Learning Map
1
Model
Model
Initializing the model, sending messages, and reading responses are the basis for all subsequent Agent capabilities.
2
Prompt
Prompt
Improve model reliability with system prompts, few-shots, and structured output.
3
Tools
Tools
Connect Python functions, Web Search, and multimodal input to the Agent.
4
Memory
Memory
Use checkpointer and thread_id to keep conversations contextual across rounds.
5
Project
Project
Combine the previous capabilities into a complete application such as Personal Chef.
Lesson Summary
1
1.1a Foundational Models
Click to enter the explanation page
Model initialization, HumanMessage/AIMessage, and basic invoke process.
- init_chat_model
- messages
- response
2
1.1b Prompting
Click to enter the explanation page
system prompt, few-shot, structured prompts, and Pydantic output.
- system_prompt
- few-shot
- structured output
3
1.2a Tools
Click to enter the explanation page
@tool, tool description, how the Agent decides whether to call the tool.
- @tool
- tool calls
- Agent
4
1.2b Web Search
Click to enter the explanation page
Supplement real-time external knowledge with search tools and let Agents aggregate the results.
- Tavily
- search
- external knowledge
5
1.3 Memory
Click to enter the explanation page
Use checkpointer and thread_id to establish multi-round conversation memory.
- memory
- thread_id
- checkpointer
6
1.4 Multimodal
Click to enter the explanation page
Let messages carry multi-modal content such as pictures, and expand model input forms.
- image input
- content blocks
- vision
7
1.5 Personal Chef
Click to enter the explanation page
Combine models, tips, tools, and memories to build a personal chef agent.
- project
- recipe
- agent workflow
Completion Criteria
- Be able to explain in your own words the problems solved by each lesson in this module.
- Can point out the key API in each example and where it fits into the Agent workflow.
- The capabilities of this module can be migrated to a new small Agent task.