AI Telegram Bots: Building with LLMs

Daniel academyInstructor: Daniel academy·intermediate·Python

This course is for developers who can already build a normal aiogram bot (handlers, routers, keyboards, FSM) and want the next skill that clients actually pay for: a bot powered by a large language model. You will not read about AI in the abstract. From the first module you build the request that goes to an OpenAI-compatible API, read the reply and the token usage back out, and wire the whole thing into an aiogram handler. Then you add the parts that separate a demo from a product: conversation memory that fits a token budget, streaming so the user sees the answer appear, tools and function calling so the bot can act and not just talk, and hard limits on cost and abuse. Every module gives you a little theory, then a task you solve in the browser, then instant feedback from the same kind of test harness a working developer runs against their own code. By the end you can take a bot from a first reply to something you would put in front of real users.

class Wallet:
    def __init__(self):
        self.entries = []

    def add(self, label, amount):
        self.entries.append((label, amount))

    def balance(self):
        return sum(amount for _, amount in self.entries)

wallet = Wallet()
wallet.add("stipend", 1200)
wallet.add("keyboard", -450)
wallet.add("pizza", -120)
print(f"balance: {wallet.balance()}")