Progress on local Hal talking over the internet building-editor

I've been working with Claude AI in adding several files to Hal, my local system that talks to Zigbee and Zwave and sends those sensor outputs to the building editor in the browser. 

Right now, it's all Zigbee sensors. AI is very powerful but it must be guided in the right direction and sometimes it takes a lot of finess to get it to work the right way. You have to know what you need done and then you need to be very clear about how to go about it and the AI will do a tremendous job of getting it done.

But.... If you're not clear, the thing will create stuff that's a bit difficult to change. so...

So yeah. I wouldn't want someone tinkering with my old c++ system with Claude AI. i have more that 1500 files all told and you could easily add something that doesn't fit what took me over 30 years to create. It's more than 12 megabyte of code. I'm only counting one directory so I believe a couple hundred more files exist.

And Claude absolutely doesn't want to see the whole thing, it would choke. But what I can do is feed little pieces to Claude. The problem is, Claude doesn't remember all the nuances of  previous chats. It works with what you send now, in this chat, and forgets everything before that. So as long as I give it just the relevant code, enough for it to understand, it works really well. That is, with clear instructions. 

The schematic system in Hal is second to none. It's a full scada system and it's a bit of a beast getting Hal to talk to the new building-editor.html system. I believe it'll take me at least 6 months to have a working system. But that's not a fully working system.

I need to have a way to pass all the schematic scada system to the building editor over the internet and show the connections between the block control system in real time, like Hal does in a local system. You can see Hal thinking and making decisions by looking at the blocks. It's so easy to debug and find a problem. 

The beauty of what happened in the power plants is that they needed a foolproof way of controlling dangerous equipment. They simply can't afford to allow ad-hoc created programs that have bugs a plenty. The beauty of blocks is that they do simple things and after debugging them for one small program,  over and over again, all the bugs show up and you have a really dependable system that won't kill people by opening 3" gas valves by mistake.

So I started this block control programming in 1991. In a working power plant. I've long since found all the bugs in the schematic system. I once ran a schematic that had no viewable gui for 9 months without having to reboot. It is rock solid. Add 3d and the system had to be rebooted every 8 hour shift.

Now, the 3d  is completely separate from the schematic logic, so I have all the confidence that it is rock solid. That's a really good thing when I plan on marketing a really interesting system. I should start soon adding the Gemma 4 AI system to the system. It'll be a new experience. It'll be a fully integrated AI system with creation and running of the automation system. 

I have several things on my side. First, I have a system that works intemittently with 3d animations as it was built to talk to 3d.  

Second Hal has a full scada block control system that is rock solid. 

Next it shows it's thinking in real time views of how the block communicate with each other. 

Next It'll be fully integrated with AI. So you'll be able to talk your home/building into being by talking to it and Hal will be able to explain the logic. 

Next you can talk to your smarphone. You can say: Set the alarms for the night. Also,  you'll be able to be on a beach in Jamaica and see a full 3d animated view of your home.

There is a feeling when you can do whatever you want without having to deal with executives who limit the way you do things.  And I have a vivid imagination of how to make the future happen. 

When I was a boy, I had a boy, in class, sitting next to me, in Canada, that would compete with me. He simply couldn't keep up. While he was involved in competition, I was involved with the fascination of how things worked. I had an insatiable curiousity of how life was put together. I really liked him and vice versa but he  could never understand why he couldn't retain the studies. He was always concerned with besting me. I tried to tell him I wasn't competing but he took it like I was trying to fool him. 

I still, today,  have friends who really believe I'm competing with them. They even tell me that I'm the most competitive person they know. It's strange to me that there is no way of getting them to understand that I love the way things work and it's not about besting them. Something that I simply could care less about doing.

It's not about mastering a discipline. It's about finding out how it exists. Humans are funny. 

Before I end this, I'm going to tell a story about a leadman who fooled me. I was having a problem with a control that took the outside air temperature and and biased how hot the hot water of the building was to heat the building. This control was an incredible machine. It was a small box that had physical linkages that could be calibrated to what temperature ranges you wanted to work at.

So I  asked him if he knew how it worked. And he said he did. Then we went out to the machine room and he proceeded to tell me how it  worked. After an hour and a half of time describing the box in detail, I began to realize he had no idea how it worked. He just didn't want me to know he didn't. To his credit, he fooled me for the entire time.

So, from that point on, I secretally believed he was a fake and was not worth going to for advice. So I basically wrote him off as a person who simply did not want to know how things worked.

Then one day, the folder that folded sheets in the laundry quit working. There was a box that controlled the whole process. It had failed.  Now, we didn't have a replacement and they were reduced to one folder instead of 2. I told him  I could fix it but I'd have to do a bit of figuring.

So we took the box to his office desk and he said he was willing to work with me in figuring out how it worked. So I powered it locally on his desk. It was obvious that transistor logic was involved. This was before chips hid their logic. It took me, literally 2 week to create a working schematic of the circuit board. One by one, I tested each stage of each  transistor. A transistor is like a wall light switch, unless it's analog. Simple, once you understand what they are.

So finally, after a long arduous but interesting 2 weeks of study and testing, I found a resistor (resists the flow of electricity) had changed it's value. The heat had made it  resist a higher level. I replaced the resistor and voila, the whole thing worked again. 

The thing is, he was intimately involved with every step i took. He stayed by my side the entire time and was genuinly interested in how the thing made it's decisions. I learned a lesson that day. Humans can live in worlds of their own making but when faced with pure honesty, they will find a beautiful form of genuine curiosity. 

And so it goes. Try not to write off a fellow human unless you see true evil, otherwise, most humans are willing to go the distance if they feel humility and character.

Pierre