Background Advisory Ventures Dispatches Contact
This WordPress Chatbot Plugin Uses Your Own Content
All dispatches

This WordPress Chatbot Plugin Uses Your Own Content

First shared in my AI club, now Frontier Table, February 27, 2023.

This is pretty cool, I think — and a bit technical, perhaps. I’ve been playing around with a WordPress plugin that has OpenAI read through all of your website content and then generate questions and answers based on it. That alone is useful in its own right, if you know SEO. Then you upload the newly generated data to OpenAI and update its models with what’s on your site.

Why does this matter? Because the chatbot can then go live on your site and not only answer whatever GPT knows, but hold custom conversations about your products, your services, your ideas.

Some context on what’s actually happening, as far as I understand it. The plugin is AI Engine, by Jordy Meow of Meow Apps, and it’s young — it only arrived in the WordPress plugin directory on December 27, and the fine-tuning guide linked above went up on January 19. Two months in the directory is a fair measure of how new this corner of the web is.

The fine-tuning step is OpenAI’s, not the plugin’s. OpenAI’s documentation says fine-tuning “is currently only available for the following base models: davinci, curie, babbage, and ada” — the original GPT-3 base models, which the same docs note “do not have any instruction following training” of the kind text-davinci-003 has. So what you get back is an older base model, without ChatGPT’s instruction training, that has been taught your material — not ChatGPT with your catalog bolted on. Whether that trade is worth it depends on what your visitors actually ask.

Still, think about what a small business gets: a question-and-answer layer, generated from pages it already wrote, standing in front of a model that also knows the rest of the world.

Very interesting — especially this line:

“In any case, the combination of fine-tuning and context window expansion, especially as supported by the rapidly evolving LLM tools ecosystem, mean customers will be able to achieve human performance & reliability — or better! — on many economically valuable tasks — in 2023!”

That is a large claim for a calendar year with ten months left in it.

Where this started

Most of these dispatches start as posts in the Frontier Table, a group of about 100 builders, founders and operators trading what's actually working with AI.

Robertson Price

Robertson Price

Serial entrepreneur who has built and exited multiple internet companies over 25 years — from search (iWon.com, $750M acquisition) to content networks (32M monthly visitors) to e-commerce (Rebates.com). He now builds enterprise AI infrastructure at Ragu.AI.