ChatGPT Sites Use Cases: 8 Insane Things You Can Build Today

ChatGPT Sites lets you describe a website and get back a live, hosted link, no designer and no code. Once you've seen it work, the real question isn't how it works, it's what's actually worth building with it. So here are eight ChatGPT Sites use cases I've built myself, with the time each one took and what came back, from a business site that captures leads to a full app with its own logins and database.

Every one of these ran inside ChatGPT Work, and every timing is a real run, not a demo promise. If you've read our guide to ChatGPT Sites, this is the "now what do I do with it" companion.

What ChatGPT Sites is, quickly

ChatGPT Sites is OpenAI's website builder inside ChatGPT Work and Codex. OpenAI describes it as a way to "create, preview, publish, and share interactive websites and lightweight apps." You describe what you want, include the word website or tag @Sites, and it builds, hosts and hands you a live link on its own chatgpt.site domain, which you can later point at a custom domain you own. It's in public beta on the Plus, Pro and workspace plans. The eight below are where I've found it earns its place.

The 8 use cases worth stealing

1. Launch a business website in an afternoon

The clearest use is validating a business idea fast. I gave it a detailed brief for a fictional B2B office-plant service called Leaf It To Us: The goal, the values, and to design the brand and logo itself. Fifteen minutes later I had a live, hosted site with a logo it created, its own images, a lead-capture form, and a separate private inbox to see who enquired. It even built a before-and-after slider showing a bare office transformed, using its own image generation. If your business is simple enough, this can be your actual site, not just a test.

2. Spin up on-brand landing pages

If you already have a brand, the trick is feeding it your style. For our own 9x community landing page, I put two files in the project folder: A brief.md for what the page should do, and a design.md teaching it our exact style guide. What came back matched our main site, pulled our next live workshop in dynamically, and enforced a business-email-only sign-up. That page is live now with people signing up through it every day. If you use agents at all, a design.md for your brand is one of the highest-return things you can write.

3. Build an interactive dashboard

This is the one I think has the biggest impact for most teams. I showed it a screenshot of a YouTube analytics dashboard built by an agency, connected it to our channel data, and asked it to rebuild the whole thing to our brand. In about fifteen minutes it reverse-engineered every metric and chart and produced a fully interactive dashboard, hover states, time filters, retention graphs and all. I then had it set a scheduled task to refresh the numbers every morning at 10am, and shared the link with my team. The number-crunching lives as Python scripts in the project folder, so nothing is a black box. Work that used to mean a two-week wait on a BI team now takes an afternoon.

4. Create a smart, dynamic form

Dynamic forms, where the next question depends on the last answer, normally mean an expensive form builder and a lot of clicking. I described the questions, the branching logic and the scoring in plain language and had it build the form as a site in just over twenty minutes. Ours routes an individual toward our community and a company toward either a booking or a contact form, depending on team, budget and whether they're the decision-maker. Writing the logic into a prompt is far quicker than assembling it by hand in a builder.

5. Give every client their own portal

If you're an agency, a freelancer or in B2B sales, per-client pages suddenly become viable because they're so fast to make. I built a reusable client portal structured around a clients folder, with a subfolder per client holding their proposal, call notes and schedule. Each client gets a unique link and an on-brand portal with their scope, signed contract, open tasks and schedule tabbed up, and it took fifteen minutes. Onboarding a new client is just a new folder and a nudge. You could also have it pull the details straight from a CRM like HubSpot or Attio.

6. Rebuild a tool you pay for

Some paid tools are simple enough to just rebuild. I use TinyPNG to compress images but keep hitting the free-tier limit, so I had ChatGPT open it in its built-in browser, work out how it functions, and recreate it as its own branded app called Pixel Ferry, then improve on it by saving every compressed image straight to our Google Drive. Half an hour of work, one API connection, and a tool that does exactly what I need without the daily cap. The lesson isn't "clone software"; it's that a small, annoying tool in your stack is often a half-hour build away.

7. Ship an internal tool that takes action

The difference between a dashboard and an internal tool is that a tool lets you do something, not just read. My example is a sales quote builder. Instead of re-explaining the pricing logic, I attached the messy Excel sheet that held it, and had it build a "Relays Quote Studio" web app that reverse-engineered every formula. Each rep logs in with their own account, prices update live, and it blocks a 70% discount because that breaks the rules of the offer. It even generates the customer-facing PDF proposal. OpenAI points to internal apps as a core use, and it's where I'd start if you want the biggest win.

8. Build a full app with logins and a database

The one that surprised me most: A proper web app with its own database and login. I had it build an expense tracker where I photograph a receipt on my phone and AI extracts the amount, vendor and category. It came with a users table and a receipts table I never configured, and login through "sign in with ChatGPT," so each user only sees their own data. It works across web and mobile off the same database. Building AI in means adding your own OpenAI API key, so mind that anyone using it spends your credits.

What ChatGPT Sites can't do yet

It's worth knowing the fences before you plan around it. Sites is in public beta, and it's on the paid tiers, Plus, Pro and workspace plans, not Free or Go. It can't process payments except through a third-party processor, and OpenAI is strict that it must not handle sensitive data like health or payment-card details, so it's no Shopify replacement. And out of the box, user logins only work through "sign in with ChatGPT," so everyone using your app needs a ChatGPT account. OpenAI's own line is that Sites suits focused internal tools, dashboards and calculators, and a more complex product may still need a real engineering project. That matches what I've seen: it's brilliant for simple sites, internal tools and prototypes, and those are exactly where most of the waiting happens in a business.

If you want the how-to rather than the what, our guide to building a website with ChatGPT walks through the first build step by step, and the full ChatGPT Work guide covers the tool it all runs inside.

The takeaway is simple. If you've been waiting on someone else to build a dashboard, an internal tool or a page, you can now build it yourself. Look at what your team is waiting on, pick the smallest one, and build that first. Teaching teams to do exactly this, from asking ChatGPT questions to shipping their own tools, is what our Cowork training programme does, function by function, on the tools you already run. If that's the shift you want to make, book a call with us.

Common questions

What can you build with ChatGPT Sites? Live business websites, on-brand landing pages, interactive dashboards, dynamic forms, per-client portals, rebuilt versions of simple tools, internal tools that take actions, and full apps with logins and a database. Each of the eight above is a real build, most done in fifteen to thirty minutes.

Is ChatGPT Sites good for internal tools? Yes, that's where I'd start. Unlike a dashboard, an internal tool lets your team take action, like a sales quote builder that enforces your pricing rules. OpenAI highlights internal apps as a core use case, and it's where the time savings are biggest.

How long does it take to build a site? In my runs, fifteen minutes for a business site or a client portal, around twenty for a dynamic form, and about thirty for a more involved app that drives a browser and connects to Google Drive. Times vary with complexity, but these are finished, hosted results, not drafts.

Do I need to code to use ChatGPT Sites? No. You describe what you want in plain language, and it builds and hosts it. You do get access to all the underlying files in your project folder, so a developer can take over later if the idea gains traction.

Can ChatGPT Sites build an app with user accounts? Yes. It can create a database and user logins with no configuration from you, using "sign in with ChatGPT" out of the box, so each user only sees their own data. For a different sign-in method you'd connect a tool like Supabase or Xano.