# Ai-to-pptx: a DeepSeek outline generator with a three-section cap

> Ai-to-pptx pairs a TypeScript frontend with a separate PHP backend so a DeepSeek outline becomes a downloadable PPTX deck. The open source build stops at three sections per detail page, has no mobile view, and treats its extra templates as a paid asset.

**SmartSchoolAI/ai-to-pptx** — Ai-to-pptx is an AI (DeepSeek) assistant for creating PPTX presentations online, with both frontend and backend open source. It uses DeepSeek and other LLMs to generate outlines, offers multiple templates for PPTX generation, and supports exporting PPTX.

- Repository: https://github.com/SmartSchoolAI/ai-to-pptx
- Website: https://pptx.dandian.net
- Stars: 1,463 · Forks: 187
- Language: TypeScript
- License: GPL-3.0
- Published: 2026-10-14 · Updated: 2026-10-14 · Language: en
- Canonical page: https://hysenlabs.com/projects/smartschoolai-ai-to-pptx

## DeepSeek writes the outline, the PHP service writes the file

Ai-to-pptx is not one program. The repository presents the frontend and the backend as both open source, yet the backend lives in a separate PHP project, `ai-to-pptx-backend`, while everything in this repository is TypeScript. That split decides what you can do on your own: with no PHP service running, the frontend has nothing to talk to. On screen the flow is a request for an outline, an outline the user can revise, a choice of template, and an export that writes a PPTX file. A large language model does the first step, and the project names DeepSeek together with other models rather than one vendor, though no page says where that model endpoint or key is configured. The address of the PHP service sits in a single file, `src/views/AiPPTX/config.ts`, and on the server side the options are a Docker image or a standalone PHP environment, with the steps kept in the backend repository rather than here.

## npm install, three scripts, two files that rebrand it

The frontend is a Next.js application, and the whole setup is three npm commands plus two file edits. `npm install` pulls the dependency tree, `npm run dev` starts the development server, `npm run build` produces the production build; the `package.json` scripts behind them are `next dev`, `next build` and `next start`, and there is nothing else to run. This is the project's own checklist, labels included:

```bash
安装: npm install

开发: npm run dev

打包: npm run build
```

Run those three in order and the development server is up. After that, the backend address goes into `src/views/AiPPTX/config.ts`, and `src/configs/auth.ts` carries the two product switches: `AppName` for the name shown in the interface and `mode` for its colour. Set both before you deploy anything, because a build that ships the sample values ships someone else's branding. No Node version is stated anywhere in the guide, even though the repository root holds an `.nvmrc` and an `INSTALL` file, so pin your own runtime before the first install.

## Three sections per page, PDF export never finished

The open source version draws its own limits, and they are worth reading before anything else. There is no membership function: users cannot register, be billed or topped up, and the project says adding that is not hard. On the PPTX detail page only the three section case is supported, because two or four sections have not been tested enough to keep, so the page is pinned to three on purpose. Mobile is absent. Export to PPTX is finished while PDF is still under development, and the ability to edit text, styles and images online, to set a personal logo and background, and to publish your own templates to a sharing platform are all marked as commercial version features. Count what the free build leaves out before you price a service on it: three sections, one working export format, no accounts, no self service templates.

## MUI screens, Capacitor plugins, no mobile build

The screen work is MUI, and the project says so in plain words. `@mui/material` is pinned to 5.10.11, with `@mui/x-data-grid` 5.17.9, `@mui/x-data-grid-pro` ^5.17.19 and `@mui/lab` 5.0.0-alpha.105 beside it. Then the contradiction worth knowing about. The open source limits say there is no mobile functionality, yet the same dependency list carries `@capacitor/core` ^6.2.0 with `@capacitor/filesystem` ^6.0.2 and `@capacitor/share` ^6.0.3, the plugins that wrap a web build in a native shell, and the top level listing has no Capacitor configuration file next to `next.config.js`. Treat those three entries as scaffolding inherited from the base template, not as a mobile app in progress. The root is untidy in a way that costs an afternoon: `package-lock.json` and `yarn.lock` are committed side by side, an `.npmrc` is present, and nothing in the guide says which lockfile wins.

The scripts block goes a little further, adding lint, format and a single icon bundling step on top of dev, build and start:

```json
  "scripts": {
    "dev": "next dev",
    "build": "next build",
    "start": "next start",
    "lint": "eslint --fix \"src/**/*.{js,jsx,ts,tsx}\"",
    "format": "prettier --write \"src/**/*.{js,jsx,ts,tsx}\"",
    "build:icons": "tsc -b src/iconify-bundle && node src/iconify-bundle/bundle-icons-react.js"
  },
```

That last line compiles `src/iconify-bundle` before running its bundler, so changing icons means running it again. The rest of the toolchain is generic admin surface: `axios` for requests, `i18next` 22.0.4 for translations, `draft-js` 0.11.7 for rich text, `jsonwebtoken` ^9.0.0 in a build that ships no login, and `html2canvas` with `file-saver` in the list.

## package.json reads 0.0.1 against a shipped 0.2 tag

Two version numbers and one date describe the state of this build. The last push was on 2026-06-12, the committed `package.json` still declares version 0.0.1, and the single published tag reads 0.2 with a date of 2025-02-10, so the tree gives you no way to tell which state that tag was cut from. Under the heading for its community group the README states that development of this project has stopped. There is no upgrade guide and no migration notes to diff against, which means the upgrade cost is yours to carry. Budget for owning the code rather than waiting for upstream fixes, and keep the PHP side of `ai-to-pptx-backend` matched to whatever shape the frontend expects at the address in `src/views/AiPPTX/config.ts`, because that contract is the only seam between the two repositories and neither side documents it.

## Four free templates the open source grant does not cover

The project's own note about scope is blunt: what the open source agreement covers is the computer software code, and the templates carried inside the system are not part of it. Four sets of templates ship free with the system, and anything beyond those needs a template authorisation bought separately. The template choice offered during generation is therefore not a catalogue of open assets, and a deck built on one of them carries rights that the code grant does not cover. `README_Make_Template.md` is the file to read when you want your own look instead, and the settings for a personal logo and background images are listed as a commercial version feature, so they are not where you start.

## The upstream base and the product layer built on it

The frontend is not written from zero. Its code notes say the code mainly comes from `aippt-react` and thank that project for opening its source so a new product could be built on top of it, and the licensing line is blunt about the split: the code here is GPL-3.0, `aippt-react` is MIT, and the templates inside are excluded from that grant. That upstream is the real alternative, and the difference in approach is concrete. `aippt-react` is the plain React base, while this repository wraps it in a Next.js application, adds the `AppName` and `mode` product switches, moves the server out into a PHP service behind `src/views/AiPPTX/config.ts`, and attaches redistribution duties the base does not carry: modify the code and you must publish your modifications in a public repository such as GitHub, mark in the code and in the running system which parts are theirs and which are yours, keep the author's copyright notices, and ship your own additions under the same terms. If publishing your modifications is not acceptable to you, start from `aippt-react` and write your own backend contract.

## Closed source deployment needs a separate commercial grant

Open source commercial use needs no contact, and it carries two conditions: your modifications stay published, and your own web page carries the URL of your open source repository at the bottom. Closed source use is the other case, and it needs a separate commercial authorisation, which is also the only route to changing the copyright text. A purchased authorisation lets you run a paid SaaS membership product and use the system yourself, and it forbids selling the system on to other users, so resale of the system as a product is ruled out. Two smaller clauses bracket the deal: templates you write yourself can be submitted, and after review they are credited against part of the fee, while an optional technical service is paid once a year by developers doing second stage development work and by nobody else.

## Conclusion

Ai-to-pptx is a finished sample more than a product you can build on for free: the code is readable and the setup is three commands, but you inherit a stopped project, a second PHP service, and a grant that excludes the templates people judge it by. Take it if you want to study or rebrand a DeepSeek-to-PPTX pipeline in TypeScript. Pass on it if you need mobile, PDF output or four section decks. Open src/views/AiPPTX/config.ts and README_Make_Template.md first and settle the template rights before you ship.

## FAQ

### How do I run ai-to-pptx locally, and do I need the PHP backend?

Yes. The frontend reads its API address from src/views/AiPPTX/config.ts, and the backend is a separate PHP repository, ai-to-pptx-backend, deployed either as a Docker image or as a standalone PHP environment.

### Does the open source version of ai-to-pptx let users edit the generated PPTX online?

No. Editing text, styles and images online, per user logos and background images, and publishing your own templates to a sharing platform are all listed as commercial version features. The open source build generates the outline, lets you revise it, and exports the file.

### Which output formats does ai-to-pptx export today?

Export to PPTX is complete and PDF export is still under development, with PNG named alongside them as a target format. The guide does not give a status for PNG.

### Can I use ai-to-pptx commercially without paying?

Open source commercial use needs no contact, provided your modifications stay published and you keep the author's copyright notices. Closed source use, and any change to the copyright text, requires a separate commercial authorisation.

### Is ai-to-pptx still being developed?

The README states under its community group heading that development of this project has stopped. The last push to the repository was on 2026-06-12.

## Sources

- [License: GPL-3.0](https://github.com/SmartSchoolAI/ai-to-pptx/blob/main/LICENSE)
- [Project website](https://pptx.dandian.net)
- [README](https://github.com/SmartSchoolAI/ai-to-pptx/blob/main/README.md)
- [Releases](https://github.com/SmartSchoolAI/ai-to-pptx/releases)
- [SmartSchoolAI/ai-to-pptx on GitHub](https://github.com/SmartSchoolAI/ai-to-pptx)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/smartschoolai-ai-to-pptx