🧩 How Can I Enhance User Engagement in Tracardi with a genAI Plugin?

This comment was shared in the Tracardi Slack #general channel:

Hey everyone! I’m excited to share my journey in developing a couple of new Tracardi plugins aimed at using genAI to help with classifying user interests and intents.

The response: "That sounds intriguing. Can you tell us a bit about how your intended plugins work?"

I replied: "Absolutely! The goal is to automatically classify web pages, tag them with relevant interests, and match these interests to the users visiting the site. For example, if a website focuses on a specific topic, our intention is to tailor engagement strategies to suit that audience."

The response: "Great concept! Have you thought about the actual functionality?"

I answered: "Yes! I’m planning to develop three plugins, but I’ll hold off on the details for now."

During the discussion, we identified several architectural considerations that needed to be addressed, such as asynchronous processing, scheduling and the overall approach.

As you may have gathered by now, I’m the one who initiated this conversation and am interested in writing these plugins.

So, where did we land with this?

Ultimately, only two plugins were necessary. Thanks to Tracardi's open-source nature, all the underlying processes are open for exploration and can be leveraged.

To learn more about the new plugins, be sure to check out the sitemap and data write plugins.

And what about managing genAI functionality? For that, we’re using the remote API call plugin.

The initial genAI plan was to leverage the always-awesome LangChain packages as a Tracardi plugin for genAI - the gotcha as to why that type of plugin didn't go ahead was the asynchronous processing considerations.

I’m confident that over time, there will be further refinements to enhance all of this.

Throughout this process, I have truly appreciated all the feedback and support from the community.

Join the conversation.

Have you considered making your own Tracardi plugin?