Build an AI chatbot with Python

Key Takeaways

hey this is Kevin I'll make this quick I just launched a brand new course called build an AI chatbot with python in this course you'll create your first AI app in 60 minutes using Lang chain and Lang graph in this video I'll tell you about Lang chain and Lang graph as well as how you can access llms for free via API if you like what you hear please check out the course it's only $9 and the link is in the description below thanks in this course you'll build a custom chat bot using an llm Lang chain and Lang graph let me explain each of these terms llm is short for large language model you've almost certainly used an llm like GPT 4 Gemini Llama Or Claude probably through a chat-based interface these models are known as Foundation models because they are very large scale and serve as the foundation for a wide variety of Downstream tasks they're trained on vast data sets and designed to be general purpose meaning they can be used as is or adapted for specific applications Lang chain is a framework for developing applications powered by llms for example if you wanted to build a custom customer service spot for your company on top of GPT 4 you could use Lang chain Lang graph is an extension of Lang chain that allows you to build more complex workflows for your application you can build complex workflows using Lang chain alone but adding on Lang graph makes certain aspects of app development way easier which is why we're going to use both in this course Lang chain and Lang graph are both free open-source python libraries their development and direction are Guided by a four profit company called Lang chain which also offers some paid features for debugging and deployment our use of Lang chain and Lang graph will be 100% free and and there are also some llms that you can use for free as I said we'll be building an app on top of an llm using Lang chain and Lang graph one of the benefits of the Lang chain framework is that you can use any llm you want we'll be writing code to tell Lang chain and Lang graph what to do and it will communicate with the llm of your choice using API requests most llm options are paid but a few options are free if you have a credit card and can afford spending a few dollars I recommend using a model from open AI who are the makers of chat GPT at the moment I'd recommend using GPT 40 mini which is fast cheap and high quality pricing is based on the number of input and output tokens you use GPT 40 mini currently costs 15 cents per million input tokens plus 60 cents per million output tokens using open ai's tokenizer we can see how tokens are used to represent words by copying and pasting this text we can see that most words and punctuation marks become their own tokens whereas some words like tokenization and open AI are split into multiple tokens for fun we can see the IDS of each token anyway the point is that you'll be charged based on how many tokens you send to the model and how many tokens it sends back if you don't have a credit card or you don't want to spend anything this page has free options my current recommendation from here is Mistral law platform because it's completely free and has very high token limits the main downside is that they're allowed to use any data you send to them to improve their model in other words your data does not remain private to you also their free plan is a bit slower than GPT 40 mini and the results aren't quite as good but that is fine while you're learning once you need your data to be private you can switch to a paid option and by then you'll have a good understanding of how many tokens you're using so you can estimate the cost of switching to a paid model if you like what you hear please check out the course it's only $9 and the link is in the description below thanks