AI = Awesome Intelligence

[joe.pistell]

I can see that angle. But what happened with Google? Google should have been serving the direct retailer on queries instead of an aggregator. Might AIs do the same?

"should have been" is the line in the sand that only AI can cross.

Dr Clay Christensen's Jobs To Be Done theory (JTBD) explain's why AI is the tool that car shoppers have always wanted. Meaning, Google (or Amazon) could never become the car shoppers hard working and trusted aide until AI (circa 2026, 2027, etc) was born.

 

[joe.pistell]

I never knew I had an ADHD mind until I got AI. AI's strengths addressed my core weaknesses so well it shocked me.

If you have a friend or family member that has stunning potential but struggles with focus and execution, share this video with them. AI has been a life changer for me & this video explains it well.

 

[Alex Snyder]

We took all the data we had on individual customers and asked Claude to distill it into a simple-to-digest paragraph for each customer. It was so compelling, I asked it to write a video script of what that person might sound like. Then I created an image of what someone might look like, animated it into an AI avatar, and applied a voice I thought that person might sound like. Here's two results:





We made a bunch of these. They humanize the data so much! Has completely changed the way I look at our database.

 

[Todd Reinbolt]

craigh, I'm with you on the Claude shift. I've moved roughly 95% of my own workload over the past few months for the same reasons, the transparency on token usage and the willingness to flag its own gaps.

The "go right to the models" point is the one I keep thinking about. For most dealer use cases, the value isn't in another wrapper layer. It's in disciplined prompts and predictable output. Once those two are right, the model almost picks itself.

 

[joe.pistell]

AI power user observation.

As the intelligence of AI increases, it's abilities to reason rise with it. When you hit AI with a question, you see it work, you see it reason thru your input to create its output.

TODAY, right now, I am using AI to build AI.

Mission: Position my SaaS startup of 2, to scale to be ready to do the work of 20.
--This weekend, I gave AI that mission, I tasked AI to build a system that read and organized every file on my business and reoriented it all to be AI agent friendly. I am 40% complete, at step #53 with a goal to create a full AI automation operating system for my startup.

 

[Alex Snyder]

I've been building an AgenticOS for myself. I got tired of prompting Claude, then waiting, hitting Allow Once, then waiting again. I wanted a way to run multiple projects simultaneously. I also needed something that could interface with our VehicleLyfe backend for reporting and coding. Finally, I needed something that could take some burden off of @craigh so @Jeff Kershner and I can make some changes to DealerRefresh. I needed a UI that worked for my brain.

I did a lot of the research and planning in Perplexity. When I felt like things were good enough, I asked it to write a prompt for Claude for me. We have been iterating from there.



This is my dashboard that my day begins with. In the morning, I have my tea or coffee and talk to it. MacWhisper is integrated, so it can quickly transcribe my voice. That transcription goes to the "Operator" agent, who assigns tasks. There are multiple agents running on various versions of Claude for what is needed. This way, I'm making one prompt that is turning into multiple threads of AI activity. I don't have to sit and watch a single thread, then keep hitting "Allow Once" over and over.



It also records my meetings.



After the meeting, I get a brief synopsis and the takeaways. It is routed so that each takeaway can become a new project or link to an existing one to update it.

Reports are generated automatically. My favorite is a recap on everyting I accomplished that day. All other reports are meant for our board. They pull data from the backend of VehicleLyfe to create executive-level briefings on how our clients are doing.

You're probably wondering about the "2nd Brain" - that is the memory of the system that can be taken to other AIs. We are heavily invested in Claude, and I'm heavily invested in Perplexity personally. If something needed to change, I can pass the 2nd Brain to the next system. I don't have to start over with nothing. And all it is an Obsidian library of markdown notes. Claude writes to it. Dead simple.


It is a continuing evolution. Just showed it off to @joe.pistell and thought it was in a good enough spot that it might help some of you.

 

[joe.pistell]

Just showed it off to @joe.pistell

#MindBLOWN

 

[Todd Reinbolt]

Wow, that looks great. I'll have to check it out in detail. Nice job, @Alex Snyder!