
| Companies using AI agents | 86% (shown on screen) |
| Instruction file limit | 200 lines (author's claim) |
| GLM 5.2 vs Claude cost | 20-25x cheaper for tokens (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.
While ChatGPT waits for a prompt to respond, an AI agent is given a goal and uses a loop to observe, think, and act autonomously until that goal is reached.
It is the practice of providing the AI with structured files (like agents.md) that define business rules and roles, ensuring the agent acts as a specialized employee rather than a general assistant.
By using a persistent text file, such as 'memory.md', which the agent reads at the start of every session and updates with new information or corrections you provide.
Using open-source models like DeepSeek or GLM via API or running them locally can be significantly cheaper (up to 25x) than using closed models like GPT-4 or Claude Opus.