Generate Product Photos From ChatGPT or Claude: Rokon's MCP Studio

Can you generate product photos from inside ChatGPT or Claude?

Yes. Rokon runs an MCP server — a connector that AI assistants such as ChatGPT, Claude, and Perplexity can plug into. Add it in your assistant's connector settings, sign in with your Rokon account, and you can generate photorealistic on-model photos of your garments, check your credit balance, browse models and poses, and track generations without leaving the conversation. Each image generates in about 30 seconds, drawing on the same account and credits as the Rokon web app.

What MCP is, in plain language

MCP (Model Context Protocol) is an open standard that lets AI assistants use external services as tools. Think of it the way apps relate to a phone: the assistant is the phone, and each MCP server is an app it can install. Without a connector, an assistant can only talk about product photography; with one, it can act — call a real service, on your account, with your permission, and bring the results back into the chat.

Rokon's server exposes 8 tools to the assistant. You never call them directly — you write plain requests, in English or Arabic, and the assistant chooses which tool fits.

What your assistant can do once connected

The same conversation can also cover poses (from the 20+ pose library), plan details, and your existing collection — which turns the assistant into a front desk for the whole studio.

How sign-in works, and why it is safe

The connection uses OAuth 2.1 with PKCE — the same class of sign-in flow used by serious SaaS integrations. Practically, three things matter. You authenticate on Rokon's own page, so the assistant never sees your password. The assistant receives a scoped token for your account, not blanket access. And you can revoke the connection at any time from your account without touching the assistant. Your uploads are treated exactly as they are on the web app: never used to train public models, never resold, and your outputs remain 100% commercially yours.

Who actually works this way

Three groups get real leverage from a chat-native studio. Merchants who already live in an assistant — drafting product descriptions, translating listings, planning campaigns — can now produce the imagery in the same thread: one conversation yields listing-ready copy and the on-model photos beside it. Developers and automation builders get a standard protocol for wiring fashion imagery into agent workflows without scraping a UI. And agencies can let an assistant orchestrate routine generation requests across client accounts while humans spend their time on review and art direction.

There is also a quieter fourth group: merchants who found the idea of an AI photo studio intimidating as a standalone product. Asking a familiar assistant in plain words turns the same capability into a conversation — no new interface to learn before the first image exists.

Ask in Arabic, get the same studio

The connector inherits Rokon's bilingual design, so the entire exchange can run in Arabic — a single instruction like «أنشئ صورة لهذه العباية على عارضة خليجية بثلاث وضعيات» is enough to start a generation. That pairs naturally with the writing work Gulf merchants already do in assistants: draft the Arabic product description and generate the matching on-model imagery in the same thread, then publish both to your Salla store together. Most photography tools bolt translation onto an English product; here the assistant is simply talking to a studio that was built bilingual from the start, with a full Arabic UI and mirrored RTL on the web side.

What a working session looks like

A typical flow: you attach front, back, and detail photos of a garment in the chat and describe what you want — model preferences, modesty settings, poses, ratio. The assistant uploads the references, starts the generation, and reports back when images are ready, usually within 30–120 seconds. You ask for pose variants or another ratio; it queues them. A useful closing habit: ask the assistant to summarize what was generated and what it cost — it can read your balance before and after the session, which makes per-drop budgeting effortless. Everything generated lands in your Rokon collection, so nothing lives only inside a chat history — you download or publish through the web app and your Salla or Shopify integration as usual.

The honest limits

Four caveats keep expectations straight. First, MCP support varies by assistant and plan — some clients gate connectors to paid tiers or roll them out by region, so check your assistant's connector settings before planning a workflow around it. Second, chat is linear: a long styling session — comparing many looks side by side, fine-tuning a scene — is faster in the web studio's visual interface than through message-by-message iteration. Third, an assistant can describe an image, but it cannot judge garment fidelity for you; the pre-publish check against the real piece stays a human job. Fourth, bulk catalog runs are not a chat feature — CSV/XLSX runs of up to 1,000 products live in the web app's Bulk Studio on the Business plan.

What it costs

Nothing extra. MCP access uses your existing plan and balance: 30 credits per standard image, 60 for 4K, identical to the web app. The Free plan's 150 credits (~5 images) work through the connector too, Pro at $20/month (~SAR 75) covers roughly 40 images, and Business at $60/month (~SAR 225) about 100 with priority queue. Your balance is shared across web and chat, so there is one number to watch — and you can ask the assistant for it.

No other fashion-photography tool currently offers its studio through an open connector standard, which makes this less a feature than a bet on where merchant workflows are heading. If you already work inside an assistant, connect Rokon and test it on a real garment with your 5 free generations (150 credits) — no credit card required.