GENERATIVE AI
A step-by-step tutorial on query SQL databases with human language
Many businesses have a lot of proprietary data stored in their databases. If there’s a virtual agent that understands human language and can query these databases, it opens up big opportunities for these businesses. Think of customer service chatbots, they’re a common example. These agents can take customer requests, ask the database for information, and give the customer what they need.
The benefit of such agents is not limited to external customer interactions. Many business owners or people in companies, even in tech companies, might not know SQL or similar languages, but they still need to ask the database for information. That’s where frameworks like LangChain come in. Such frameworks make it easy to create these helpful agents/applications. Agents that can talk to humans and at the same time, talk to databases, APIs, and more.
LangChain is an open-source framework for building interactive applications using Large Language Models (LLMs). It’s a tool that helps LLMs connect with other sources of information and lets them talk to the world around them. One important concept in such frameworks is the Chain. Let’s take a look at this concept.
What are Chains?
Chains are advanced tools in this framework that combine LLMs with other tools to perform more complicated tasks. Specifically, chains are interfaces that use a sequence of LLMs along with other tools, such as SQL databases, API calls, bash operators, or math calculators, to complete a complex job. An example could be our application receiving input from a user and passing it to our LLM model; then, the LLM calls an API. The API responds to the LLM, and the LLM takes the response to perform another task, and so on. As you can see, it is a chain of inputs and outputs where, in many parts of this sequence, we have LLM models handling the situation.
Now it’s time to get our hands dirty and start coding a simple LLM-backed application. For this application, we are going to make…
