How to deploy a project from ChatGPT

1. Ask the AI for a deployable project

Ask for the source files, dependency manifest, and a README with the exact start command. Specify whether this is a background process (a bot or worker) or a web app that listens for HTTP requests. Ask the AI to list required ENV variable names only; it does not need real tokens.

2. Check the ZIP structure

The project files should be at the archive root, not hidden inside an extra nested folder. For example:

my-bot/
  main.py
  requirements.txt
  Procfile
  README.md

For Python, Deployka uses requirements.txt. For Node.js, include package.json with a working scripts.start and a lockfile for reproducible installs. ZIP uploads can contain up to 20 MB and 300 files. Leave out .env, tokens, node_modules, virtual environments, and local database files.

3. Set the right start process

For a Telegram bot or other background process, a Procfile command can be worker: python main.py. Replace main.py if your entrypoint has another name. A Node.js worker usually runs the existing start script from package.json.

A web app must listen on 0.0.0.0 and use the PORT environment variable supplied by the platform. For FastAPI with an app object in main.py, for example: web: uvicorn main:app --host 0.0.0.0 --port $PORT. Do not hard-code an unrelated port or use a development server in production.

4. Upload the code, set ENV, and deploy

1. Create a project and choose its type; you do not need to set up a VPS or Linux server.
2. Upload a ZIP, provide a GitHub repository, or start from a suitable template.
3. Review source checks and fix any reported blockers before deployment.
4. Add required values to project ENV, such as BOT_TOKEN or OPENAI_API_KEY. Never publish them in a ZIP, README, screenshot, or support message.
5. Click Deploy and wait for the build to finish.

5. Verify more than the running status

For a bot, send it a test message and confirm that it replies. For a web app, open its HTTPS address and check the endpoint you need. Then review Logs for a successful start and no repeating errors.

Common first-deploy errors

• ModuleNotFoundError or an install error — check the package name and dependency manifest.
• Telegram Unauthorized — update BOT_TOKEN in ENV; do not put it in source code.
• A web app does not respond — check that it listens on 0.0.0.0:$PORT, not only localhost or a fixed port.
• SQLite readonly — put the database on persistent storage, for example /data/bot.db, and make sure the application code actually uses that path. An ENV value alone does not change code that ignores it.

Fix the source, ENV, or start command and redeploy the same project. Buying a plan does not fix a code error.

Free and Nano plans

Free Nano is a separate free plan with 128 MB RAM, one project, and limits on term, idle activity, and slot availability; the applicable terms are shown before launch. Nano Reliability provides 256 MB and a persistent slot for the paid term. Choose what fits your project and check the exact terms in the interface before deploying.

More detail: ZIP, ENV, and logs documentation · Telegram bot template · SQLite and persistent data · Update ZIP without a new project