A bot that squeezes any article down to five lines

Here's the idea in one line: a Telegram bot you send a long article or link to, and it sends back a short summary in five lines. A long text shrinks to its core in a couple of seconds. And you'll actually use this every day — yourself.

Why this one

Most "first projects" you build once and forget. This one stays in your Telegram, because it genuinely saves you time.

And here's the neat part. It feels like squeezing text takes some heavy magic. Really, all the "magic" is one good prompt under the hood. The bot is just a pipe: it takes the text, hands it to the model, returns the answer. Build it and you'll see that with your own eyes.

What you'll learn

• Bots and reacting to commands. A bot waits for a message and replies to it. That's the same "comes in → reply goes out" loop big services run on.
• Calling an API. Your bot talks to two services: Telegram (get the message, send the reply) and a model (squeeze the text). Request goes out, answer comes back.
• "The prompt is the feature." Summarizing isn't a separate technology. It's just an instruction to the model: "squeeze this into five lines." A good prompt is your main feature.

A ready starter prompt

Don't tell the agent "make a summary bot" — it'll start guessing the language and the output. Give it context, an example and limits:

The difference is that the strong prompt leaves the agent no room to guess: the flow is clear, the summarizer prompt is spelled out, the error behavior is set. The first result lands much closer to what you wanted.

What you end up with

You'll have a new contact in your Telegram. Send it a link to a twenty-minute read — a couple of seconds later, five lines of substance come back. You skim them and decide whether the full thing is worth it. No feed, no ads between paragraphs. Just what you asked for.

Start small, finish it — and you'll have a bot you actually use.