Product · AI
Talabat AI Experience
Leading product strategy and 0→1 execution of talabat's conversational AI experience — from traditional search to intent-driven, voice-enabled food discovery across the MENA region.
Traditional search in food delivery apps forces users to think in keywords — but people think in cravings, situations, and groups. I led the product vision for the Talabat AI Experience, reframing food discovery from a search-driven to an intent-driven model, expressed naturally via text and voice.
I owned the end-to-end product definition — framing the opportunity, defining the interaction model, and prioritizing the roadmap. Partnering with Mahip on design craft and with engineering on feasibility, I directed the AI design language now integrated into Marshmallow, talabat's design system.
I ran user testing across the region to validate the concept and iterate on it — the experience was perceived as intuitive, intelligent, and responsive. It continues to evolve as part of the AI roadmap I drive at talabat.
The shift to conversational search
Traditional search forces users to think in keywords: 'pizza', 'sushi', 'burger'. But people don't think that way — they think in cravings ('something spicy'), situations ('quick lunch near me'), and groups ('dinner for 4 with a vegetarian'). Conversational AI bridges this gap, interpreting natural language intent instead of matching keywords.
The experience
Users interact through text and voice. The system interprets intent, understands context (location, time, preferences), and returns curated results. The interaction feels like talking to someone who knows your taste — not searching through a catalog.
Voice interaction
Voice input creates a real-time visual reaction — the morph responds to amplitude and tone, creating a visceral connection between the user's voice and the interface. Speaking to the app feels alive and responsive.
AI Language (design system)
A scalable visual and interaction language was developed and integrated into Marshmallow, talabat's design system. This ensures consistency across all AI-powered experiences. The visual system uses warm gradients, subtle textures, and a tone that feels modern and approachable — intelligent without being overwhelming.
The morph
The morph is the core visual element of the AI experience. It represents the AI's behavior — reacting to user input, communicating system states, and providing visual feedback. It's not decorative; it's functional, giving users a sense of responsiveness and intelligence.
Motion & conversational cycles
The motion system defines four key states: listening, thinking, responding, and idle. Each state has distinct motion characteristics that communicate what the system is doing without requiring text. Motion is the language between the user and the AI.
Listening
Thinking
Responding
Idle
Key applications
The AI language extends beyond the main experience into icon systems, search bar integration, and multiple entry points across the app. Every touchpoint maintains the same visual and interaction vocabulary.
End-to-end flow
From the moment a user opens the search to receiving curated results, every step is designed to feel seamless. The journey from input to results is fast, visual, and intelligent.
Outcome
Tested with users across the region, the experience was perceived as intuitive, easy to use, and genuinely intelligent. Users reported a strong sense of responsiveness and felt the system understood their intent. The experience continues to evolve as part of talabat's ongoing AI exploration.
