Large language models open a whole new door to how people can interact with the web — but what does that mean for your business? And how can AI models boost your brand's signal and enhance your website’s customer journey?

‍

What is a large language model (LLM)?

A type of artificial intelligence (AI), LLMs are basically algorithms that take a ton of data (think billions of words), mix it with sophisticated deep learning, and generate — and predict — new content. You can feed a query — a question — into the LLM, and it will give you a response derived from that large number of words, making it an excellent tool for weeding through lots of information and getting you an answer quickly.

Typical LLM applications include:

• Summarizing, translating, and predicting text
• Answering questions
• Classifying and categorizing content
• Generating code
• Making (better) chatbots

‍

LLMs aren’t born — they’re trained (and can be branded) 

Large language models don’t just show up one day, FedExed from the stork. They must be “trained” on the data that will serve as their knowledge base. Language is complex, after all, so the model needs to sort through giant swaths of text and identify things like basic patterns, nuances in phrasing, different dialects, etc., to make sense when they return an answer. Next, the model can zero in on a more specific dataset that will be relevant to its exact application.

What this means is that you can train an LLM on the specifics of your industry, company, and customer base, creating a branded LLM. Once you’ve trained the model to understand your business through the information you present, the way it’s composed, and the voice and tone it uses, it can act as an agent of that business to address customer questions and needs as they come up. 

In other words, you can shape highly personalized and customized experiences for your customers thanks to the wonders of natural language processing (NLP).

‍

From chatbot to virtual guide

Chatbots don’t have the best reputation. We’ve all had that experience: landing on a webpage and foolishly consulting the pop-up window in the bottom right-hand corner only to lose ourselves in a vortex of unhelpful, pre-scripted responses. Clippy could do better in most cases.

A branded LLM changes all that. Suppose a GPT (short for generative pre-trained transformer, or rather, the model used to build chatbots) has the full context of your site, as well as insight into how users landed there. In that case, it can anticipate the user’s needs and deliver the right answers immediately upon arrival.

Here’s an example: Let’s say someone searches for information about specialized law services and visits your page. You could greet them with an article on that topic, then follow it up with an executive summary of your company’s unique approach to your niche area of law, as well as an abstract of other areas that may be of interest to that user based on their search. And that’s great, but the experience wouldn’t be complete without a thoughtful follow-up. So you can also round out the interaction by inquiring as to whether the user’s question has been answered and continue to converse and offer up information from there, all using the terminology and voice of your specific brand. The interaction is conversational, highly customized — and natural. 

But that’s not all. Maybe the user is a prospect scouting out your capabilities. In this case, they could land on your site and ask about your experience with, say, a particular law in your state. Your branded LLM would then pull up summaries of, and links to, relevant articles, case studies, and service pages that feature your experience with that thing. 

Now you’ve got a customer liaison and sales rep all wrapped up in one tidy LLM. 

‍

Hyper-personalization via branded LLM is the future

In the above examples, users can either query the site and dig into more detail as they please or receive personalized recommendations as they continue to interact with your site. The hyper-personalization opened up by branded LLMs fosters more authentic interactions between your brand and your customer base. Your site visitors also get a more tailored user experience with a designated, informed guide that can make sure they’re getting what they need — and fast. This all happens rather than asking the user to take the traditional route, to either “search” for a piece of information (and be served up a warren of other links, but no “answer”), or click through the typical main navigation items until they find what they’re looking for.

See the potential? 

‍

Did you enjoy this article? Read more like it on the Edgar Allan blog.

Check out more articles from Edgar Allan on AI:

AI: Friend or Foe?

Creating digital artwork using custom code and AI with Slater 

Slater early-release: An open letter to the community

Say hi to Edgar Allan on LinkedIn, X (Twitter), YouTube, and TikTok. We’d love to hear from you!

Take a look at the work Edgar Allan has done by checking out our case studies.