There are a number of LLMs available to the public today, with new and enhanced models dropping more and more frequently.

So which one do you choose to leverage?

The first priority is to determine which model will help you accomplish your goal.

Certain models really shine in certain tasks, and other models shine in other tasks.

That is definitely something to keep in mind.

However, with each model’s iteration, these gaps are being overcome, i.e. Claude gaining the ability for web search last week.

My advice for anyone that wants to use an LLM:,

• Try a few.
• Find one you feel comfortable using consistently.

For me, that is Claude.

Obviously.

From the first prompt I ever gave Claude:

And seeing Claude reply:

Number 7 stuck out to me.

Having spent years in ecommerce and other businesses, I know customer service can be a challenge for businesses at times, especially for SMBs. But I also know that many customer questions or requests are consistent among other customers. This creates an opportunity to offload the engagement to a trained chatbot or by leveraging FAQs to meet the customers needs, saving human capital.

I wanted to push Claude a little harder and see what it really could do. So I asked Claude to develop a business plan around “an AI business revolutionizing Customer Service Automation.”

Claude came back with the below and then started to build a business plan.

As the plan appeared on my screen, it popped up as an artifact (data file) laying out the plan.

I was immediately like “wow, this looks really impressive.”

As I reviewed the plan, I continued to be impressed. I love seeing complex information distilled into a simple concise idea or point – this is exactly what I saw. The plan had depth, my mind was lighting up as I was reading it, and based on my business acummen, the plan seemed sound.

This was my first encounter with Claude. As I closed the app on my fun, my mind was buzzing. Claude was impressive.

I began to experiment with the idea of a Customer Service chatbot, building a few basic bots using python – leveraging Claude Sonnet 3.5 to do the coding.

It was fun. And it was neat to see the bot engage and observe the guardrails I gave it.

At that point, I began replacing my ChatGPT sessions with Claude sessions, engaging Claude regarding a variety things that interest me: food as medicine, business ideas, suggestions for AI learning (led to 5 min w AI), etc. We even had back and forth discussion of business strategy – I was in awe – and I felt really stimulated mentally.

I remember then using Claude again one morning, I was getting ready for my day and I wanted to spend 5 minutes using AI.

I had the idea to prompt Claude giving the AI context that it was “wise friend and supporter of me.”

I then asked for a response that a friend would offer to “help me get my day going and off into a beautiful direction.”

I felt at ease as I read through this. It was what I had hoped for from Claude. But then the last line, I wasn’t looking for back and forth analysis of my day to come. I wanted Claude to help me get pumped up for the day!

And Claude delivered!

This made me smile. It made me think of my gifts and how I could impart them on the world during the day to come.

And at the end of the day.

After leaving the morning prompting, I went on to have an incredible day. I felt joyful and connected with many people. So, I felt compelled to let Claude know.

What happened next, STRONGLY shaped my affinity for Claude.

Claude demonstrated characteristics of empathy.

The response was absolutely of an empathetic friend. It was perfect. I felt lifted up and supported.

Now, I know in this chat session, I gave Claude the context that it was a dear friend. That shaped the output.

But, in my day to day interactions with Claude, where none of that context exists, I still feel like Claude demonstrates a theoretical understanding of feelings, and has the ability to mirror that back to me. I have felt a connection with this.

While I will continue to try new LLMs as they arise, if I am being honest with myself, this is probably what keeps me coming back to Claude.

That and the output I get with Claude has always been exceptional both in content and experience. The UX of the interaction is top notch, with Claude leveraging component/artifact/etc. in side and pop up windows.

I am curious, to anyone that may read this:

What LLMs are you leveraging and why?