Build an application that uses a language model and stays reliable when the model is wrong, slow or unavailable. You build Larkfield's ticket assistant step by step, from one request to validated answers, safe tools and tested failure paths.
Read every lesson as a guest, or sign in to enroll and track your progress.
Enroll to save the lessons that you complete. It is free.
By the end of this course, you will be able to:
A language model can read a support ticket and draft a good reply in seconds. Putting that into an application is a different job. The model sometimes returns text that your code cannot read, invents an order number, asks a tool for something it should not see, or does not answer at all because the provider is busy.
In this course, you build an application that expects all of this. You start with one request and read every part of it. Then you turn prompts into specifications that you can test, turn answers into validated data, and let the model ask for tools while your code keeps control of what runs.
Larkfield’s help desk gives you one assistant for the whole course: it routes each ticket to a team, says whether a person must check it, drafts a reply and can look up an order. By the end, it retries safely, logs what every call costs, and passes tests for every failure path you have seen.
You do not need an account or a paid key. Every lesson runs with recorded responses from real models, and you can switch to a free local model or your own key at any point. Diagrams and annotated outputs carry the main ideas, so the text stays short.
You completed Python for Practical AI Engineering and LLMs, How They Work and How to Choose One, or you have the same skills. You do not need an API key or a paid account. Every lesson runs with recorded responses, and you can use a free local model or your own key if you want live calls.
Courses to take first: Python for Practical AI Engineering, LLMs: How They Work and How to Choose One.
People who can write Python and know what a language model does, as in LLMs, How They Work and How to Choose One, and want to build a real application on a model API. Developers, data scientists and professionals moving into AI engineering.
Experienced AI practitioners designed this course. AI helped us draft the content; people chose the teaching approach and the examples, and reviewed every lesson. The outputs that you see are the actual results of running the code. How we make our courses
Send your first request. See what an LLM API is, run Larkfield's assistant against recorded responses without an account, read a request and a response field by field, and put the model behind one boundary in your code, with keys kept safe.
Treat a prompt as a specification. Write the task, its audience, limits and output contract; mark untrusted customer text clearly; and compare prompt versions on a fixed set of tickets instead of one good answer.
Make the model's output data you can trust. Define a schema, handle refusals, cut-off and malformed answers, and check the meaning, because valid JSON can still be wrong.
Let the model ask for tools, and let your code decide. Describe a read-only tool, check every argument and permission before it runs, and stop loops that repeat or run too long.
Hold a conversation and show answers as they arrive. Choose which history to send, stream text without trusting it too early, and tell users clearly when something fails.
Make it reliable and affordable. Retry rate limits and timeouts safely, log usage and cost for every call, and test every failure path with recorded responses.
Three end-to-end case studies on open data. Extract product recalls into structured records, answer questions with read-only tools, and classify banking requests and compare an LLM with a fine-tuned model. Complete at least one for the certificate; each one is a portfolio piece.
Complete at least one case study for the certificate. The others are optional. What each one covers
Use every module in one practice project. You collect evidence, and check it against the course outcomes.
Realistic projects from start to finish, on real data, with every output explained. Each one makes a portfolio piece. Complete at least one for the certificate; the others are optional.
Extract product recalls into structured data3 hr
Turn public product-recall notices into validated structured records with a schema, business checks and a measured error rate on a hand-checked sample.
Answer questions with tools over open data3 hr
Build an assistant that answers questions about an open dataset only through validated, read-only tools, with a bounded loop and checked numbers.
Classify banking requests with an LLM2 hr
Classify customer banking requests with prompts and structured output, compare prompt versions on a fixed set, and compare accuracy and cost with a fine-tuned small model.
Loading the reviews…
Took this course? Sign in to review it. Learners can write a review after they complete 25% of the lessons.
