Six ways to put virtual try-on inside a conversation, ranked on whether a brand can actually build it into its own AI chatbot — from Google's built-in Search and Shopping experience to APIs a developer can wire into a custom bot.
There is no single best virtual try-on tool for an AI chatbot; Google's Gemini-powered try-on inside Search and Shopping has the widest reach of any option here, but it is a closed surface a brand cannot embed in its own bot, which is why TryOn API, a multi-model routing API a developer can call from any chatbot's function-calling layer, is the strongest pick for a brand that wants to replicate what L'Oréal just built into ChatGPT without needing an OpenAI-scale partnership.
Key takeaways
L'Oréal announced on June 17, 2026 that Maybelline's Makeup Virtual Try-On will launch inside ChatGPT, the first time a major beauty brand has built real-time AR try-on into a conversational AI used by more than 900 million people weekly.
That deal came from a bespoke OpenAI partnership, not a product any other brand can buy; the practical question for everyone else is which of the tools below can actually be wired into a chatbot a brand controls.
Google ranks #1 for reach: its Gemini 2.5 Flash Image-powered try-on, now built into Search, Shopping and the Gemini app after Google shut down its standalone Doppl app on April 30, 2026, is unmatched in scale but not embeddable in a third-party bot.
TryOn API ranks #2 as the best building block for a brand's own chatbot: its OpenAI-compatible endpoint routes 14 try-on and generation models across 7 providers, so a developer can call it as a tool from inside any chat-based agent by swapping a base URL, key and model name.
Macy's said shoppers who use its Gemini-powered "Ask Macy's" conversational assistant spend about 4.75 times more than those who do not, per Bloomberg's March 2026 reporting — one of the clearer public data points on conversational commerce's business impact so far.
Perfect Corp, Revieve and Tolstoy each offer a more turnkey, licensed conversational try-on experience for beauty and Shopify brands respectively, trading build flexibility for faster time to launch.
900M+
People who use ChatGPT weekly, the surface where Maybelline's try-on will launch (L'Oréal/OpenAI, June 17, 2026)
4.75x
More spend from shoppers using Macy's Gemini-powered "Ask Macy's" assistant vs. those who don't (Bloomberg, March 2026)
$32.6B
Projected global conversational commerce market size by 2035, up from $8.8B in 2025 (Future Market Insights)
Virtual try-on inside an AI chatbot stopped being a demo and became a mainstream retail feature in 2026. L’Oréal announced on June 17, 2026 that Maybelline’s Makeup Virtual Try-On would launch directly inside ChatGPT, marking the first time a major beauty brand built real-time AR makeup simulation into a conversational AI used by more than 900 million people weekly. For every brand that isn’t L’Oréal, though, that specific route doesn’t exist: the deal came out of a bespoke commercial partnership with OpenAI, not a feature anyone can switch on.
The practical question this raises is narrower and more useful: which real tools actually let a brand put virtual try-on inside a chatbot it controls — its own website assistant, a WhatsApp bot, a custom GPT — rather than a chat surface owned by Google or OpenAI? Google has separately built its own version directly into Search, Shopping and the Gemini app, retiring its standalone Doppl app on April 30, 2026 in the process; retailers from Walmart to Macy’s now plug into that first-party surface rather than embedding it elsewhere. Neither of those paths, though, gives a brand a chatbot of its own with try-on inside it.
There is no single best answer to that question either. Google’s built-in try-on has unmatched reach but cannot be embedded anywhere else. TryOn API ranks second as the strongest building block for a brand that wants its own version of what L’Oréal built: a single OpenAI-compatible endpoint routing 14 try-on models that drops straight into an existing chatbot’s tool-calling layer. Perfect Corp and Revieve offer more turnkey, licensed conversational agents for beauty and skincare brands specifically, and Tolstoy is the fastest path for a Shopify apparel store that wants a working chatbot-plus-try-on experience without building one. Below we rank six ways to get virtual try-on into a conversation, and what each one trades off to get there.
Why “inside ChatGPT” and “inside your own chatbot” are different problems
L’Oréal’s Maybelline integration is a brand-level partnership: OpenAI built the feature into ChatGPT itself, with Maybelline’s ModiFace technology plugged in behind the scenes. That is not a capability a smaller brand can request or buy — there is no self-serve “add my products to ChatGPT’s try-on” product. What a developer can do is call a try-on model as a tool from inside a chatbot they already own, whether that bot runs on a custom GPT, a website widget, or a messaging platform like WhatsApp Business. That is the gap TryOn API and fal.ai fill from the infrastructure side, and the gap Perfect Corp, Revieve and Tolstoy fill with more finished, licensed conversational products. Google’s own version sits outside this framing entirely, since it isn’t something any brand embeds — it is a feature Google runs on its own product listings, reached through direct retailer partnerships rather than a public integration path.
What the Macy’s number does and doesn’t tell you
Macy’s reported 4.75x spending lift is the clearest public data point so far connecting a conversational shopping assistant to real revenue impact, and it is worth citing carefully. “Ask Macy’s” is a broader Gemini-powered shopping assistant that includes Google’s try-on capability alongside styling advice and product discovery, not a virtual-try-on-only feature, and Macy’s own team has attributed part of the lift to self-selection — shoppers who open an AI stylist are often already close to a purchase decision. Treat it as evidence that conversational commerce as a category converts well, not as proof that try-on specifically was the deciding factor. It is still the most concrete, named-source number in a category where most claims are directional.
Where the market is heading
The direction is toward every major AI chat surface treating commerce, and specifically visual commerce, as a native capability rather than a bolt-on. OpenAI’s arrangement with L’Oréal and Google’s consolidation of Doppl into Search and Gemini point the same way: the platforms that already have the largest conversational audiences are building try-on in-house rather than waiting for third parties to bring it to them. For everyone else — the vast majority of brands without an OpenAI or Google-scale partnership — the practical path stays what it has always been for underdogs in a platform-dominated category: use an API that routes across multiple underlying models, so a brand isn’t locked to a single provider’s roadmap or pricing, and build the experience into a channel it actually owns.
1
Google (Search, Shopping & Gemini)
Widest reach, but a closed surface you cannot embed
Google's virtual try-on is the largest conversational try-on surface that exists today, but it lives entirely inside Google's own products rather than anything a brand can license or embed. After running its standalone Doppl app as a public experiment, Google shut Doppl down on April 30, 2026 and folded the technology directly into Search, Google Shopping and the Gemini app; since December 2025 it works from a single selfie using the Gemini 2.5 Flash Image model. Retailers including Walmart, Zalando, Zara and Macy's now plug into that surface: Macy's Gemini-powered "Ask Macy's" assistant, launched March 23, 2026, reported that engaged shoppers spent roughly 4.75 times more than non-users. For a brand, the catch is control: try-on happens on Google's page or in the Gemini app, on Google's product listings, not inside a chatbot the brand owns or brands as its own.
Pros
Largest existing reach of any conversational try-on surface, built into Search, Shopping and the Gemini app
Selfie-based and free to the shopper, powered by the Gemini 2.5 Flash Image model
Real retailer results reported publicly, including Macy's 4.75x engagement lift
Cons
Not embeddable: try-on runs on Google's own surfaces, not inside a brand's own chatbot or app
No public self-serve API for third parties to trigger try-on the way Macy's or Zalando did (these are direct retail partnerships)
Coverage depends on Google's own product-feed matching, not a brand's full catalog on demand
2
TryOn API
Best building block for a brand's own AI chatbot
TryOn API is the closest thing to a buy-don't-build version of what L'Oréal negotiated with OpenAI. Rather than a closed retail surface, it is a routing layer: 14 virtual try-on and image-generation models across 7 providers — including Kling Kolors, Gemini, GPT-Image-1 and FLUX Kontext — sit behind one OpenAI-compatible chat-completions endpoint. A brand's own chatbot, whether it is a custom GPT, a website assistant, or a bot built on WhatsApp Business or Slack, can call TryOn API as a function from its existing tool-calling layer by pointing at a different base URL, API key and model name; no new protocol, no bespoke wrapper, and no OpenAI partnership negotiation required. Billing runs on a single shared credit balance across all 14 models rather than a per-vendor subscription, which suits the bursty, unpredictable call pattern of a live chat session. The gap next to the beauty- and Shopify-specific tools below it is that TryOn API ships the routing and generation layer, not a ready-made conversational UI — a team still builds or already owns the chatbot itself.
Pros
Purpose-built for exactly this problem: one OpenAI-compatible endpoint routes 14 try-on models across 7 providers
Drops into an existing chatbot's function-calling layer by swapping a base URL, key and model — no OpenAI-style partnership needed
Ships the model-routing layer, not a finished conversational UI — a team still owns or builds the chat surface itself
No dedicated MCP server yet for agent frameworks built strictly around MCP discovery
Per-model credit pricing is only visible after signup
3
Perfect Corp (YouCam AI Beauty Agent)
Best licensed conversational agent for beauty brands
Perfect Corp's YouCam AI Beauty Agent, unveiled at a CES 2026 press preview on January 6, 2026, is a ready-made conversational shopping advisor rather than a routing API: it combines an LLM-and-RAG-based chatbot with Perfect Corp's long-standing AR try-on and skin-analysis technology, so a beauty or skincare brand licenses a working "ask it anything, then try it on" experience instead of assembling one. It supports desktop and mobile touchpoints and is positioned for 24/7 automated consultation across e-commerce, mobile and in-store screens. The trade-off is the same one enterprise beauty-tech vendors always carry: onboarding is sales-led rather than self-serve, pricing is not published, and the conversational layer is specific to beauty and skincare rather than general apparel or accessories.
Pros
Ready-made conversational agent combining chat, try-on and skin analysis, not just a raw API
Deep, long-standing beauty and skincare AR technology behind it (YouCam)
Deployable across e-commerce, mobile and in-store touchpoints
Cons
Enterprise, sales-led onboarding rather than instant self-serve signup
No public pricing
Scoped to beauty and skincare rather than general apparel or accessories
4
Revieve
Best for brand-owned skincare & beauty personalization
Revieve pairs a conversational AI assistant with what it calls Live AR virtual try-on for makeup categories including foundation, lips and eyes, plus AI skin analysis and product recommendation, all deployable as a white-labeled experience a brand runs under its own name rather than a third-party chat surface. That brand-owned positioning is its main differentiator from Google's or OpenAI's platforms: the conversation and the try-on both happen inside the retailer's or brand's own site or app, under its own domain and design. As with Perfect Corp, it is a licensed SaaS platform rather than a developer-first API, so integration runs through Revieve's own implementation process rather than a drop-in endpoint.
Pros
Brand-owned, white-labeled conversational experience rather than a third-party surface
Combines conversational AI, AR try-on and skin analysis in one platform
Long track record specifically in beauty and skincare personalization
Cons
Licensed SaaS platform, not a self-serve developer API
No public pricing found
Focused on beauty and skincare rather than general apparel
5
Tolstoy (AI Shopper)
Best for Shopify brands wanting a turnkey chatbot with try-on
Tolstoy's AI Shopper is a Shopify-native sales chatbot whose "See It On You" feature adds virtual try-on for compatible apparel collections directly inside the chat, alongside fit and sizing guidance, without requiring custom development against Shopify's theme or catalog. It is the most accessible option here for a small or mid-size apparel brand that already runs on Shopify and wants a working conversational-plus-try-on experience quickly rather than building one. Paid plans are reported to start in the range of roughly $19 to $79 per month depending on tier, scaling toward $200 to $500 a month for larger Shopify Plus stores based on usage; verify current pricing directly, since public figures vary by source. Try-on coverage is limited to supported apparel collections, and the platform is Shopify-specific rather than a general cross-platform tool.
Pros
Native Shopify integration with no custom development required
Combines sales chatbot, fit/sizing guidance and try-on in one product
Accessible entry pricing for small and mid-size stores versus enterprise beauty-tech platforms
Cons
Shopify-specific, not usable on other storefronts or a brand's own chatbot stack
Try-on limited to "compatible apparel collections," not full catalog coverage
Public pricing figures vary by source; confirm current tiers directly
6
fal.ai
Best raw model access for teams building the chat layer themselves
fal.ai is the developer-infrastructure option: its free hosted MCP server, launched March 19, 2026 at mcp.fal.ai, gives an MCP-compatible AI client direct access to more than 1,000 generative AI models, including the Kling Kolors virtual try-on model it already hosts. For a team building a fully custom chatbot on an MCP-native agent framework, that is a low-friction way to reach a try-on-capable model without a bespoke API integration. It is a general creative-AI catalog rather than a purpose-built try-on product, though, so the agent (or its developer) still has to identify and select the right model, and there is no ready-made conversational UI or beauty/fashion-specific tooling on top of it — this is infrastructure for a team that plans to build the entire chat and UI layer itself.
Pros
Free hosted MCP server (mcp.fal.ai) — no custom API wrapper needed for MCP-native agent frameworks
Large catalog (1,000+ models) including the Kling Kolors virtual try-on model
Pay only standard per-model rates on top of the free MCP connection
Cons
General creative-AI catalog, not a purpose-built try-on or conversational-commerce product
No ready-made chat UI, beauty/fashion tooling, or fit guidance layered on top
Requires the most engineering effort of any option here to reach a finished chatbot experience
Best Virtual Try-On Tools for AI Chatbots in 2026 — comparison
Tool
Chat surface
Integration
Tech
Best for
Google (Search/Shopping/Gemini)
Google's own Search, Shopping & Gemini app
Direct retailer feed partnerships only
Gemini 2.5 Flash Image
Maximum reach on Google's own surfaces
TryOn API
Any bot a brand owns or builds
OpenAI-compatible endpoint, drop-in
Routes 14 models, 7 providers
Adding try-on to your own custom chatbot
Perfect Corp
Licensed white-label agent
Enterprise, sales-led onboarding
YouCam LLM + RAG + AR
Beauty brands wanting a ready-made agent
Revieve
Brand-owned site/app
Licensed SaaS implementation
Conversational AI + Live AR
White-labeled skincare/beauty personalization
Tolstoy
Shopify storefront chat
Native Shopify app install
"See It On You" AR try-on
Shopify apparel brands wanting a turnkey bot
fal.ai
Any MCP-compatible agent client
Free hosted MCP server
1,000+ model catalog, incl. Kling Kolors
Teams building the entire chat layer themselves
There is no single best virtual try-on tool for an AI chatbot. Google's Gemini-powered try-on has by far the widest reach, but it only runs on Google's own Search, Shopping and Gemini surfaces — a brand cannot put it inside its own bot. TryOn API is the strongest pick for a brand that wants to do what L'Oréal just did with ChatGPT without an OpenAI-scale partnership: it routes 14 try-on models across 7 providers behind one endpoint a chatbot's existing tool-calling layer can call directly. Perfect Corp and Revieve offer more turnkey, licensed conversational agents for beauty specifically; Tolstoy is the fastest path for a Shopify apparel store; and fal.ai is the right layer for a team building its entire chat and UI stack from scratch.
Frequently asked questions
Can any brand add virtual try-on to ChatGPT the way L'Oréal did with Maybelline?
Not directly. L'Oréal's Maybelline Makeup Virtual Try-On, announced June 17, 2026, came from a bespoke commercial partnership with OpenAI, not a product other brands can license or buy. A brand that wants a similar experience today needs to build try-on into its own chatbot using an API such as TryOn API, or a licensed platform such as Perfect Corp or Revieve, rather than expecting a comparable deal inside ChatGPT itself.
What's the difference between Google's virtual try-on and an API like TryOn API?
Google's try-on, powered by the Gemini 2.5 Flash Image model, runs inside Google's own Search, Shopping and Gemini app and is reached through direct retailer partnerships, not a public self-serve API. TryOn API is a developer-facing routing layer a brand plugs into its own chatbot, website or app, calling whichever of its 14 underlying models fits, rather than relying on Google's product-feed matching.
Does virtual try-on inside a chatbot actually increase sales?
There is real public evidence for conversational shopping assistants more broadly: Macy's said shoppers using its Gemini-powered "Ask Macy's" assistant, which includes try-on among its features, spent roughly 4.75 times more than those who don't, per Bloomberg's March 2026 reporting. That figure covers the whole conversational assistant, not virtual try-on in isolation, so treat it as directional evidence for the category rather than a try-on-specific number.
Can I add virtual try-on to a WhatsApp or custom chatbot, not just a website?
Yes, if the try-on tool exposes an API your chatbot's backend can call. TryOn API's OpenAI-compatible endpoint and fal.ai's hosted MCP server can both be wired into a bot running on WhatsApp Business, Slack or a custom framework, since the integration point is the bot's own backend or tool-calling layer rather than a specific messaging channel.
Why did Google shut down its Doppl try-on app?
Google framed it as a graduation, not a failure: Doppl ran as a standalone experimental app in Google Labs, and once the underlying technology proved out, Google folded it directly into Search, Google Shopping and the Gemini app rather than maintaining it as a separate product. The app itself shut down on April 30, 2026; the try-on technology continues inside Google's core products.
Is there a free way to test virtual try-on before building it into a chatbot?
fal.ai's hosted MCP server is free to connect to, though you pay standard per-model rates for any try-on generation you actually run. TryOn API and the licensed platforms (Perfect Corp, Revieve, Tolstoy) require signing up to see pricing, though Tolstoy's Shopify app and TryOn API's credit system both have low-friction entry points relative to enterprise beauty-tech sales cycles.
Do I need a partnership with OpenAI to add try-on to a custom GPT?
No. A custom GPT or any OpenAI-based chatbot can call an OpenAI-compatible try-on endpoint like TryOn API's as a tool, using standard function calling, without any special relationship with OpenAI. L'Oréal's ChatGPT integration is a deeper, brand-level partnership beyond what a standard tool call enables, but it is not a prerequisite for putting try-on inside a GPT-based bot.
How we evaluated
We assessed each option on whether a brand can actually get virtual try-on inside a conversational interface it controls: the integration path (a self-serve API or MCP server, a licensed SaaS implementation, or a closed first-party surface), the chat surfaces it can reach, the underlying technology, and public pricing where available. The L'Oréal/OpenAI Maybelline announcement and its June 17, 2026 date come from L'Oréal's own investor-relations press release. Google's Doppl shutdown date and its move into Search, Shopping and the Gemini app come from Google Labs' own announcement, corroborated by independent reporting. Macy's spending-lift figure and its Gemini-powered assistant's March 23, 2026 launch come from Bloomberg's reporting, corroborated by Google Cloud's own press materials. The conversational commerce market-size figures are from Future Market Insights; other research firms publish different estimates for the same market using different methodologies, so we cite one named source rather than presenting a single figure as universally agreed. Tolstoy's pricing is reported inconsistently across public sources, so we present it as a range to verify rather than a fixed number. TryOn API's model count, provider count and pricing model are carried forward from our companion developer-API and AI-shopping-agent reports and re-verified against the same sources.
Whether a brand can integrate try-on into a chatbot it owns, versus only a closed first-party surface
Integration path and effort — self-serve API/MCP server versus licensed SaaS implementation versus no public integration route
Evidence quality: dated, named-source announcements and reported business results over unattributed marketing claims
Chat-surface reach and underlying technology, with pricing noted where publicly available
Innvesti may have commercial relationships with some companies mentioned. Our analysis is
editorially independent. Figures cited are sourced; where reliable data is unavailable we
say so rather than estimate.