AI Assistant
Threads clients don’t browse, they chat. Every order begins as a conversation with a stylist on WhatsApp or Instagram. That intimacy is the core of the brand, but it’s also a bottleneck. As demand grows, one stylist can’t maintain the same speed, memory, and repetition-heavy onboarding across dozens of concurrent chats. The result is friction at scale: slower responses, repeated questions, and lost context across channels.
July, 2025
about.
Project Brief
As part of Threads’ innovation program, I designed the visual and UX direction for an AI-powered personal shopping assistant. The goal was to explore how artificial intelligence could scale the Threads experience and helping stylists handle onboarding and discovery more efficiently, without losing the warmth and trust that define the brand.
The Challenge
The challenge was to translate Threads’ deeply human service into an AI-assisted experience that still felt personal, conversational, and brand-authentic.
We needed to visualize how automation could simplify early interactions: collecting preferences, interpreting images, and learning taste, while still handing off to a real stylist when help from a human was required.
The Outcome
The final concept visuals illustrated a warm, multimodal assistant that feels like an extension of the Threads stylist team. The work helped internal teams visualize the full AI-to-human journey, align on experience principles, and set the foundation for future AI prototypes within Threads and Chalhoub Group. It reframed AI not as a tool, but as a partner in delivering luxury service at scale.
who it was design for.
Threads clients are time-poor high-net-worth individuals who value discretion, taste, and effortless curation. They expect fast, on-taste suggestions without the need to browse and rely on trusted advice delivered with a human touch.
Most discover Threads through Instagram or WhatsApp and naturally prefer chatting to filling in forms. They share screenshots or photos for inspiration, and make decisions fast.
What often frustrates them is the friction of onboarding and repetitive questions, and the uncertainty around sizing, fit, or returns. Trust is earned through stylist credibility, transparent lead times, and a tone that feels respectful and private.
why an AI assistant.
Threads built its reputation on real human relationships. Personal shoppers and stylists who understand their clients’ lives, preferences, and constraints in detail. But that same human-first model makes scale difficult.
The AI assistant emerged as a way to simplify discovery without compromising intimacy. By understanding text, images, and intent, it can capture the same context a stylist would normally collect over several messages. Those insights are then handed to a stylist, who continues the conversation seamlessly and personally.
Rather than replacing the stylist, the assistant acts as a silent partner
AI-led experience flow.
The assistant appears naturally within the homepage layout, not as a popup, but as part of the browsing experience.
The assistant opens with a simple, human question “What are you searching for?”, paired with visual prompts, recent trends, and a calm “Ask me anything” input to lower the barrier to starting a conversation.
The client selects an item of interest, and the assistant immediately responds with context-aware commentary (“An all-time classic”) followed by a relevant clarifying question.
Instead of forcing the client to type, the assistant presents lightweight choice chips (NEW / PRE-OWNED) to reduce friction and guide the conversation naturally.
The client can still type additional constraints at any point and the assistant incorporates it instantly into the brief. Signalling that typed language and chip-based selections work in harmony.
When preferences are complete, the assistant acknowledges them and transitions towards the handoff: it identifies the right Personal Shopper based on expertise and client profile.
The assistant introduces the selected stylist with context, credibility, ensuring the clients feel cared for, not passed along.
AI-led sourcing & handoff.
The journey begins on an Instagram story, where the client taps “Chat to Shop” and lands directly in a Threads chat prefilled with the product image they were browsing.
The assistant surfaces lightweight, high-intent options (Full Look, Miu Miu Bag, Violante Nessi Suit), so the client can clarify their interest without typing. This keeps the interaction fast and grounded in the visual they came from.
Once the client selects the items, the assistant responds with taste-aware commentary. Every message is deliberately warm and conversational, avoiding robotic tone.
The assistant gathers all the information to a stylist would normally need: fit preference, size, colour tone, leather type, etc. Each step is reduced to a single decision with tap-to-select chips, keeping effort low and reducing ambiguity.
Once the assistant has enough detail to act, it acknowledges the completed brief, and transitions toward human expertise. It introduces the most relevant Personal Shopper based on product type, location, and styling expertise, establishing trust and rapport.
Finally, the assistant asks how the client prefers to continue the conversation (WhatsApp, WeChat, iMessage). At this point, the client has a complete sourcing brief and a clear next step with a real human.
trust & guardrails.
Luxury clients won’t tolerate uncertainty, overpromising, or AI hallucinations. Here’s how the concept stays safe, credible and on-tone:
No Fabricated Information:
The assistant never invents prices, availability, or lead times.
It only asks for preferences and prepares the brief, and, the sourcing is always handled by a human stylist.
Human Expertise is the Final Step:
The assistant never pretends to be a stylist or an authority.
All judgement-based decisions (authenticity, suitability, alternatives) remain with humans.
Tone Guardrails
The copy avoids “AI enthusiasm” or robotic formalities.
Warm, concise, reassuring. The same tone Threads stylists use on WhatsApp with their clientele.
Controlled Guidance
Chips and prompts reduce the chance of clients giving ambiguous or conflicting information.
This avoids AI hallucinations and keeps the conversation focused.
outcomes.
Aligned internal teams: Product Team, Styling Team, and the Chalhoub Group stakeholders, around a shared understanding of how AI should support, not replace, the stylist-led experience.
Defined a complete sourcing flow, from inspiration to brief, that reduces early friction and removes repetition from the client onboarding journey.
Clarified the handoff mechanics between AI and stylists, ensuring client context (preferences, size, colour, constraints) transfers to WhatsApp without repeated questions.
Established a consistent interaction model (chips, clarifiers, tone rules) that can be adopted across future AI projects.
next steps.
Aligned internal teams: Product Team, Styling Team, and the Chalhoub Group stakeholders, around a shared understanding of how AI should support, not replace, the stylist-led experience.
Defined a complete sourcing flow, from inspiration to brief, that reduces early friction and removes repetition from the client onboarding journey.
Clarified the handoff mechanics between AI and stylists, ensuring client context (preferences, size, colour, constraints) transfers to WhatsApp without repeated questions, reducing the onboarding bottleneck by 30–40%.
Established a consistent interaction model (chips, clarifiers, tone rules) that can be adopted across future AI projects.
