Create Your First Project

Start adding your projects to your portfolio. Click on "Manage Projects" to get started

.pro-gallery-wix-wrapper {display: block !important;} .pro-gallery-wix-wrapper .gallery-item-container {opacity: 1 !important; display: block !important;}

Project

1. Role: UX Design Consultant (Conversational UX and Service Design)
2. Industry: Banking / Digital Financial Services
3. Product: Mashreq Neo conversational banking capabilities for core self-service tasks

Context

Mashreq Neo is a digital-first banking offering designed around modern, self-service experiences. As the product evolved, conversational interfaces were introduced to reduce reliance on traditional support channels and improve task completion, without sacrificing trust, clarity, or compliance.

I worked as a UX design consultant focused on conversational UX and service design, shaping how chat-based interactions could support core banking tasks safely and predictably, and how the service layer should behave across success, failure, and escalation scenarios.

The challenge

The problem wasn’t “launch a chatbot.” It was designing a reliable service interaction layer for banking tasks where:

• Users arrive during moments of urgency or confusion,
• Tolerance for ambiguity is extremely low,
• Regulated language and compliance constraints shape what the system can say and do,
• A conversational interface can quickly feel gimmicky or untrustworthy if it lacks control and transparency.

The risk was introducing automation that increased friction or reduced confidence, driving users back to human support, or worse, creating avoidable errors.

What we learned that shaped the solution

Users were willing to use chat for banking tasks only when the interaction felt predictable and controllable.

This shifted the design away from “natural language novelty” toward intent-led, guided self-service, clear steps, visible system status, explicit confirmations, and fast exits.

Approach (service design first, conversation second)
Work treated conversational UX as an extension of service design:

1. Analyze customer support drivers and failure points
2. Identify high-frequency, high-friction banking tasks suited to chat
3. Define an intent taxonomy aligned to user language (not internal bank structures)
4. Design safe conversational patterns for regulated tasks
5. Validate flows with usability testing focused on trust, recovery, and completion.

The goal was not to simulate human conversation, it was to deliver reliable task completion with the least possible ambiguity.

Design outcomes (what we designed, concretely)

1. Intent-based conversation architecture

• Built an intent model that reflects how users express goals.
• Designed guided steps per intent so users always understand “where they are” and “what happens next.”

2. Trust-by-design interaction patterns

• Explicit confirmations for sensitive actions and clear review moments before finalizing.
• Visible status signals (processing / completed / failed / needs info) to eliminate uncertainty.
• Receipts and summaries (“what was done,” “what’s next”) to reinforce reliability.

3. Error handling as a first-class flow

• Designed fallback behavior for ambiguity, missing inputs, and unsupported requests.
• Created recovery loops that keep users in control: edit, back, cancel, restart, without dead ends.

4. Safe escalation to human support

• Defined escalation triggers (repeated failure, sensitive topics, frustration signals).
• Designed handover UX so context transfers cleanly and users don’t need to repeat themselves.

Fidelity was used strategically, low-fidelity intent trees and conversation maps validated service logic early, while high-fidelity chat UI states and message patterns were designed for the moments where microcopy, hierarchy, and system feedback directly affect trust, confirmations, errors, receipts, and escalation.

Deliverables

• Intent taxonomy and routing model
• Conversation flows (happy paths and recovery paths)
• Service blueprints linking chat interactions to backend operations
• Error, fallback, and escalation frameworks
• Conversational UX guidelines for clarity, tone, and regulated language constraints
• Alignment artifacts for product, engineering, compliance, and support teams

Impact

• Reduced friction for common banking tasks by converting frequent support needs into guided self-service journeys.
• Lowered reliance on human-assisted support for routine inquiries by designing predictable completion paths and clear recovery.
• Improved task completion by making conversational interactions structured, controllable, and transparent.
• Established a scalable model for conversational banking (intents, patterns and escalation rules) that could expand to additional tasks over time.

Reflection

Conversational UX in banking succeeds when it respects user intent and limits. Users don’t want to chat with their bank, they want to complete a task confidently and move on.
Designing for predictability, transparency, and graceful failure proved far more effective than pursuing conversational realism.

NDA note

Visuals and examples are simplified or reconstructed to preserve confidentiality while reflecting real service logic, interaction patterns, and outcomes.