Build a smart Slack bot with language models

Build a smart Slack bot with language models
A diagram showing how the summarization slack bot is architected.

Ever wanted to build an intelligent Slack bot? Check out our guide that walks you through building a Slack bot powered by language models. In this case, the bot receives a URL to an arxiv paper, and it returns a brief summary of the abstract.

Guide: https://docs.cohere.ai/slack-app-summarize-example
Starter Code: https://github.com/cohere-samples/cohere-slack-starter-app/

There are many additional ways to inject intelligence into a Slack or Discord bot.

Want a bot that is triggered when somebody asks a question in the channel? That’s done by classifying messages to determine if they’re a question or not:

Classifying messages as questions or statements

How would you want the bot to respond? One way is to retrieve the most similar question from the FAQ, and post its answer as a response to the question:

Auto responding to a question using semantically similar question and response

Want a bot that is triggered when a user posts an abusive message? That is also done by another classifier that determines if a message is abusive or neutral. How should this moderation bot respond to an abusive post? Perhaps it notifies an admin of the offending post.

If you need any help figuring out the best way to use Cohere for your bot, we’d love to hear from you! You can post in our community forum.