AI Agent Foundations: From LLM to Workflow
A beginner path arranged from published AI Agent articles: understand LLMs first, then Agents and Harnesses, then put Skills and Workflows back into real work.
- Who it is for
- For people new to Agents, or already using AI without yet connecting the concepts to a practical way of working.
- Source material
- The “Agent: Past, Present, and Future” series and related AI Native articles
- Current status
- Pricing and access details are to be announced
Course structure
The path connects published source reading into three stages. Guided course lessons, practice feedback, and progress tracking are still being prepared.
- 01
Understand the LLM first
Start with next-token prediction, then establish the basic coordinates of Prompt and Context.
- What does an LLM actually do?Read original article (Chinese) →Course guide in preparation
- Prompt and Context: what are you giving AI?Read original article (Chinese) →Course guide in preparation
- 02
See how an Agent is made
Move from post-training and chat products to Harnesses, and understand the difference between talking and getting work done.
- Why a bare LLM is not a chat assistantRead original article (Chinese) →Course guide in preparation
- Agent = LLM + HarnessRead original article (Chinese) →Course guide in preparation
- 03
Put an Agent into your work
Choose an appropriate Agent form, then turn repeated experience into Skills and Workflows over time.
- Chatbot, GUI Agent, and CLI AgentRead original article (Chinese) →Course guide in preparation
- Turn experience into Skills and WorkflowsRead original article (Chinese) →Course guide in preparation
- Move from using AI to working AI NativeRead original article (Chinese) →Course guide in preparation
What this page lets you decide
- 01
Place LLM, Prompt, Context, Agent, and Harness on one coherent map.
- 02
See the difference between choosing an Agent tool and giving it repeatable conditions for work.
- 03
Begin a first AI collaboration loop around a real working context.
Release, purchase, and learning access
This course uses existing articles as its knowledge base. The course edition will add an intentional sequence, practice, feedback, and progress beyond the original reading; its format, price, and release are still being confirmed.
This is the public course page. When the course opens, it will be the one authoritative route for sign-in, payment, and the learning centre.
Before the course opens
If the original articles are public, what changes in the course edition?
The original reading will remain public. The course edition will arrange it into a learning sequence and add explicit goals, practice, feedback, and learning records; those course features are not released yet.
Can I start now?
You can begin with the public source reading identified for each lesson. Course tasks, progress, and future paid access will be updated on this page.
When will purchase open?
Scope, price, and access for the first release will be updated here once confirmed. There is no purchase or presale flow today.