Help Center
Learn how appproduce turns your ideas into working apps, and how credits are billed.
What is appproduce
appproduce lets you build apps by chatting: describe what you want in plain language, and the AI handles requirement analysis, planning and coding. You preview as it builds, then publish with one click or download the source to deploy yourself. No coding required.
Get started in four steps
Describe
Create a project and describe the features, pages and data you want. The more specific you are, the better the result.
AI builds
The AI analyzes, plans and codes step by step, with a live preview on the right.
Refine by chat
Not happy with the result? Keep chatting to have the AI adjust it, iterating until it's right.
Publish or download
Publish for a live link, download the source to deploy yourself, or export miniapps to WeChat DevTools.
Project types
Pick a type when creating a project. The AI adapts its generation strategy to each scenario:
Core capabilities
Step-by-step AI generation
Requirement analysis → planning → coding, fully automated, with key points open to your input.
Live preview
Renders as it generates — change one thing, see it instantly.
One-click publish
Get a standalone link after publishing; share to the plaza or use it directly.
Source download
Export a ready-to-run project with a local backend for your own development and deployment.
Document parsing
Upload Word/PDF/PPT/Excel/images to auto-extract text (with OCR) as requirements or source material for generation.
Web search
Search the web during generation for up-to-date content (consumes credits per query).
Knowledge base
Upload documents as references so the AI generates with your business context in mind.
URL ingestion
Report projects auto-detect URLs in your description and collect their content as source material.
Credits & billing
Billing unit
Billing is in credits; see the recharge center for the exact ratio.
Free quota
Every new account gets free credits to start creating. Free credits are valid for one month.
When you're charged
Credits are only consumed when the AI actually generates, edits, searches the web or parses documents. Browsing, preview and publishing cost nothing extra.
How much
It depends on project complexity and how many revisions you make. A clear, complete description reduces unexpected usage.
Recharge & earning credits
Recharge plans
Choose a plan in the recharge center. WeChat Pay and Alipay are supported.
Invite friends
Invite friends to sign up via your referral link to earn bonus credits.
Share your work
Share your project links and earn rewards when visitors open them.
Redeem codes
Redeem codes for credits in the recharge center (some codes may have an expiry).
Tips & tricks
Master these and your results will match your intent — while spending fewer credits.
How to use the knowledge base
Feed your own documents to the AI so it generates from your business context instead of guessing.
- 1In the editor toolbar, click the "Knowledge base" button (book icon) next to "Versions" to open the panel.
- 2Upload documents — PDF, Word, PPT, Excel, CSV, HTML, Markdown, TXT are supported; image-heavy PDFs are OCR'd automatically.
- 3Once uploaded, the platform indexes them and auto-retrieves relevant passages during generation — no manual referencing needed.
- 4Turn on "Strict compliance mode" to answer strictly from your docs, pointing to official sources rather than inventing when uncovered.
KB retrieval consumes a small amount of credits per use. PPT projects have no runtime AI, so the KB isn't shown for them.
How to describe requirements and revisions clearly
The more specific you are, the less rework — and the fewer credits spent.
- 1Start with "who it's for and what problem it solves", then list key pages, features and data fields.
- 2Provide reference examples, URLs or field lists so the AI aligns with your expectations.
- 3For revisions, point to the specific page/element and state "what it is now vs. what you want" — avoid vague "make it nicer".
- 4Break big changes into a few small iterations — steadier and cheaper than rebuilding from scratch.
How to use platform skills
The platform has several AI-invokable capabilities that trigger automatically based on project type and description — no manual switches.
- 1Document parsing: automatically reads text from your uploaded Word/PDF/PPT/Excel/images (with OCR).
- 2Knowledge base retrieval: vectorizes your docs and finds relevant passages during generation.
- 3Web search: fetches up-to-date public info on demand to correct time-sensitive content.
- 4Document export: report projects can export to Word/PDF in one click.
These are platform-exclusive capabilities; if you download the source to deploy yourself, you'll need to wire up the corresponding services.
How to write a proposal report
"Proposal report" is a project type built for structured long-form reports, producing a full chaptered document.
- 1Choose the "Proposal report" type when creating a project.
- 2Describe the topic, key points and target readers; URLs in your description are auto-detected and their content collected as source material.
- 3Upload research materials (PDF/Word/PPT/Excel/images) when you need more sources — they're used as evidence too.
- 4After the report is generated, export it to Word or PDF from the top of the preview.
URLs that fail to collect are listed explicitly — sources are never fabricated.
How to roll back a version
The platform auto-saves before each generation, so you can return to an earlier state anytime — and the rollback itself is undoable.
- 1In the editor toolbar, click the "Versions" button (clock icon) to open the version panel.
- 2You'll see auto snapshots (created before each generation) and versions you saved manually.
- 3Click "Restore" on a version and confirm; your current code is auto-saved first, so you can undo the rollback.
- 4At important milestones, save a version manually with a note to find it later.
Rollback is unavailable while a generation is in progress — wait for the current round to finish.
How to use search
Let the AI reference the latest information from the web to avoid stale content.
- 1In the chat input area, click the "Web search" button (globe icon) to open the search panel.
- 2Enter keywords, review the results and sources, then click "Merge into request" to bring them into the chat.
- 3You can also tick web search during the analyze/plan stage so the AI references real-time info directly.
Web search consumes credits per query and only appears once the platform has web access enabled.
FAQ
What if generation is interrupted?+
Interruptions are usually caused by the model provider dropping the connection or rate-limiting. It typically recovers on its own; you can also click the "Continue" button to resume from where it stopped, or wait a moment, click "Back" in the browser, and re-enter the project.
How do I deploy the system?+
Download the source, or download a Docker deployment package. Note that the package does not include the base software environment — see the included documentation after downloading for details.
Is downloading the source free?+
Downloading the source itself is free, but only paid users who have recharged can download it.
What if I'm not happy with the result?+
Keep chatting to have the AI revise it — iterate as many times as you like. Interrupted or failed generations aren't charged.
Can I get an invoice?+
Yes. Paid recharge orders can request an invoice from your billing page, with personal or company titles supported.
Still have questions?+
Click "Online support" in the footer to scan and add our support team for one-on-one help.