So I challenged Anything to build a Service Request Portal from scratch: user authentication, a database, form validation, status tracking, and a dashboard. This review breaks down everything I found.
I’ll also compare Anything to alternatives like Bolt.new and tell you exactly who should (and shouldn’t) use this platform.
What Is Anything (Create.xyz)?
Anything is an AI-powered app builder that turns natural language prompts into fully functional web applications with real databases, authentication systems, and backend logic.
Unlike traditional no-code builders like Webflow or Bubble that rely on drag-and-drop interfaces and pre-built components, Anything works more like having a developer sitting next to you.
You type “Build me a service request portal with user authentication and a dashboard,” and within 30 seconds, you’re looking at a live preview with actual working code.
The platform doesn’t lock you into templates. It generates custom applications from scratch based on your specifications, complete with:
- Full-stack architecture: Frontend UI, backend API routes, and database schemas
- Real authentication: Sign-up, login, and user management out of the box
- Database integration: Actual relational databases with proper table relationships
- 30+ integrations: ChatGPT, Google Maps, email services, payment processors, and more
Who Is Anything For?
Anything works best for people who need custom functionality quickly and don’t want to be constrained by templates.
Here’s who gets the most value from this platform:
- Startup founders and entrepreneurs testing business ideas can use Anything to build MVPs in hours instead of weeks.
- Developers doing rapid prototyping will appreciate Anything’s ability to scaffold entire applications instantly. You can describe a complex app structure and get a clean Next.js codebase as your starting point.
- Small business owners and operators who need internal tools can build exactly what they need without hiring developers.
- Freelancers and consultants building client projects can use Anything to deliver custom solutions faster and more profitably.
Anything Pros and Cons
- Full-stack apps generated in under 60 seconds
- Real React/Next.js code you can download
- Conversational editing feels natural and fast
- Automatic mobile-responsive layouts without configuration
- Database schemas created based on prompt logic
- 30+ integrations including AI and payment tools
- One-click publishing with custom domain support
- Live preview updates instantly as AI builds
- Authentication and user management built-in automatically
- Clean, production-quality code with proper structure
- No vendor lock-in; host anywhere you want
- Technical errors occasionally appear in the logs console
- The credit system limits prompt iterations per month
- Documentation assumes some technical familiarity with concepts
- Steep learning curve for database relationships
Try Anything free and see how fast you can go from concept to working app. No credit card is required for your first project.
Anything Features
- Natural language app generation from prompts
- Full React/Next.js code export capability
- Built-in user authentication and authorization
- Relational database with automatic schema creation
- 30+ integrations (AI, maps, payments, email)
- One-click deployment with SSL certificates
- Conversational editing via chat interface
- Mobile-responsive design generated automatically
My Hands-On Experience with Anything (Create.xyz)
I decided to put Anything (formerly known as Create.xyz) to the test. My goal was to build a functional Service Request Portal where users could sign up, submit maintenance requests (like plumbing or cleaning), and track their status.
I recorded the whole thing to see exactly how many hurdles I’d have to jump over.
1. Getting Started: Signing Up and First Impressions
When I first landed on the Anything homepage, I wasn’t met with a complicated dashboard or a gallery of templates.
Instead, there was just a big, white prompt box that said “Turn your words into mobile apps, sites, tools, and products.” It’s a bold design choice because it puts all the pressure on the user to know what they want to build right away.
I looked at the top right and clicked the “Get started” button. This opened a clean signup modal with three main options:
- Sign up with Google
- Sign up with Apple
- Use an email address
I decided to go with my email. After I typed it in and hit “Submit,” the site didn’t ask for a password. Instead, it moved to a “Magic Link” screen.
I had to hop over to my Gmail, where I found an email from hello@createanything.com with a 6-digit code. I typed that back into the Anything site, and I was in.
The interface that appeared next was basically a blank slate. On the left, there was a thin sidebar for my projects, and the rest of the screen was dedicated to that central prompt box.
I noticed a small label at the bottom that said “v0 Max,” which I assume is the engine doing the heavy lifting. There was no long tutorial or annoying pop-ups asking about my company size. I was just… there.
What I thought about this was:
The signup was remarkably fast. I’m so used to builders asking me twenty questions about my job title and “intent” before letting me see the tool, so this was a breath of fresh air. It feels very modern, almost like a developer tool disguised as a consumer app.
2. My First Prompt Attempt: Describing the Portal
Now came the moment of truth. I had a specific list of requirements ready for my Service Request Portal. I didn’t want to just say “make a portal”; I wanted to see if the AI could handle specific logic and fields.
I pasted my vision into the box. Here is exactly what I asked for:
- Project Name: Service Request Portal.
- Core Features: User authentication (sign up/login), a service request form with a dropdown for service types (Plumbing, Electrical, etc.), a date picker, and an urgency selector.
- The Dashboard: A place where users can see their submitted requests, filter them by status (Pending, In Progress, Completed), and see color-coded status badges.
- User Profile: A page where users can manage their name, email, and phone number.
I didn’t notice any obvious character limits as I was typing, which was great because I was being quite wordy. Once I hit the arrow button to send the prompt, the screen changed entirely.
What I thought about this was:
Entering the prompt felt very natural. I liked that I could just dump my list of “must-haves” into the box without worrying about formatting it into a special code language. It feels like the tool actually wants you to talk to it like a person.
3. Watching the AI Build My App
As soon as I hit enter, the “Builder” interface took over. This part was actually pretty exciting to watch.
On the left side of the screen, I saw a file explorer start to populate. Folders and files like app/api/services/route.js and components/Header.jsx started appearing one by one.
The AI didn’t just dump a finished product on me; it showed me its thinking process. A chat-like window on the side gave me updates like:
- “I’ll build you a professional Service Request Portal with authentication.”
- “Let me start by setting up the foundation.”
- “Now let me create the backend API routes.”
Within about 30 seconds, a live preview of my landing page appeared on the right side of the screen. It was a clean, blue-and-white design with a big headline: “Home Services, Simplified.”
It even created “features” cards for “Easy Requests,” “Track Progress,” and “Save Time” without me specifically asking for those decorative details. It filled in the gaps of my prompt with logical, professional-looking content.
What I thought about this was:
Watching the code files appear in real-time was very satisfying. It gave me confidence that there was actual logic being built under the hood, not just a static image. I was impressed that it understood the need for a backend API right away without me having to explain what an API is.
4. Assessing the Output: Does the App Actually Work?
After the AI finished its “thinking” and the file explorer stopped flickering, I spent some time really digging into the actual app it produced.
It’s one thing for an AI to claim it built a “complete portal,” but it’s another thing to have a functional, high-quality product. I went through every page to see if it actually followed my instructions.
The first thing I checked was the Landing Page. Honestly, I was expecting something basic, but it gave me a polished “Home Services, Simplified” hero section with a clean blue-and-white color palette. It included:
- A clear “Create Free Account” call-to-action button.
- A feature list (Easy Requests, Track Progress, Save Time) with relevant icons.
- A fully functional header that changes based on whether you’re logged in or not.
When I moved to the Service Request Form, I was genuinely impressed. I had asked for specific fields, and the AI delivered them perfectly. It didn’t just give me generic text boxes; it built a logical, high-quality form:
- Service Type: A proper dropdown menu with Plumbing, Electrical, Cleaning, and Landscaping.
- Description: A large text area for details.
- Preferred Date: A real date picker. I noticed it even added logic to prevent picking dates in the past, which is a professional touch I didn’t even ask for.
- Urgency: A selector with Low, Medium, and High options.
Finally, I checked the Profile Page. It was straightforward but complete, allowing users to edit their name, phone number, and address.
The email was set to “read-only,” which is standard practice for security, showing me the AI understands basic app logic.
Even the branding change I requested, switching “Service Portal” to “QuickFix Home“, was applied everywhere: the logo, the landing page, and even the footer copyright.
What I thought about this was:
The quality of the output is significantly higher than I expected. It didn’t just “mock up” a site; it built a functional application with real logic, form validation, and database connections.
This part satisfied my prompt 100%, and the attention to detail, like the date picker logic and color-coded badges, makes it feel like it was built by a human developer.
5. Exploring Design Customization and Branding
The initial app looked good, but it was a bit generic. The header said “Service Portal,” and I wanted something more specific and personalized for my use case.
I wasn’t sure how to customize the site Anything had just built for me. Unlike traditional website builders where you might hunt for a “Settings” panel or “Theme” options, I decided to just ask directly in the chat interface at the bottom of the screen.
I typed: “How do I edit and customize the website?”
The AI responded almost instantly with a comprehensive guide. It even listed the specific files I might want to edit, like the landing page (/apps/web/src/app/page.jsx) and the header component (/apps/web/src/components/Header.jsx).
What stood out to me was how conversational the instructions were. Instead of technical jargon, the AI gave me plain English examples like “Change the primary color from blue to purple” or “Rename ‘Service Portal’ to ‘HomeHelper.'” This felt approachable, even for someone who might not be familiar with React or Next.js file structures.
Encouraged by this response, I decided to test it out immediately. I typed: “Change ‘Service Portal’ to ‘QuickFix Home'”
The AI got to work. I could see it “thinking” for about 1.8 seconds, then it showed me its process. It identified that “Service Portal” appeared in multiple places: the Header component (both desktop and mobile versions), the landing page header, and the footer. It then used what it called a semantic_edit function to update both files automatically.
Within seconds, the AI responded: “Done! I’ve updated your site to use QuickFix Home instead of Service Portal.” It listed exactly what changed:
- Header logo text (both desktop & mobile)
- Landing page branding
- Footer copyright
What I thought about this: The speed and accuracy were impressive. Instead of manually hunting through multiple files and components, I just described what I wanted in plain English and the AI handled all the implementation details. The fact that it caught both the desktop and mobile versions of the header showed it understood the code structure deeply, not just doing a simple find-and-replace.
One thing worth noting: You’re not locked into using just the chat interface. If you’re comfortable with code, you have complete freedom to directly edit any file in the project. I clicked on the “Code” tab at the top of the screen and saw the full file explorer with the entire project structure: apps, web/src, components, utils, everything.
6. Setting Up the Backend and Database
This is where most no-code tools get really confusing, but Anything handles it differently. I clicked on the “Database” icon in the far left sidebar.
A screen appeared showing me the actual tables that the AI had generated based on my prompt. I saw:
- auth_accounts and user_profiles for the login system.
- services for the actual maintenance requests.
I could actually see the schema for the services table. It had columns for id, user_id, type, description, date, urgency, and status.
There was also an “SQL Runner” where I could write raw queries if I knew how. I didn’t need to use it for this test, but it was nice to see that the data wasn’t locked away in some mysterious “black box.”
I could also see the “Backend API” section which listed out paths like GET /api/services and POST /api/services. It felt like a real developer environment.
What I thought about this was:
This part was very powerful. Most “app builders” just give you a spreadsheet-style table, but this felt like a real relational database. The fact that it automatically linked the “services” to the “user_id” impressed me because that’s usually the part where beginners get stuck.
7. Testing the Integration Superpowers
I wanted to see what else this portal could do, so I checked out the “Integrations” or “Add-ons” menu. The documentation mentioned over 30 built-in integrations. I saw a huge list that included:
- AI Models: ChatGPT, GPT-4 Vision, and Audio Transcription.
- Design Libraries: Chakra UI and shadcn/ui.
- Utilities: Google Maps, Google Search, and Resend (for sending emails).
- Data: Movie databases, SEO keyword research, and web scrapers.
The way you add these is by typing a “/” in the chat. For example, if I wanted to send an email to the homeowner when a request was completed, I would type “/Resend” and tell the AI to “Send an email notification when the status changes to Completed.”
The integration list is huge, covering everything from QR code generators to US Sales Tax calculators. It makes the tool feel like it has infinite room to grow.
What I thought about this was:
The integration list is probably the strongest part of this tool.
8. When Errors Started Appearing
It wasn’t all smooth sailing. While I was clicking around the “Database” and “API” sections, I noticed some red text in the “Logs” console at the bottom of the screen.
I saw the following exact error message:
name: “ClientFetchError”, message: “Failed to fetch. Read more at https://error.anything.com/auth/…”
It seemed like the authentication system was having trouble connecting to the backend for a moment. I also saw a warning about “Integrations may be temporarily unavailable” during a transition to their new builder.
To try and fix the fetch error, I did what every tech person does: I hit the “Refresh” icon in the preview window. That seemed to clear it up, and the “Sign In” page loaded correctly after that. It was a small glitch, but it reminded me that this is still a very complex system running in a browser.
What I thought about this was:
The error messages are a bit technical. If you aren’t a developer, seeing “ClientFetchError” might be scary. However, I liked that there was a log console at all. Most no-code tools hide the errors, which makes them impossible to fix. Here, I could at least see what was wrong.
9. Responsive Design and Mobile View
Since many people would use a “Service Request Portal” on their phones while looking at a broken pipe, I needed to see if it worked on mobile.
At the top of the preview window, there were two icons: a desktop and a phone. I clicked the phone icon.
The “QuickFix Home” layout adapted immediately:
- The navigation links moved into a “hamburger” menu.
- The hero image and text stacked vertically.
- The “New Request” button on the dashboard became a full-width button at the bottom.
Everything looked great. It didn’t look like a desktop site shrunken down; it looked like a native mobile app. I didn’t have to write any special code or click any “make responsive” buttons. The AI just knew to use flexible layouts.
What I thought about this was:
This part impressed me. Usually, making a site look good on mobile is a huge headache that involves a lot of manual resizing. Anything handled it automatically, and the resulting mobile UI looked better than some professional apps I use daily.
10. Publishing to the World
Once I was happy with the portal, I wanted to see how hard it was to actually “go live.” I clicked the “Publish” button in the top right corner.
A panel slid out with a few options:
- Web and Backend: This was checked by default.
- Domain: It gave me a random URL: service-request-portal-250.created.app.
- Database Changes: It warned me that “This migration will create new tables.”
I clicked “Publish changes,” and a loading bar appeared. About 20 seconds later, I got a “Live” badge. I clicked the link, and I was looking at my actual, functional website in a new tab.
I was able to click “Create Free Account,” sign up, and see the dashboard. It wasn’t just a preview anymore; it was a real URL I could send to anyone.
What I thought about this was:
The one-tap publishing is amazing. Compared to setting up hosting, SSL certificates, and database connections on a traditional server, this is a hundred times faster. It makes it extremely easy to test an idea with real users without spending a whole weekend on DevOps.
11. Can I Actually Own This Code?
One of my big fears with AI builders is “vendor lock-in.” I wanted to see if I could take my code and leave if I wanted to.
I clicked the “Code” icon at the top of the file explorer. This opened a window with several options:
- Embed Project: Gave me an iframe code to put the app on another site.
- Download Project: This would give me a ZIP file of the entire React/Next.js project.
- Copy Code: I could copy specific files.
This was a huge relief. The code was clean, using standard libraries like Tailwind CSS and React. If Anything ever goes out of business, I could literally download my files, put them on my own server, and the app would still work.
Final Verdict: My Honest Assessment
After spending an hour building the QuickFix Home portal, I’ve realized that Anything is less of a “website builder” and more of an “app factory.”
The Good:
- Speed: I went from zero to a live, functional portal with a database and auth in under 10 minutes.
- Conversational Editing: Asking the chat to change the branding or colors is much faster than clicking through menus.
- Real Code: The fact that you can see and download the React code is a huge plus for anyone who cares about ownership.
- Integrations: The list of third-party tools is incredibly deep.
The Not-So-Good:
- Technical Errors: You might see some “Fetch Errors” or technical logs that could be confusing for total beginners.
- Credit System: You have to be careful with your prompts so you don’t burn through your monthly limit on minor tweaks.
- Learning Curve: While the prompt is easy, understanding how the database and APIs connect still requires a little bit of “logic” thinking.
Compared to other tools like Softr or Glide, Anything feels more powerful but slightly less “polished” in terms of the editor UI.
It’s perfect for someone who wants to build a specific, functional tool quickly and doesn’t want to be limited by what a template can do. If you can describe it, this tool can probably build it.
Pricing & Plans
Anything offers three tiers based on a credit-based system, where credits are consumed each time the AI generates code or runs integrations. Here’s how the pricing breaks down:
| Plan | Price (Monthly) | Price (Yearly) | Credits/Month | Best For |
|---|---|---|---|---|
| Free | $0 | $0 | 3,000 (one-time) | Testing the platform, small hobby projects |
| Pro 20k | $19 | $15.83/mo ($190/year) | 20,000 | Freelancers, small business tools, regular builders |
| Max | $199 | $165.83/mo ($1,990/year) | 220,000 | Agencies, complex apps, power users needing advanced AI |
Annual plans save you 2 months (17% discount)
Payment & Refund Details
- Payment methods: Credit card via Stripe
- Refund policy: No refunds due to per-generation AI costs, but contact support for unusual circumstances
- Unused credits: Additional purchased credits roll over; base plan credits expire monthly
- Cancellation: Cancel anytime from your dashboard; takes effect at billing period end
My Recommendation
Consider Max only if you’re building at scale (3+ complex apps monthly), need automated testing, or want the fastest AI responses for rapid iteration. For most users, Pro 20k offers the best value.
Alternatives to Anything
Anything excels at generating full-stack applications from conversational prompts, but it’s not the only AI-powered app builder available.
If you’re evaluating options, Bolt.new (by StackBlitz) is the closest alternative worth considering. Both tools use AI to generate complete applications with real code, but they differ significantly in their approach to iteration, deployment, and pricing.
| Feature | Anything | Bolt.new |
|---|---|---|
| Ease of Use | Conversational chat interface; requires logical thinking about features | In-browser IDE with AI chat; more developer-focused interface |
| Best For | Custom business tools, client portals, MVPs needing databases and auth | Rapid prototyping, frontend-heavy apps, developer experimentation |
| Mobile Apps | Mobile-responsive web apps; can export to App Store | Web apps only; mobile-responsive but no native deployment |
| Backend & Data | Full relational database with schema generation, API routes included | Limited backend; uses Web Containers (browser-based Node.js) |
| Design Flexibility | AI-generated designs; can edit code directly or ask AI to modify | Full code editing in browser; more manual control during generation |
| Integrations | 30+ built-in (ChatGPT, Stripe, Google Maps, email services) | Fewer pre-built integrations; relies on npm packages |
| Code Ownership | Download complete Next.js project as ZIP; host anywhere | Download or deploy to StackBlitz; open-source friendly |
| Deployment | One-click publish with custom domains and SSL | Deploy to StackBlitz, Netlify, or Vercel; requires more setup |
| Pricing | Free (3k credits) / Pro $19/mo (20k credits) / Max $199/mo (220k credits) | Free tier available; Pro starts at $25/mo for enhanced features |
Choose Anything if: You need a complete business application with authentication, database relationships, and backend logic generated automatically. It’s ideal for non-developers or entrepreneurs who want to describe their requirements conversationally and get a production-ready app without touching code.
Choose Bolt.new if: You’re comfortable with code and want more hands-on control during the generation process. Bolt excels at rapid frontend prototyping and gives developers a familiar IDE experience while leveraging AI for scaffolding. It’s better for quick experiments or projects where you’ll heavily customize the generated code yourself.
Final Verdict on Anything
Anything delivers on its promise. You can genuinely go from idea to working app in minutes.
The AI understands complex requirements, generates clean React/Next.js code, and handles backend logic that would normally require a developer. The conversational interface feels natural, and owning the code eliminates vendor lock-in concerns.
However, it’s not perfect. The credit system requires careful prompt management, technical errors occasionally surface in the logs, and non-developers may struggle with database concepts. The platform assumes you can think logically about features and workflows, even if you can’t code.
