A chatbot that doesn't just answer questions — it sells sneakers at 2am with the same personality the brand projects everywhere else.
Comet is a Gen-Z focused D2C sneaker brand with a vibrant personality, a highly engaged social media community, and a product drop culture that drives intense, time-sensitive customer demand. The brand had built real equity with its audience — but its customer support infrastructure was designed for a different scale.
As traffic grew — especially around product drops — the gap between what customers expected and what a manual support team could deliver became impossible to close. Sizing questions went unanswered. Style queries arrived at midnight. Cart abandonment climbed during the exact moments when responsive, on-brand support would have converted the sale. An AI sales assistant was the answer — but it had to sound like Comet, not like a chatbot.
Three people were let down by a support model that couldn't keep up with Comet's community — a shopper who wanted instant answers, a support lead drowning in volume, and a growth lead watching support costs eat into margins.
Gen-Z sneaker enthusiasts encountering a new drop on Instagram at midnight had no way to get immediate sizing answers — and by the time a response arrived the following day, the purchase had already gone to a competitor.
🌙 Off-hours sizing queries converting to competitor salesSupport leads during product drops faced hundreds of simultaneous messages covering the same recurring questions — order tracking, sizing, and style comparisons — consuming the entire team's capacity while still leaving queries unresolved.
📨 Repetitive ticket volume blocking strategic support workGrowth leads watched support costs scale linearly with traffic while cart abandonment peaked during the highest-demand drops — spending more on support while still losing sales to unanswered pre-purchase questions.
💸 Support costs scaling without conversion improvementGen-Z customers don't wait. Zara's decision to buy a limited drop happens in a window measured in minutes, not hours — and the questions standing between her intent and her purchase are almost always the same ones: sizing, styling, availability, delivery. A manual support team operating within business hours couldn't cover those windows, and no generic chatbot could replicate the brand personality that made Comet worth buying from in the first place.
The support operation was a ceiling on growth. Nikhil's team was handling escalating volumes with no relief mechanism, and the repetitive nature of the tickets — the same sizing charts, the same tracking links, the same return policy questions — was consuming bandwidth that should have been focused on complex customer relationships. The AI needed to be both a sales tool and an operational relief valve, simultaneously.
“Comet's community expects instant, on-brand responses at the exact moments support teams are offline and tickets are at their highest.”
The AI was designed to give Zara an instant answer at midnight, give Nikhil his time back, and give Dev a support model that doesn't cost more as it scales.
Deployed a RAG-based AI assistant trained on Comet's brand voice, product catalog, and Gen-Z communication style — responding in the same authentic, playful personality that built the brand's community.
Human-CentricityImplemented a recommendation engine analyzing preferences, browsing behaviour, and style intent to suggest specific sneaker models, colourways, and sizes — helping Zara find what she came for before she goes elsewhere.
Human-CentricityIntegrated with the order management system to handle tracking, size exchanges, and return assistance without human intervention — removing Nikhil's most frequent ticket categories from his queue entirely.
SustainabilityThe AI operates continuously — drops at midnight, questions during weekends, queries during high-traffic events — without shift changes, breaks, or service degradation. Zara gets an answer when she needs it.
ResilienceBuilt intelligent escalation logic that detects when human expertise is needed — transferring to Nikhil's team with full conversation context so the customer doesn't have to repeat themselves.
ResilienceImplemented feedback mechanisms so successful interactions improve future recommendations — the AI gets better at selling Comet's products with every conversation it completes.
SustainabilityZara asks about sizing at midnight during a drop and gets an instant, accurate, on-brand answer. The window between intent and purchase closes before her attention moves elsewhere.
Nikhil's team no longer handles the same 20 questions on repeat. Routine queries are captured by the AI — the support team focuses on the situations that actually need human judgement.
Dev's support costs dropped while conversion improved. The AI replaced a linear cost model with a fixed-infrastructure model — support now scales with traffic without proportional cost increases.
Every AI interaction sounds like Comet — the same tone, the same energy, the same personality the brand projects on social media. The chatbot is a brand asset, not a utility.
RAG-powered conversational AI trained on your brand voice — delivering instant, personalised sales assistance 24/7 while dramatically reducing support costs.