure transaction automation and AI-driven workflows.

Open CX

An AI-powered omnichannel support platform with a strong automation focus, but a limited government footprint.

Intella

A regional leader in Arabic NLP and sentiment analysis, specialising in call-centre analytics.

A slice of the 100+ company audit across the global and MENA Voice-AI landscape.

Where MIRA wins

Three things set MIRA apart from everyone in that audit.

Adaptive conversation

Where competitors run static flows, MIRA adjusts its tone and pacing in real time, based on live sentiment.

Deep integration

It slots into hospital and corporate workflows, meeting industry-specific needs instead of forcing a generic one.

Localised NLP

Superior Arabic speech recognition and sentiment analysis give it a real edge across MENA markets.

User interviews

20 in-depth interviews across healthcare and corporate support — including leaders at Talabat, Cisco, Intercom, and the Head of Customer Support at Yango.

User Research

What the interviews revealed

Two audiences, two very different sets of needs — and a clear mandate for how the AI should behave.

Healthcare professionals

Patient-facing, emotionally sensitive calls.

85%said the AI needed a calm, reassuring tone with patients.
70%reported patients felt discomfort with robotic responses.
60%found the lack of personalised responses frustrating.

Corporate agents

High-volume, efficiency-driven support.

90%preferred clear, concise responses with less back-and-forth.
75%cited slow system integration as a key barrier.
65%needed quick resolution without AI misinterpretations.

On-site observation

At Reem Hospital

Observed 15 emergency calls — 60% needed an agent override when the AI missed the urgency.
Medical staff wanted better AI workflow suggestions for confirming and rescheduling appointments.

At Axiom

Call misrouting from AI misunderstanding pushed average call times up by 20%.
Agents often fell back to manual handling, which chipped away at trust in the AI.

The impact

Targeted design changes moved the numbers that mattered — for resolution speed, efficiency, trust, and adoption.

Faster resolution

Minimum call-resolution time dropped to 10 seconds.

Operational efficiency

Call-handling times fell by 31.5%.

Higher satisfaction

Support efficiency rose 5% at Reem Hospital and 43% at Axiom.

Faster onboarding

Adding a new organisation went from 42 minutes to 1.5 — a structured, no-code flow replaced manual setup.

More AI adoption

Confidence in AI-led interactions lifted enterprise usage by 20%.

Better accuracy

AI misinterpretation dropped by 45%, building trust and reliability.

Less manual handling

Human intervention fell by 35%, freeing up agent time.

User flow & information architecture

Managing customer interaction — the core call-handling logic. (Flow diagrams are blurred for NDA.)

Appointment data validation — checking appointment details end to end.

File upload & validation, and MIRA’s entire modular flow — every module, connected.

The product

The platform spans sixteen-plus screens — analytics, calls, chats, appointments, the knowledge base, and organisation management. Here they are in full.