Meet Rai by RAKBANK

Building the persona, name, and tone of voice for a bilingual AI banking assistant in the UAE - a product persona. 
RAKBANK was designing an AI assistant that started first as a universal menu which later turned into an AI conversational bot. At the time, along with the team we worked on understanding user intent - creating the appropriate scenarios and use cases and creating prompts to ensure the digital assistant hallucinates less. After launching the team wanted us to create a persona and feel for this digital assistant but since it's entirely a customer experience feature the Research & Design team was the one to spearhead it.  
Our role
Persona Design · Naming Strategy · Tone of Voice System · Bilingual Content Strategy · Cultural Research

Our approach

We ran the project in four phases. Each one fed the next. Here's the Figma link for the full project. 

1. Context and risk mapping
Before any creative work, we mapped the strategic and operational risks — legal, brand, ethical, cost, and gender bias considerations. We aligned on team, dates, development, and roll-out. The point was simple: AI persona work fails when teams skip the risk conversation, then have to undo decisions later.

2. Research designed for cultural depth
We built a research framework around five emotional and cultural dimensions: personality traits, gender perception, naming, tone and language, and cultural fit. Every question was paired with a clear extraction goal — what we'd learn from each answer. The research wasn't just about preferences. It was about uncovering subconscious cultural cues that shape how UAE users perceive trust, warmth, and authority in a banking context.

3. Benchmarking the landscape
We audited local and international assistants across banking, telecom, services, and government. The big finding: local-first wins. Government platforms like TAMM and Rashid showed how cultural resonance, naming strategy, and bilingual execution build trust. International assistants like ChatGPT, Gemini, and Meta AI left their names in English on Arabic pages and defaulted to masculine grammar — making them feel foreign and impersonal.

The benchmark validated four strategic pillars: cultural resonance, tone adaptability, bilingual consistency as a USP, and deliberate gender clarity in Arabic.

4. Persona framework - anchored in tonal spectrum
Instead of jumping to one fixed persona, we built a framework with two anchor archetypes — *The warm guide* and *The efficient partner* — at opposite ends of a tonal spectrum. We tested each archetype across four interaction states: greeting, apology, escalation, and confirmation/success. Then we let the research findings tell us where on the spectrum Rai should live.

Results:

60% of RAKBANK’s user base is now completing their daily banking tasks via Rai.

Previous
Previous

Jeem by Extensya

Next
Next

The Portal