Tonita

By Tonita, Inc.

Overview

Tonita is a search engine for apparel and accessories, built to work with how you actually think about shopping. It brings together products from hundreds of stores, thousands of brands, and millions of items — all in one place, from clothing and footwear to cosmetics and everyday essentials. There isn’t just one way to shop. Sometimes you’re focused on price or practicality; other times it’s about style, fit, materials, or longevity. What matters can shift with the situation — work, travel, special occasions, outdoor use, or everyday wear. Tonita brings these options together, making it easier for you to browse, compare, and make an informed choice. Within ChatGPT, Tonita works best as part of a conversation. You can explore broad questions, half-formed ideas, or specific constraints naturally — whether that means thinking through what to buy for a particular occasion, learning about materials or categories, or narrowing in on a specific kind of product. ChatGPT helps shape your thoughts into useful searches, and Tonita surfaces a wide set of relevant results for you to browse, compare, and refine as the picture becomes clearer. (For more advanced tools like virtual try-on, saving favorites, and deeper browsing, Tonita can also be used directly at Tonita.co.) Tonita is built to support the way you actually think about shopping — by priorities, context, and tradeoffs — rather than forcing everything into rigid categories. This includes the ability to search and explore across aspects like: • Budget and value: From fast fashion and sale finds to higher-end pieces, investment items, and everyday staples. • Style and aesthetic: Minimalist, maximalist, classic, streetwear, formal, playful, utilitarian, fashion-forward, or timeless — and combinations of these. • Trends and longevity: What’s popular right now, last season’s styles, and pieces meant to stay relevant beyond a trend cycle. • Use case and setting: Work, travel, special occasions, outdoor use, events, lounging, or everyday wear. • Fit, sizing, and comfort: Different body types and needs, including petite and extended sizing, relaxed or tailored fits, and comfort-first designs. • Materials and feel: Natural fibers, synthetics, performance fabrics, lightweight layers, structured pieces, or materials chosen for warmth, breathability, or softness. • Durability and care: Easy-care basics, machine-washable items, wrinkle-resistant pieces, or garments meant to be worn, repaired, and kept longer. • Sustainability priorities: Interest in materials, production methods, longevity, or brands taking different approaches to environmental impact. • Brand familiarity and discovery: Well-known labels alongside emerging brands, niche designers, and alternatives to the usual options. • Putting a full look together: Apparel, footwear, accessories, jewelry, and cosmetics, considered side by side rather than in isolation.

Tools

tonita-get-listing-details

ChatGPT

Get detailed information for one or more product listings. When to use this tool Call tonita-get-listing-details whenever you need to know more about specific products that have already appeared in search results. Common scenarios: - Shopper asks about a product: "Tell me more about that Nike shoe" — you need the full description, available sizes, and images. - Preparing for visual search: Before calling tonita-search with query_images, you need the image_id from this tool's response. - Comparing products: The shopper wants to compare two or three items side by side — fetch all their details in one call. - Understanding before searching: The shopper says "find me more like this". Call this tool first to understand what "this" is (its style, color, brand, price range), then construct a well-informed search. What you get back For each listing ID, the response includes: - url: The product page URL on the retailer's site. - text / structured fields: Title, description, price, brand, sizes, and other metadata (varies by retailer). - thumbnail_image: An object with image_id and image_url — the image_id is what you pass to tonita-search query_images for visual search. Examples Get details for a single product ``json {"listing_ids": ["thursdayboots.com!6c66be1f3d6c7f00"]} ` Get details for multiple products to compare `json {"listing_ids": [ "thursdayboots.com!6c66be1f3d6c7f00", "nike.com!a1b2c3d4e5f67890" ]} ` Workflow: shopper says "more like this" about a listing 1. Call tonita-get-listing-details with {"listing_ids": ["thursdayboots.com!6c66be1f3d6c7f00"]} 2. Read the response: it's a "navy blue wool overcoat" with image_id "89e667d28336d70f" 3. Call tonita-search with retrieval_queries=["navy blue wool overcoat"], query_images=[{"image_id": "89e667d28336d70f"}], prioritize_visual_similarity=true`

tonita-search

ChatGPT

Search Tonita's product catalog to find clothing and apparel items. When to use this tool Call tonita-search whenever a shopper wants to find, browse, discover, or compare products. Common scenarios: - Direct product search: "Find me red dresses under $100" - Filtered browsing: "Show me Nike running shoes, not from Amazon" - Visual similarity: "Find items that look like this jacket" (using a reference image from a previous tonita-get-listing-details call) - Exploratory shopping: "What summer dresses do you have?" How search works Search operates in two phases: 1. Retrieval — The engine fetches candidate products using retrieval_queries. Each query string is matched independently against the product catalog, and candidates are pooled together. More queries = broader candidate pool (up to a few variations). 2. Ranking — The pooled candidates are re-scored using reranking_query (falls back to the first retrieval_queries entry if omitted), combined with structured signals like product_type, gender, color, price, and brand preferences. When prioritize_visual_similarity is enabled and query_images are provided, the engine also retrieves visually similar products and visual-similarity scores influence the final ranking. Important rules - One product type per call. Searching for "shirts and pants and shoes" in one call produces poor results. Make three separate calls instead, each focused on one type. - Keep queries objective and literal. The search engine has no pop-culture or trend knowledge. "hot girl walk shoes" won't return good results — use "women's white athleisure sneakers" instead. - Never reference other products in queries. Don't write "similar to the Patagonia jacket". Instead, call tonita-get-listing-details first, read the product details, then construct a new descriptive query based on them. - Use listing details for visual search. When the shopper says "more like this" about a previously shown product, call tonita-get-listing-details to get the image_id, then pass it in query_images with prioritize_visual_similarity set to true. - Omit what you don't know. It is OK to leave optional fields as null. For example, if the user has no brand or color preference, leave those fields null — do not guess. - Reuse `chat_id` within a conversation. Pass the same optional chat_id on every related tool call in one shopping thread (tonita-search, tonita-get-listing-details) so exchanges can be correlated. Omit when the client has no session concept. - Full details on search. By default, structuredContent.listingCards stays lean (listingId, title, currentPrice). On ChatGPT/widget clients, richer cards (brand, URL, images) also arrive under result _meta.listingCards for the host UI — tools-only clients (generic MCP) do not include that _meta at all. Set return_full_details=true when the client LLM needs brand/URL/images in structuredContent itself (required for tools-only clients) — and always pair that with a modest `max_results` (e.g. 5–10). Do not pull full details for large carousels. Examples Simple search ``json { "retrieval_queries": ["red satin midi dress"], "product_type": "dress", "gender": "Womens" } ` Search with brand and price filters `

json { "retrieval_queries": [ "running shoes for daily training", "men's cushioned running sneakers", "lightweight jogging shoes for men" ], "product_type": "running shoes", "gender": "Mens", "restrict_to_brands": ["Nike", "Adidas", "New Balance"], "price_constraints": "under $150" }

` Search with color `

json { "retrieval_queries": ["leather crossbody bag"], "product_type": "bag", "gender": "Womens", "color": "black", "color_hex": "#000000", "price_constraints": "under $200" }

` Visual search (find products that look like a reference listing) ``json { "retrieval_queries": ["minimalist white low-top sneaker"], "product_type": "sneakers", "gender": "Womens", "query_images": [{"image_id": "89e667d28336d70f"}], "prioritize_visual_similarity": true }…

Capabilities

Interactive

Example Prompts

Click any prompt to copy it.

Links

Documentation Privacy Policy

Similar Apps