
| Agent coding usage | 40 hours over 25 days (author's claim) |
| Estimated agent cost | Hundreds to thousands of rubles/month (author's claim) |
| Safety rules | 5 essential rules (shown on screen) |
Who will this analysis save an hour?We’ll send the link with the “actually” takeaway — they won’t have to watch the whole video.
A chatbot is reactive; it waits for you to ask a question and then stops. An AI agent is proactive; you give it a goal, and it autonomously chooses which tools to use and which steps to take until the goal is achieved.
No. Agents are slower, more expensive, and less predictable than traditional automation. Use rigid 'scenarios' for tasks that follow the same steps every time, and only use agents for tasks that require text analysis or flexible decision-making.
The main risks include high costs from infinite loops, 'hallucinations' leading to incorrect actions (like sending a wrong email), and security vulnerabilities like prompt injections where external text tricks the agent into leaking data.
Yes. Tools like Claude Projects, Custom GPTs, and Yandex AI Studio allow you to build agents by simply writing instructions in plain language and uploading relevant files or connecting basic tools.
The video recommends using the MCP (Model Context Protocol). It acts like a universal standard (similar to USB) that allows different AI models to connect to your CRM, databases, or files without custom integration code for every model.