In this lesson, you’ll learn how to make your Colang 2.0 bots more conversational and adaptive by integrating Large Language Model (LLM) capabilities using the llm module. We’ll see how LLM flows can help your bot handle a wider range of user messages and generate more natural replies. By the end, you’ll understand the difference between strict, rule-based flows and the dynamic flexibility that LLMs bring to Colang.

Without LLMs, your bot can only respond to the exact phrases you define. If a user says something unexpected, the bot won’t know how to reply.

For example:

import core

flow user asked for help
  user said "help" or user said "support"

flow bot provide help
  bot say "How can I assist you today?"

flow handling help
  user asked for help
  bot provide help

flow main
  activate handling help

Sample conversation:

> help
How can I assist you today?

> I need assistance with my account
(no response)

Here, the bot only responds to "help" or "support". Any other input is ignored. This isn’t much different from using traditional string comparison if statements. To move beyond these limitations and make your bot truly conversational, you can bring in the power of LLMs.

To enable LLM-powered features in your Colang project, import the llm module along with the core module at the top of your .co file:

import core
import llm

• import core: Loads the essential flows and actions from the Colang Standard Library.
• import llm: Adds LLM-driven flows and actions to your project.

By importing the llm module and activating the llm continuation flow, your bot can handle a much wider variety of user inputs. The llm continuation flow lets the LLM generate responses for anything not covered by your defined flows.

import core
import llm

flow user asked for help
  user said "help" or user said "support"

flow bot provide help
  bot say "How can I assist you today?"

flow handling help
  user asked for help
  bot provide help

flow main
  activate llm continuation
  activate handling help

Sample conversation:

> can you give me some advice
Sure! What kind of advice are you looking for?

> help
How can I assist you today?

> tell me a fun fact about space
Did you know that a day on Venus is longer than its year?

• The first and third inputs are handled by the LLM.
• The second input matches the help flow.

The llm module also enables more conversational and varied responses using the bot say something like $text action. This action prompts the LLM to generate a response similar to your prompt, but in more natural language.

Example:

import core
import llm

flow main
  user said something
  bot say something like "What brings you here today?"

Sample conversation:

> hello
Hi! What brings you to chat with me today?

Here, the LLM creates a friendly, context-aware greeting, rather than repeating the exact phrase.

By importing and activating the llm module in Colang 2.0, you can transform your bot from a rigid, rule-based system into a flexible, intelligent conversational agent. The llm continuation flow and bot say something like action allow your bot to handle a wide range of user inputs and respond in a more natural, engaging way.

In the next practice, you’ll get to experiment with LLM flows and see how they can improve your own Colang projects.