Phone call with Steve. We had to play a little bit of phone tag, due to connections. Once we got connected, we had a good conversation. The main things that we were talking about were dealing with removing friction and rub on the client side of the equation. Making things easier and easier. We talked briefly spoke about some current projects like Metrc updates, merchant processing, and CardPointe and Clover server issues and threading. Here are some of my other notes:

- One of Steve's biggest goals is to get AI fully integrated into the adilas system. Almost to the point of a simple chat-based system that could handle and process natural language and then be able to help the user do anything inside of the adilas system. Not just navigation, education, training, and consulting, but actual physical tasks. Instead of clicking buttons, navigating, and running reports, you could just talk with the system, and it would be able to help you. That is one of his main goals. It may take some steps to get there, but a cool vision.

- Along with the idea above, of a system that could do anything (AI super system), he was talking about a flow process like this... From left to right... Something comes in, it could be a bill, receiving inventory, doing a sales transaction, whatever... The middle would be the chat window and super simple interface (almost nothing structured). You tell the system what you want to do and it helps you do it. If it needs more information, it would just prompt you and help hold your hand. Then on the right, or the output window, you could get your results and/or confirmation that the task was completed. The results don't even have to be reports that we have programmed. It, the output and/or results, could flex, based on what was asked and/or being worked on.

- I was expressing that I liked the idea, and I could see how it could really help in some situations. This is just me, but I don't think that every person wants that. Yes, it could be an awesome option and could speed certain things up. However, if it was a repeated process, that would be a pain to keep telling it what to do. It might take longer to explain it than it would to click two quick buttons (from a pre-built interface). There are a lot of assumptions being made. One, it (AI) would have to have a super deep knowledge of the system and all of its possibilities. Someone would have to help set things up. We would need to record those instructions in order to repeat those processes. Things change as time goes by, someone would have to be able to edit things, etc. We would also want it to keep learning on the fly. Technically, each person, even across the same business, would have specific needs.

- Without being a "Debbie Downer", I could see something like this being possible as a phase 4 or 5 of working with the AI agents. Phase 1 would be integration and doing simple existing navigation and use of tools. Phase 2 would include education and training on existing pieces. Phase 3 could be where we start letting the AI agent have access to raw data through API's and special AI tooling. Phase 4 we would have to introduce ways of creating some kind of assembly or package for the AI agent to follow (recipe/build type mentality). Then maybe a phase 5 where the system is trained and enabled to help with all kinds of stuff. This could be super deep. Say a person has a picture of something. Do they want to enter it into the system, store info, check something off, etc. It's a little too open right now. This is my take on it... but it would have to be a phase 5 ish type thing.

- As Steve and I were talking, we were talking about AI, bots, robots, agents, etc. Steve was saying that they are all combining into just the AI (artificial intelligence). The words bots, agents, etc. may go away. They are somewhat merging (lines are blurring). There is some mixing going on.

- I loved his ideas on helping our clients get rid of heavy learning curves, manuals, processes, etc. Decrease the friction and the rub. I love that. It just takes time and money and development (and a plan) to get there.

- We talked about the fact that some (most) people don't want to watch a hundred videos or read a huge, big user guide. They want it to be easy.

- Some of what we were talking about might even be the next steps (future) beyond the value add-on core model and/or the fracture UI buildout.

- Trying to listen to what our clients are saying and telling us. That is important. This came from Russell, long ago - Our clients want something that is easy, powerful, and looks nice. If they can get those three things, it will sell.

- I wanted to record these ideas... part of the idea farming stuff that we are doing. We record things, plant them in the ground (sit on them or think about them), and then finally roll around to see if we can make something out of the ideas. Ideally, we get more and more activity on certain things and that helps us know what is being asked for and/or required. Fun process, but it does take time and feedback loops (over and over again).

- For me, when I got home, I scribbled down some notes and drew some funnels, mixing of tools, and even possible stacked or reverse funnels. All of this deals with getting something, mixing it together, using possible tools, and then getting an output (of some sort). Sometimes, once you get an output or result, you have to remix it or send it through another funnel to get what you really want. Break it down, transform it, summarize it, expand it, or whatever... some kind of action to either pull, mix, create, and/or alter something. As a note for me, I have some old graphics that I was working on back when I was doing some developer intern training. These guys had to produce something (desired output or a result), but the starting spot varied. They could use tools and then produce the desired output. I might tie back into that concept.

- Dealing with the concepts for the developers (listed above - inputs, funnels, mixing, blending, and getting outputs). I found some of the old entries. They were in 2/14/2015 and then again on 3/6/2015. If you want, check out these entries and look at the image galleries for expanded visuals. I also added some of the old handwritten notes and scans to this element of time. These entries also include some concepting on the 3D calendar. All playing through around the same time. Kinda fun. Enjoy!

Recording notes from Thursday. Asking ChatGPT how to do more effective branching and knowledge sharing. See attached for a PDF copy of the chat results. This may not be super important for the full adilas story, but it helps me get some ideas and direction on how to link things together. Here is the full chat link, if needed:

https://chatgpt.com/share/69c1afc8-bd60-8007-b1b1-e0fc774ff9be - chat branching and sharing help

Working with Hamid. Showing him what I am doing with AI and how the flow goes. We started a new chat session with ChatGPT. Hamid and I worked on it together for the first two hours. He then had to leave. I then finished it up. This was a wonderful session going over the 3D data assembly line and how it ties into world building. Lots of great explanations, discovery stuff, and explaining "the gap" between operations and accounting. See attached for numerous resources. The full chat log, with ChatGPT will be listed below. This was an awesome session.

https://chatgpt.com/share/69bcdbd3-9560-8007-ad64-15d36d272a43

Recording notes and doing some organizing and research. Everything is moving so fast. Reading up on some tools that we use. Adila university - entry # 4029. Basically, a ladder of possible tools that all of us use until something breaks. Then we have to virtually go up the ladder to the next best tool. Fun entry.

Working with Shannon. We jumped into some older documentation for the adilas user guide. We were in the financials and accounting section, section 11. We read over the first few entries and then got to the section new school accounting. It was mostly done, but not fully done. We started working and Shannon was asking what do you really want here? We did some quick brainstorming and then I showed Shannon a small graphic from years back. See attached for a copy of that graphic. She then recommended that we use voice to text, explain what we wanted to, and then see if ChatGPT could interpret our babblings. It did a great job! Fun chat session with ChatGPT on some accounting concepts. There are some hidden gems in the comments, after it interpreted what I was saying. Good stuff. As a note, there is a web link to the chat session as well as PDF version attached to this element of time.

https://chatgpt.com/share/69bc3d35-4230-8007-a52d-e6b549493127