You already know what a form does: it collects answers. An AI form does something a normal form cannot. It reads what the visitor typed, uploaded or said, thinks about it, and gives them something back on the spot. A quote. A summary. A first answer. An image. A shortlist.
That is what AI Forms in AI Engine are for, and they are the feature almost nobody knows about, so this is the plain explanation.
A form versus a chatbot
A chatbot is a conversation: open ended, turn by turn, the visitor steers. A form is structured: you decide the fields, the visitor fills them, and one submit produces one result. Most business questions are forms in disguise. “How much would this cost?”, “Is my file the right format?”, “Which plan fits me?” do not need a chat. They need three fields and a good answer.
Forms are also easier to trust with a budget. One submission is one AI call with a known shape, not a conversation that can wander for twenty turns.
What an AI form can take in
- Text, short or long, with the usual validation.
- Images, which the model can actually look at. A photo of a room, a screenshot, a product.
- Audio, recorded in the browser or uploaded, transcribed and used as input.
- Files, PDFs and documents, with multi upload and a staged list so people can add or remove before submitting.
- Conditional logic, so a field can appear only when an earlier answer makes it relevant.
What it can give back
An answer written for that submission. An image generated from the fields. A file. The result renders as formatted text, so a step by step reply looks like one instead of a wall of text.
The form is a block or a shortcode, so it goes anywhere: a page, a sidebar, a product tab, a synced pattern.
Five forms people actually build
- Lead qualification. Company, need, budget, timeline. The AI writes a two line assessment for the visitor and a summary for you, and the real conversation starts already sorted.
- Quote or estimate. Dimensions, materials, options. Out comes a first estimate and the assumptions behind it, before anyone books a call.
- Image intake. Upload a photo, get an opinion: is this the right part, what condition is it in, what would you recommend.
- Document first pass. Drop a PDF, get a summary and the three questions the visitor should ask next.
- Support triage. Describe the problem, attach a screenshot, get the most likely fix and, if it is not that, a ticket with everything already filled in.
None of these need code. The prompt behind the form is where the work goes: what the AI is, what it must return, what it must never claim.
How it connects to the rest
The form uses whichever model you picked in AI Engine, so switching from one provider to another is a dropdown, not a rebuild. It runs on your own API key, so a quiet form costs nothing and a busy one costs cents. Usage limits apply to forms the way they apply to chatbots, so a viral form stops rather than surprising you on the invoice.
Developers get a small JavaScript API, prefill and setField, to fill a form from the page around it without manual event tricks.
The honest part
AI Forms is a Pro feature. The free version of AI Engine gives you the chatbots, the content tools and the MCP server; forms sit in AI Engine Pro, alongside Knowledge and Function Calling. If your use case is one of the five above, that is usually the first Pro feature that pays for itself, because it does the thing a human used to do by email.
Ten minutes to a first form
- In Meow Apps → AI Engine, open the Forms tab and create one.
- Add fields. Start with two, you can always add more.
- Write the prompt: who the AI is, what to produce, and the one thing it must not do.
- Add an output block, place the form with the block or shortcode, submit a test.
- Set a usage limit before you publish the page, not after.
If you want the form to answer from your own content rather than general knowledge, pair it with Knowledge, the same way you would for a chatbot. The chatbot walkthrough covers that part.
