Building Your Personal AI Learning Tutor Agent
A tutorial to set up a Shinkai AI Agent for personalized learning plan creation.
This tutorial walks you through creating a Shinkai AI Agent that acts as a personal learning tutor. This agent will interactively help you build structured learning plans with curated resources for any topic.
You’ll build an AI tutor that :
- Adapts to the depth of knowledge wanted.
- Generates relevant subtopics.
- Collects your feedback.
- Finds top YouTube videos, websites, and research articles.
- Organizes resources into a progressive learning plan.
- Read online material and help you understand.
- Export the learning plan.
Step 1 : Conceptualize Your Learning Tutor Agent
Before diving into tools or Agent setup, first define your Agent’s purpose and capabilities :
- Personality : Helpful, enthusiastic, interactive.
- Core Task : Collaboratively build a structured learning plan.
- Key Features : Subtopic generation, resource finding (videos, sites, articles), transcript summarization, progressive content ordering, user feedback integration.
- Output : A well-organized learning plan with clear descriptions and links.
- Extra Features : Download a webpage, export learning plan to markdown file.
Understanding these requirements will guide your tool selection and Instructions creation.
Step 2 : Install Required Shinkai Tools
Based on your Agent’s concept, ensure the following tools are installed in your Shinkai Desktop App. They are available on the Shinkai AI Store :
- YouTube Search
- YouTube Transcript Summarizer
- Google Search
- Arxiv Search
- Download Pages
- Markdown Exporter
Additional search tools are available in the Shinkai AI Store. You could add relevant spcecialized search tools to your Agent.
Should you require a new tool, try Shinkai AI-assisted tool builder.
Step 3 : Configure Each Tool
Next, navigate to the Shinkai Tools tab (wrench icon), and set up the configuration of each tool requiring it. While many tools have sensible defaults, review and adjust settings if needed.
- YouTube Search : Enter your SerpAPI key.
- Download Pages : You can specify your Chrome path, but the tool most likely can detect it.
- Other Tools : Here other tools don’t require any setup.
Note that for additional tools, or for different Agents, you might have to configure more tools. Some tools can have advanced configurations.
You will also be able to override the tools configuration specifically for your AI Agent in its own configuration.
Step 4 : Create a new AI Agent
- In the Shinkai app, click the Agents button (robot icon) on the left-side menu.
- Click + Add Agent.
Step 5 : Define Agent Name and Description
- Agent Name : A unique explicit identifier (e.g., “Learning tutor”)
- Description : A brief summary of its role (e.g., “AI tutor for creating personalized learning plans with videos, webpages and articles.”).
Step 6 : Craft the Instructions
The Instructions guide the Agent’s behavior, tasks, and interaction style. Paste the following detailed prompt into the “Instructions” field :
Notice we use tags to facilitate the LLM understanding. It’s not mandatory, but it’s a good practice.
Step 7 : Select and Configure the LLM/AI
Choose an LLM/AI to power your Agent and configure its parameters :
- Select an LLM/AI : Pick a capable model (e.g., shinkai_free_trial, gpt_4_1). Note you can even pick an existing AI Agent.
- Advanced options (optional) : enable/disable stream, adapt the parameters for tailored LLM behavior.
Step 8 : Add Agent Knowledge (optional)
The “Knowledge” section lets you attach files/folders for the Agent to reference. For this specific tutor Agent, it’s optional as it primarily fetches information dynamically. You could add files on general learning strategies if desired.
Step 9 : Enable Tools
In the Agent’s configuration screen, under “Tools”, enable all tools installed in Step 2 :
- YouTube Search
- YouTube Transcript Summarizer
- Google Search
- Arxiv Search
- Download Pages
- Markdown Exporter
The Agent can only use tools that are explicitly toggled on here.
Note that here you can override tools configuration specifically for this Agent. It won’t affect the configuration of the tools for the rest of the platform.
Step 10 : Schedule task (optional)
Shinkai lets you automate tasks execution by AI Agents on precise schedules.
We don’t need it for our Learning Tutor AI Agent, as we use an interactive process.
Step 11 : Save and Interact with Your Learning Tutor Agent
Click Save. To use it, open a new chat. Select your learning_tutor from the model selector. Custom Agents appear at the top.
You can now start interacting with your AI Learning Tutor.
Initiate a conversation by stating a topic you want to learn. For instance :
- You : “Hi ! I’d like to learn about Quantum Computing.”
The Agent should then begin its process : asking for a level of kownledge depth, proposing subtopics, asking for feedback, and guiding you through building the learning plan as defined in its Instructions.
Below are screenshots of a session with our Learning Tutor AI Agent.
Troubleshooting and Fine-tuning
- Agent Not Following Instructions : Review the Instructions for clarity. Consider a more capable LLM or adjust Temperature.
- Tools Not Used Correctly : Ensure tools are installed (Step 2), configured (Step 3), and enabled for this Agent (Step 8). The Instructions must clearly instruct tool usage.
- Long Waits : Be patient ; resource fetching and summarization take time.
- Iterative Development : Refine your Agent’s Instructions based on interactions.
Considerations
- Tool Reliability/Quotas : External tools may have limits or fail.
- Information Quality : Always critically evaluate resources.
Conclusion
You’ve now built a personalized AI Learning Tutor Agent in Shinkai !
This demonstrates how combining conceptualization, tool setup, detailed instructions, and LLM configuration creates powerful, specialized AI assistants.
For further assistance, explore the rest of Shinkai’s documentation, contact support, or join the community on Discord.