मुख्य सामग्री पर जाएं

technical · Agentic AI & Tool Use

AI Agents Fundamentals

Build an AI agent you can read, limit and count, from a one-page plan to a run against a real model.

कोर्स के अंत तक, आप ये कर पाएँगे:

  • Tell an agent from a chatbot or fixed workflow by what each does at a surprise
  • Write a system prompt in five parts and test it against fixed edge cases
  • Turn a Python function into a tool with a description and argument schema
  • Judge an agent by running its test set several times and counting correct runs
  • Read a trace forward to find the first wrong step and fix it there

$9/माह से9 घंटे5 अध्याय20 पाठ

कोर्स अभी अंग्रेज़ी में हैं।

“An agent is not a chatbot with tools. It picks each step from the last result. Here is how to build one and know whether it works.”

इस कोर्स के बारे में

यह किसके लिए है

  • You write Python and have called an AI model from your own code
  • You want to build an AI agent you can understand, control and fix, without a ready-made framework hiding the steps
  • Your team has an AI system and needs to know whether it is really an agent, and whether it should be

आपको क्या चाहिए

  • You can read and write basic Python
  • You have called an AI model from code before, for example through the OpenAI or Anthropic API

आप कौन से कौशल अभ्यास करेंगे

  • AI agent design
  • System prompts
  • Tool calling
  • Checking AI output
  • Human approval steps
  • Python
  • Reading run logs
  • Cost and risk checks

कितना समय लगता है

कुल मिलाकर लगभग 9 घंटे

  1. अध्याय 1 · What is an AI agent, and when is it worth building one?2 घंटे
  2. अध्याय 2 · What does an agent need before you write its code?1.5 घंटे
  3. अध्याय 3 · How do you make a model's output reliable enough to act on?1.5 घंटे
  4. अध्याय 4 · How does an agent use tools?2 घंटे
  5. अध्याय 5 · How do you build, run and debug the agent?2 घंटे

इसमें क्या शामिल है

  • 20 पाठ
  • 83 पठन
  • 20 हल किए गए उदाहरण
  • 50 अभ्यास
  • 39 त्वरित जाँचें
  • 20 आम भ्रम
  • 5 प्रोजेक्ट
  • 20 क्विज़
  • अध्याय के अंत की 5 जाँचें

पाठ्यक्रम

5 अध्याय, 20 पाठ

पूरा कोर्स: हर अध्याय और पाठ। विस्तार के लिए अध्याय पर टैप करें। शुरू करने पर पाठ खुलेंगे।

    • Which Part Did It?
    • Would Your Prompt Sell a Tahoe for $1?
    • What Can a Read-Only Agent Still Lose?
    • The Spec Said Metric. Who Checked?
    • Same Question, Two Answers
    • Is That Citation a Real Case?
    • Make the API Hold the Shape
    • The Count Went Up. Did It Help?
    • Why Can't the Model Call a Function That Works?
    • Can a Tool Definition Stop a Wrong Call?
    • Whose Rows Did the Model Just Read?
    • The Quote Timed Out. Did It Send?
    • What Will You Open on Thursday?
    • The Short Function That Is the Agent
    • A Freeze You Ask For, a Key You Hold
    • Where Was the Fault Made?

करियर

यह कोर्स आपको कहाँ तक ले जा सकता है

वे भूमिकाएँ जिनसे यह कोर्स जुड़ता है, और वे कौशल जिन्हें कोर्स पूरा करने के बाद आप अपने रिज़्यूमे और LinkedIn पर लिख सकते हैं।

5
जिन भूमिकाओं से मेल खाता है
AI EngineerML EngineerSoftware EngineerPlatform EngineerPrompt Engineer

वे कौशल जो आप बनाएँगे

कोर्स पूरा करने के बाद इन्हें अपने रिज़्यूमे या LinkedIn प्रोफ़ाइल में चिपकाएँ।

  • Agent design and scopingagent architecture
  • System prompt writing and agent stateprompt engineering
  • Tool definition and tool callingfunction calling
  • Agent permission control and safe hand-offsAI safety controls
  • Agent evaluation and output checkingLLM evaluation
  • Agent loop implementation, tracing and debuggingagent orchestration

प्रश्न

प्रतिबद्ध होने से पहले

In an agent, the model chooses an action, reads what it returned and chooses the next action from that result, until it is done or a rule in the code stops it. A chatbot answers one message, and the next move is yours. The course names an agent's five parts: the model, the system prompt, the tools, the state and the loop code.

No. You build the agent loop in plain Python so every part stays visible: the loop, the tool registry, structured output and the traces. You should already be comfortable calling an LLM API and writing Python.

ReAct alternates reasoning and acting: the model decides the next step, calls a tool, reads the result and decides again. Chapter 5 has you wire that loop from scratch, with a step limit, a named end reason and a record of every step, so you can see why a run ended where it did.

In chapter 4 every tool returns its outcome, success or failure, in one fixed shape the model can read, and the code retries only a failure worth retrying, never a call with a side effect. Chapter 5 adds traces, so when a run goes wrong you can read forward to the first wrong step.

Pathfinder, a supplier-search agent. It takes a sourcing question, searches a mock supplier catalog, compares candidates on lead time and risk, and holds any quote request until a sourcing analyst approves it. You plan it on one page in chapter 1, write its prompt and limit its tools in chapter 2, and build, run and debug it in chapter 5.

More often than most teams expect. Lesson 1.3 gives you a test for whether a task needs an agent at all. A fixed workflow or a single model call is often cheaper, faster and easier to debug, and the course shows how to tell the difference before you build.

With boundaries set before the first run: a short list of tools and argument values the agent may use, limits enforced in code rather than only in the prompt, and a person approving any action that cannot be undone. Every run also leaves a trace you can read afterward.

Engineers who have called an LLM API and written Python and now want to build agents they can reason about. It is relevant to roles such as AI engineer, LLM application developer and AI solutions engineer.

Tutorials usually show an agent that works once. Here each chapter pairs one part of the build with the failures it prevents, and you practice it on the same project.

Python 3.12 or newer with uv, Pydantic for structured output, a small Python mock of the supplier catalog, and tool calling through the OpenAI Responses API.

Only a little. You should have called an AI model from your own code at least once, and be able to write a Python function, read JSON and use a terminal. The course explains tool calling, structured output and the agent loop from the start.

No. You write the loop, the tool registry and the trace yourself in plain Python, using a model API directly, so you can see what each part does. Pydantic is used to define argument and answer shapes.

Pathfinder is the agent you build across the course. It searches a mock supplier catalog, answers with supplier ids it can ground in tool results, and has one side-effecting call that waits for your approval at the terminal.

Most of the build and testing runs offline with a scripted model that replays fixed replies. The final chapter runs against a real model, which needs an API key; the runs are short.

आज ही शुरू करें