Calicat: a product-design-engineering platform that generates high-fidelity prototypes from natural language

Calicat is a one-stop collaboration platform for product, design and engineering, built by the team behind ProcessOn. Its central premise is simple: people who know nothing about design should still be able to produce high-fidelity prototypes. The method is to move the whole chain — describe the requirement, generate the interface, iterate on it — into a single editor. You state what you want in natural language, AI produces a first draft, and then a human refines it round by round using the component library, auto layout and style replacement. Product, design and engineering therefore talk inside the same file instead of guarding deliverables in three different formats.

What problem it solves

The real cost of the prototyping stage is rarely the drawing itself; it is the back-and-forth of clarification. A product manager writes a requirement in prose, a designer understands something else, and engineering builds whatever they imagined — rework follows. Calicat tries to compress those three steps onto one canvas: a requirement description becomes a visual page, and the page connects directly to components and delivery, with no layer of retelling in between. For independent developers and small teams this compression matters even more, because nobody has to learn a complex design tool first in order to put an idea in front of other people and test it.

Main features

• AI design assistant: Type a requirement in natural language to generate a first draft, then keep iterating on the generated pages instead of starting from an empty canvas.
• Multimodal design input: Turn sketches into prototypes by uploading a photo of a hand-drawn page, which AI reads for elements and structure; a browser extension can also extract the design of an existing web page and convert it into an editable file.
• Requirement document generation: Document cards are produced automatically during design work, turning vague verbal ideas into a structured module that resembles a PRD.
• Component library and styling: A built-in component library covering common scenarios (including TDesign), with style replacement and auto layout, so resizing and spacing changes do not require aligning every element by hand.
• AI streaming prototype generation: Extend from a single page to a full design set, which suits cases where many pages must share one visual language.
• Support for multiple design scenarios: Beyond product interfaces, it also works for social media covers, editorial layout and PPT page design.
• One-stop collaboration: Members collaborate in real time inside the same project, and finished work can be exported as design files or pushed to the development team.

How to use it

1. Sign up: Visit the Calicat website at https://www.calicat.cn/ and create an account.
2. Create a project: Enter a project name and a short description to set up the design file.
3. Describe the requirement: Explain the page you want in natural language in the input box, and AI generates a preliminary design from it.
4. Refine the design: Pick suitable components from the library and polish details with auto layout, style replacement, the image library and the icon library.
5. Add other input methods: If you have a sketch, upload or photograph it so AI can recognise it and turn it into a high-fidelity page; if you want to reference an existing site, extract it in one click with the browser extension.
6. Capture the requirement document: Use the automatically generated document cards to record design intent as a deliverable specification.
7. Collaborate and hand off: Invite team members and communicate in real time; when finished, export the file or push it straight to development. The team has also announced an MCP Server to connect the development stage.

Where it fits

• Product prototype validation: A product manager writes a requirement as a description and gets a discussable page within minutes, validating the direction before committing design resources.
• Faster interface design: Designers use AI for the first draft and spend their energy on ideas and detail work rather than rebuilding basic frameworks.
• Social media content production: Covers and editorial layouts can be generated directly, which suits operations teams that publish visuals at high frequency.
• Document and presentation layout: Automatic layout keeps long documents and PPT pages consistently presentable and reduces manual adjustment.
• Cross-role collaboration: Product, design and engineering share one project file, cutting format conversion and information loss.

Limits and things to watch

AI-generated prototypes still need human review. The tool is good at turning a description into a reasonable page structure, but it cannot judge whether the business logic holds up or whether information hierarchy is appropriate. Complex interaction states, permission branches and error flows usually have to be added by hand. The component library covers common scenarios, while a highly customised visual system still has to be designed yourself. AI also misreads natural language, and the vaguer the description, the more rework follows — writing the requirement clearly is faster than letting AI guess repeatedly. For current plan details, quotas and limits on collaborator numbers, refer to the official website.

FAQ

Is Calicat suitable for someone with no design background at all?

Yes. The entry point is a natural language description, and with the component library and auto layout you can get a structurally complete page without drawing anything; producing a refined visual result still requires some aesthetic judgement.

Can a hand-drawn sketch become an editable prototype?

Yes. After you upload a sketch image or take a photo, AI recognises the elements and layout and converts them into a high-fidelity page, which you can then keep editing in the editor.

Can the generated prototype be handed to developers?

You can export design files or push them directly to the development team. The team has mentioned an upcoming MCP Server intended to further connect the development and delivery stage.

How is it different from traditional prototyping tools?

Traditional tools are mainly about manual drawing, and efficiency depends on how skilled the operator is. Calicat puts generation first and leaves judgement and correction to the human, which fits the stage where an idea needs to be validated quickly.