Phone call with Steve after our AI meeting. Going over tons of ideas and options... Exciting times.

- On the 3rd party solutions page, add in the other API's that Nxtlinq may need to access their data. Do this all behind the scenes so that a user doesn't have to go in and turn any of those settings on. That way, all of these API sockets will be monitored and controlled via the 3rd party solutions channel.

- We can handle the user level permissions - meaning allowing a user to turn the AI agent on/off at will. The corporation may turn it on, but not all users may want it. Basically, the corp level will be the top level and the user level will be on an individual basis level.

- Earlier today, we were talking about subsets of aliases - Steve had the idea... What about multiple levels deep? Aliases of aliases of aliases... We can totally keep it all strait but what if a user does something all the time... What if the AI agent was able to make its own aliases for common tasks. Talking about predictiveness and interpreting input and intent - We even talked about slang or certain terms that may be used in different regions.

- "Adi go" - command prompt - We already have "Adi" as our little blue dog avatar (helper or mascot). Maybe using "Adi" as a command prompt. This would be similar to how smart phone users can talk with Siri or Alexa. Ours would be "Adi". Idea from Steve.

- Sell what we have

- The value of hyperlinks - mixing command line, Gui interfaces, web, and API - we are built for this

After Steve and I got finished, I was wrapping up for the day. I was really grateful that everything was going good, meeting wise, direction, etc. I was saying a little prayer and had a number of thoughts pop into my head. Call it inspiration, direction, or whatever. I'll take it and I love it! Here are my notes:

- On the user set aliases (making their own prompts), for the AI quick search prompts, what if we used a "#" (pound or hashtag sign) for all user specific prompts. That would not compete with any of our existing prompts. For example: #xyz or #something or #somethingelse... - we could even reclaim certain existing prompts by just add the hashtag in front of it. Another example may be... The prompt or keyword for clocking in/out has an alias of "my time" - what if a user wanted to use that for his/her own prompt for project time tracking. They could create "#my time" or "#time" and it would totally work. Even though both of those values are normal keywords by themselves. It could be really cool. Imagin a hashtag and at least one letter, more if needed. Super simple for quick speed.

- Be able to build and show the prompts and what they are tied to. Simple user interface to setup and maintain their own AI quick search prompts. Very similar to the icon payee top links in the snow owl theme, but even more, and easier to get to... they would be clickable and usable in the AI quick search text fields. Anywhere in the system... That would be super cool! Even better than the icon payee top links (existing link building or special button feature inside of adilas right now).

- What if we used the "go mode" page (new AI quick search prompt page) to show their own prompts or even a fracture type interface - simple tiles and buttons - (small sample image)

- When creating your own buttons, be able to name them, color them, set an icon, a link (advanced - they have to know the URL or web link) or a simple quick search prompt. Maybe even tie the user created prompts and buttons together or make it a dual-purpose database action. Some of the buttons may need to open up a new window and some (most) will just jump to that new page. Just some ideas.

- What about being able to tie multiple prompts together in a mini list, process, or mini prompt recipe? That would be super cool. Imagine something like this: start cart, choose customer, show x or add y to the cart, simple cash checkout - simple English or natural language instructions and the system would help them walk through that process... kinda hold their hand. It won't just do all of them at once, but it will know what the process should be. We could easily create those one-to-many relationships and virtual steps or phases of the process. Just for fun, it makes me think of a data assembly line or assembly line for data. Cool stuff!

- Along this multi-prompt line... We already allow for duplicating invoices to the cart. What if we used a template type scenario and setup a master invoice and then were able to duplicate the exact thing in one click. If changes are needed, they could modify it, but cut out all of the searching and such. What about a partial duplication process... once again a mini recipe for certain steps or data assembly line controls.

- We have a page inside the current system called "my favorites". It currently has a number of buttons on that page. This was one of my todo list notes from a couple of days ago... It totally fits here (see below).

todo: Get back to the ideas of the my favorites page inside of adilas... it got taken over and a bunch of hardcoded buttons put on it. But it does have some potential. See this help file - help.cfm?id=418&pwd=favorites - or visit this page - top_secret/secure/my_favorites.cfm. Added this note on 10/14/25 - it would be so cool to let the users choose anything that they want in the system and then organize it and name it. Cody Apedaile started a draw your own interface (but got pulled off of it for budget reasons). Anyways, the goal is still there to help create your own interface and organize your own buttons as you see fit.

-  Super simple, using old school tech with a modern or new school flare... it will work

- What about helping with memory... We could help hold small comments, save/hold this or that, or even helping to string things together... Maybe help them with their memory... It won't tax the server very hard (store simple things in session - per login). Imagine small AI quick search prompts like clear memory, save to memory, view memory, etc. Super simple stuff. Use JSON storage if saving memory in a more long-term fashion. Simple interface to get things in/out of memory and even use the AI quick search field to help. That would be cool.

- This may not be ready yet - but what if we record and then help them go from one thing to the next to the next... simple session memory with a display... If we know a series of AI quick search prompts... It wouldn't be that hard to display something with a next button, help them step through the process.

Meeting with the Nxtlinq guys. Going over progress. Then talking about multi-tenant options (settings, domains, corps, keys, secrets). We setup some small dates and goals to get an MVP up and live. Other small talks on different subjects. Fairly quick meeting.

Working on help files and AI quick search prompts. Got on a meeting with Steve. We went over a bunch of things. Things move so fast sometimes. We were talking about where we are heading, MVP levels for the AI stuff, questions about making the AI agent multi-tenant and able to handle different domains, corp id's, keys, secrets, etc. We spent a little time, and I was showing Steve what this can do (current AI quick search prompts), small dog and pony show, bouncing around, looking things up, showing some future plans, and making some decisions.

Here are a few other notes from our meeting:

- We already have so many AI quick prompts and keywords. What if allowed our users to setup and create their own aliases or mini prompts? I'm kinda cheating here... It is now hours later (the next morning, when I am recording this). But, we came up with some good ideas and had some inspiration on this topic later that evening (last night - click to see EOT # 12392 for more info).

- The value of simple text look-ups vs straight AI or Id based look-ups. You have to start somewhere, but ideally, you get it tuned into just normal natural language stuff. The computer wants the Id's but normal people want to speak their normal language. Find that happy middle ground or even favor the human side... a bit...

- If we go more to natural language prompts, if a decision can't be made, provide roll-over fail safes - options if things don't work out.

- We are seeing at least four levels of permissions right now. They are corp level (world level), group/category level, page level, and user level.

- In order to make the AI agent into a multi-tenant agent, it will need a way to dynamically deal with domain names, sub domains, corp keys, corp id's, and users. I will try to put some things together on this topic.

- Steve and I were talking about multi-threading and other threading questions... keeping things straight from server to server and between users and users. In this case, for AI, between serves and AI agents and AI agents and users within certain corporations. 

Working with Russell. Talking with Chuck for a few minutes and catching up. They, Russell and Chuck, were on a previous meeting. After Chuck left, Russell and I did a little flashback and were talking about CSS and modern tech. Sometimes it is hard to try to keep up with all of the changes. The old basics are still super important. A developer needs html, JavaScript, CSS, SQL, and some kind of backend technology.

As we were working today, we were talking about exposure, in development, leads to some new ideas and knowing what is possible. Not all exposure is good, but certain kinds of exposure can really help you imagine what is possible. Once you know that, it isn't super hard to then go try to figure it out. Granted, the term "isn't super hard" is relative...

We did some training on the VS code interface, some speed tips, some Git tips (code repository and branching stuff), and then spent some time tweaking a little bit of adilas code. We changed some older hardcoded values (in the CSS) to use flex box instead of older direct positioning. Good little training session.

Working with Shannon. Small recap on what we were working with last time. Talking about templates and going from static to dynamic templates. We briefly talked about when I started creating web pages, each one was a static page (copy and paste) from template of sorts. Then once we added database connections, we really were able to use a dynamic, data-driven template page. That was a huge building block.

We then switched over to adding new content for the fracture plans. Working on the modular architecture sections. At the end of our session, we were talking about old CSS and more modern CSS stuff. We even looked at some examples (some of our old stuff). Crazy how fast things are changing and what it takes to keep up in this high-paced world.

As a side note, we were working on some new documentation and using AI to help us refine some of the content. It started to joke around with us a bit. That was kinda fun. It added a couple of sarcastic prompts. It made me laugh and Shannon and I had a small discussion on "joke settings" and "changing levels" on those settings.

Working on help files for sub inventory attributes and templates. Also started to work on help files for parent attributes and high-level templates.

Small data clean-up for a client. I had some old code but had to tweak it a bit. The new request was specific for just certain changes.