I feel kind of like I've been in a cave for the last 90 days, so it's really nice to be able to take a breath and look around. We've been deep in the trenches, and a lot has happened. Lots of lessons were learned, new appraiser tools were built, and the community has grown significantly. I want to give a huge welcome to all of the new members who have joined us recently! It's an exciting time to be part of this group, and I'm thrilled to have you here as we continue to push the boundaries of what's possible in our industry. Let's get into it.

Lots were learned in Project Prometheus. Specifically, we learned about OpenClaw. Though the first steps for most people into an agentic AI system are pretty primitive, was it unwise to start just a few weeks after it came out? I don't think so, because the point of the experiment was never necessarily to be successful, as much as it was to have an experiment to learn things, to grow, to develop tools, and all of that was accomplished.

The reason we called it Project Prometheus is because Prometheus was the Titan that brought fire down from the gods to the humans. In a way, I wanted to bring the fire of AI to appraisers.

The other side of the coin was possibly this will be Project Icarus. Icarus, of course, was the man who wanted to fly like the gods. He created himself some wings, put them together with wax, flew too close to the Sun, and the Sun melted the wings and he fell into the ocean and died. I don't think we fell into the ocean and died. In fact, I would say we were 75% successful. Let me explain.

The project showed that much of the back end for appraisal work can be automated. We were able to take an engagement letter, extract the pertinent information from that engagement letter, put it in a spreadsheet, go out to county websites and other public data sources, and get owner names, parcel numbers, legal descriptions, plat maps, GIS data, Zestimates, assessed values, taxes, and sometimes even sketches. Anything that could be gathered in the pre-research process mostly could be found and put in a work file.

Next, that data was put into the appraisal report. The AI was able to distinguish what type of appraisal this was, open up the correct template, fill out the correct areas with some difficulty, and deliver that zap to Pandora, our anonymous appraiser.

You might remember that our anonymous appraiser had a part-time assistant when she began. The truth is she was never able to fully bench that part-time assistant. This is good news and bad news. The bad news is we are not quite there with AI. The good news is we are not quite there with AI. In other words, those of you that have assistants are not going to have to give them a pink slip anytime soon.

What did happen, however, was her assistant was able to in some ways supervise the AI. I believe that's the direction that AI is going to take us as appraisers, but we are also going to be reliant upon our staff to become the directors of the AI.

Thank you for sticking with me and the support through this project. It's been a lot of fun. I have to admit I'm glad it's over, but a lot was learned along the way. By the way, all of the tools that we developed are available on GitHub which you can find a link to inside the membership area. Enjoy!

This home defies gravity

Open Claw came out in January of 2026. It became popular in February of 2026, and by mid-February, I was using it and convinced enough by it that on March 1st we began Project Prometheus.

Here's what I learned along the way: Open Claw is a harness. It allows you to be able to build out systems in a somewhat secure and controlled environment. Now, I say somewhat because there are some anomalies there and some gates and barriers that need to be set up. Probably the biggest lesson I learned was the difference between deterministic and non-deterministic agentic AI.

It looks like this: there are certain things that can be built out, similar to software, that just work every single time. Then there's the non-deterministic part of AI where it kind of acts like a human. It has a brain and makes decisions. It's somewhere in the middle that you find success, because if you make everything deterministic, you might as well write software. If you make everything non-deterministic, you might as well let a rogue player do whatever he or she wants. Neither is ideal.

Once I started to employ Claude Code to help fix and write the programs for Open Claw, things improved dramatically. They never got perfect because I think we still have a ways to go. There are other agentic AI harnesses out there, and moving forward I'll probably try some of them out. Hermes is one that is pretty popular. We'll see how that goes. Frankly, I've got friends that are simply just using Claude Code to build out programs that sit on their servers. That also is quite appealing.

In the end, you should be using something. The way that you learn to harness the power of AI is by getting in and getting your hands dirty.

Point #1: Edge Gallery

Edge Gallery is the ability to have an LLM on your phone or personal device that can be run internally. I downloaded it using Gemma 4 so that I could use it on the plane in airplane mode. You do not have to be connected to the internet, and you have a fully functional, powerful AI at your fingertips.

Point #2: Wise

I used to use Xoom for payments overseas, but got banned a few years ago. Wise has been my backup. I've not been super impressed (had some hiccups), but it does work and it's similar in cost. If you have assistance in the Philippines, you might want to check it out.

I didn't talk about it much publicly, but I was approached by two and a half investors/partners/entrepreneurs recently with big projects regarding AI and appraisers. I'll be honest with you, there were stars in my eyes as I contemplated the opportunities. These were situations where a lot was at risk, but a lot was to be gained if it succeeded. It also meant a great deal of time over the next several (not months, but) years.

Though I enjoy building things and I enjoy starting new businesses, I ultimately decided to go a different direction. Let me tell you why. For many years, in fact as long as I have been coaching, I've been teaching about missions and visions and what brings you true joy. In fact, that's what The Appraiser Coach is all about: helping people understand their mission in life and realizing that appraising head down 60 to 70 hours a week is not it. You probably are not the 0.1% of people who appraise properties because they just love it so much they can't stand to do anything else. Instead, it is a means to an end. It puts money in your wallet and food on your table, but finding that balance is essential.

Ultimately, I made a decision to go in a different direction because there was a week where I spent all day working on the farm and another day with my grandchild. It was that week that made me realize that that's what I ultimately love. I've got a comfortable life. I love coaching. I'll continue to do that, but ultimately it's about slowing down and spending more time with the people and the things that I love.

Now, go create some value!