Your storefront answers questions now.
An agentic shopping assistant built into Idukki. Shoppers ask “what should I wear for hot yoga?”, your storefront returns a curated look blending UGC + your catalog. Text, voice, or image input. Brand-safe. In-stock-aware. Live in 30 minutes.
- 28 locales
- In-stock-aware
- Voice + image + text
- What should I wear for hot yoga?
- Hot yoga = breathable, sweat-wicking, four-way stretch. Here’s a complete look from your favourites:
in stock
in stock
in stock
- Cross-Back Sports Bra · Slate$48
- High-Rise Align Short · 8"$68
- Sweat-Wick Crew · Eucalyptus$54
Inspired by @maya.k- Anything cheaper?
- Sure, under $120 total:Same fit, value picks from your ‘Essentials’ line · 3 items
Try a prompt
- “What should I wear for hot yoga?”
- “Find me a sustainable blazer under €200”
- “Match this photo, neutral palette, ankle boots”
- “Outfit for a beach wedding in September”
- “Skincare routine for combination skin in winter”
- “Furniture for a 12-square-metre studio”
Not a search bar. A stylist.
Free-form input. Look-level output. Inventory + tone + brand-safety baked in.
Natural-language prompts
Shoppers type or speak the question. The stylist parses intent (occasion, budget, palette, sustainability, fit) and returns a curated bundle.
Image-in, looks-out
Drag a photo of an outfit you love, get the closest match from your catalog + the UGC where customers wore something similar.
Voice prompts
iOS + Android speech APIs deliver hands-free shopping. Great for mobile, voice assistants and in-store kiosks.
Shop the answer
Every result is a clickable look: products tagged, UGC linked, add-to-cart inline. The stylist closes the loop.
Brand-safe by default
Refuses out-of-scope prompts (politics, medical advice), only recommends in-stock SKUs in your shipping region, and respects your tone-of-voice guide.
Trained on your library
Per-brand fine-tune on top of a shared base. Knows your catalog, your UGC, your size + colour conventions, your seasonal drops.
Four steps. Conversation to cart.
- Step 1
Listen
Stylist receives the shopper’s prompt (text, voice, or image). LLM parses occasion, budget, fit, palette, sustainability flags and any explicit constraints.
- Step 2
Retrieve
Two-pass retrieval: first against your tagged UGC (so real-customer photos anchor the answer), then against your catalog with semantic + filter ranking.
- Step 3
Curate
Returns a 4-8 piece “look” bundle: hero UGC + matching products + complementary cross-sell + sizing tip. Multi-currency, locale-aware.
- Step 4
Convert
Inline add-to-cart, save-to-wishlist, share-to-friend. Conversation history persists per session so “show me cheaper” works.
Same idea, half the friction.
StoryStream pioneered the AI stylist for luxury + automotive, enterprise contracts, weeks of setup. Idukki ships the same agentic surface to mid-market DTC: self-serve, one rate per session, UGC-first, live in 30 minutes.
Full Idukki vs StoryStream comparison- Shopify product recsRanks SKUs alone. No language input. No UGC.
- StoryStream AI StylistEnterprise contract, sales-led, weeks to set up.
- OpenAI plugin / GPT shopNo brand-safety guardrails, no real-time inventory, no UGC.
- Idukki AI StylistSelf-serve, one rate per session: UGC-first, in-stock-aware, voice + image input.
Priya Anand
Founder, Folio Studio (agency)
“Our PDP search bar became a stylist. Average session went from 1.8 to 5.4 product views, and our beauty client saw a 31% PDP CR lift in week three.”
FAQ
How is this different from Shopify product recommendations?+
Shopify recs rank SKUs based on co-purchase + view history. The AI Stylist accepts free-form questions ("what should I wear for hot yoga?"), pulls in UGC (real customer photos in your catalog), and assembles outfits, not single-product picks. It also handles image and voice input.
Where does it run? On Idukki servers or my checkout?+
The stylist runs in the Idukki edge layer. Prompts are tokenised + embedded in your workspace; the LLM call hits an Anthropic Claude endpoint via Idukki’s service. No prompts leave your workspace, no shopper PII is sent to the model, and we honour your DPA.
Can I control the tone of voice?+
Yes. Each workspace has a tone-of-voice config (formal, playful, technical) + a refusal policy (what topics the stylist won’t touch). You can also pin facts the stylist must say (e.g. "always mention our free 30-day returns").
What about out-of-stock SKUs?+
In-stock-aware by default. We sync inventory hourly (5-minute on Plus) and never recommend a product you can’t fulfil to that shopper’s shipping region.
How is the model fine-tuned?+
Per-brand transformer fine-tune on top of a shared base. We index your catalog (titles, descriptions, attributes), your UGC (auto-tagged), and your knowledge base (size charts, FAQs, brand guidelines). Re-indexing happens on every catalog update.
Does it speak multiple languages?+
Yes, 28 locales. The stylist responds in the shopper’s language even if your catalog is monolingual; it translates SKU titles + UGC captions on the fly.
Can shoppers upload a photo?+
Yes. Image input runs through our brand-safety vision model first (no NSFW, no third-party logos), then the stylist returns the closest catalog matches plus UGC where customers wore something similar.
What does it cost?+
Included on Pro + Enterprise. Starter plan gets 1,000 stylist sessions/month included; overage at $0.01/session. Predictable per-session rate, no per-source or per-language surcharges layered on top.
The storefront that answers back.
Hand us a CSV of your catalog and an IG handle. We’ll deploy a stylist trained on your library to a sandbox of your store within one business day.
Stop renting six tools. Ship one.
Spin up your first widget in 4 minutes. Migrate from Bazaarvoice, EmbedSocial, Tolstoy or Videowise in a day. Your CFO will love us. Your CRO already does.
- No credit card
- Cancel anytime
- SOC 2 + GDPR
