Cheers everyone, I read some threads in here a few months ago about using antigravity to spin up software and it caught my attention. I had an idea I tossed around in my head for a couple years now but never had the means to actually build something to test it out so I committed to learning how to really use AI past just a glorified search engine.
I watched a few free courses on how to navigate the dashboard and use Claude and figured I’d learn the best by actually just messing with the stuff.
I’m a 10 year journeyman lineman with absolutely no technical background but I wanted to build a sort of field reference/knowledge tool that I had always wanted as an apprentice. The idea started as basically a “wrapper” with a RAG layer of actual industry specific content and safety documents. And also a quiz feature that would test knowledge of specific topics and actual field procedures which I had ai generate in bulk with multiple choice and open response answers and I’d go through and verify the information was actually accurate. Which it consistently would get safety critical facts wrong as well as say things that wouldnt make sense in physical reality so that took quite a bit of time trying to correct.
So yea a glorified chat gpt with a quiz feature if you will but I never built anything before so i wanted to try it out.
Once I felt I had enough quiz questions and a small base layer of RAG documents to pull from I gave the link to a few guys I work with/worked with in the past to check out and I got the ok from them to include some analytics on my end so I could see how they used it but I don’t even think they knew what I was talking about lol.
I realized a few things with the data. First thing was that the AI responses were pretty terrible. Said stuff that didn’t make sense or were dangerously inaccurate but it did awesome on actually explaining electrical theory and how equipment works and so on… due to the fact that it probably knows everything an electrical engineer would know.
The next thing was that one user in particular, an apprentice I work with, used it to ask about specific procedures and how to do them basically in preparation for how to attack a job we were about to do in the near future which I thought was interesting. Of course AI right now has no grounding in reality of how to perform any physical task of working on energized high voltage lines so the answers were terrible and resorted to the safest possible response. Of course no job is the same and there are so many variables that could change the thinking pattern of someone who’s about to do something like say replacing a pole mounted transformer.
They had access to it for a couple weeks and not everyone used it but there were a select few who reached out to me separately saying how “cool” they thought it was and how I was “onto something” but I know that doesn’t mean anything. These few also used it several times unprompted. I think that the trust in the initial project was killed pretty quickly because it clearly just didn’t sound like someone that performs the work and it would say things that are physically impossible.
After the initial test I took the fact that it was generally used to prepare for specific procedures, walk through a past trouble call, or to learn a thought process that only comes from years of experience in the field.
The original app was bad at this and it would regularly hallucinate even though I had RAG content made specifically for certain questions. There was higher friction in my eyes because they had to go to the app with a question vs just opening it and having stuff right in front of them so I pivoted.
Instead I decided to make basically a content library of how to do certain procedures, how to troubleshoot different service calls, specific things I knew I struggled with understanding as an apprentice and so on. The idea was to essentially have a sh*t load of specific stuff that usually only lives in guys heads written down and super simple to understand. I made templates for different doc types and basically just talked with Claude about a topic or a situation and we’d extract all my knowledge into these templates that became user facing content. These docs would also serve as RAG layers in the chat function. Which I haven’t exactly nailed down yet but the hope was to open it and immediately have something useful in front of you.
I’d record voice memos of my own day or trouble calls I was dispatched to talking about what I found how I fixed it and my thought process and it was honestly pretty cool to do because I finally found a use for all my experience.
Built it on nights and weekends with every spare minute I have which sometimes isn’t much with my current job and family but I was locked in, maybe too much so that I got caught in the builder loop of let me make this perfect for 4 months and avoid putting it out there.
Long story short I said screw it and gave the link to the same guys that used v1 and had the analytics on it’s been about 5 days now and no one has really used it much past the first two days or really said anything about it to me which I’m taking so far as it might be more of a nice to have and not necessarily a genuine product.
I’d be curious to see what anyone has to say about any of this regarding a better route to validation or pivoting to something I’m not fully seeing.
I know there is opportunity within the utility industry even just due to the fact that the demand for electricity is going to increase 10 fold in the coming years as well as data centers. Current infrastructure just can’t handle it. There also seems to be a pretty big gap from guys retiring to guys coming into the trade.
One obvious struggle for what Ive been working on is the fact that for a physical trade half of the work is knowing how to do something and the other half is actually doing it. Only so much can be taught and comprehended through text on a screen so that’s definitely where i fall short with this.
I’ve been apprehensive to give this out to people I don’t know because much of the information is safety critical and I don’t have anything in place currently to protect myself if they misunderstand what’s written or the chat hallucinates.
I’ll answer any clarifying questions anyone has on this whole thing and thanks in advance for anyone who has some insight!
Constructive criticism is welcomed..
Thanks,
J
P.s. the real problem that brought this project to mind was the “engineering” tone current text books are written in as well as being subject to how well you understand them and how much the journeyman training you knows. You’d be surprised how many guys do this job that don’t understand how stuff works.
Current models can explain theory like the back of their hand but they don’t have the specifics of how to actually do stuff or the seasoned journeyman thought process… of course someday sooner than I think they probably will though.
I watched a few free courses on how to navigate the dashboard and use Claude and figured I’d learn the best by actually just messing with the stuff.
I’m a 10 year journeyman lineman with absolutely no technical background but I wanted to build a sort of field reference/knowledge tool that I had always wanted as an apprentice. The idea started as basically a “wrapper” with a RAG layer of actual industry specific content and safety documents. And also a quiz feature that would test knowledge of specific topics and actual field procedures which I had ai generate in bulk with multiple choice and open response answers and I’d go through and verify the information was actually accurate. Which it consistently would get safety critical facts wrong as well as say things that wouldnt make sense in physical reality so that took quite a bit of time trying to correct.
So yea a glorified chat gpt with a quiz feature if you will but I never built anything before so i wanted to try it out.
Once I felt I had enough quiz questions and a small base layer of RAG documents to pull from I gave the link to a few guys I work with/worked with in the past to check out and I got the ok from them to include some analytics on my end so I could see how they used it but I don’t even think they knew what I was talking about lol.
I realized a few things with the data. First thing was that the AI responses were pretty terrible. Said stuff that didn’t make sense or were dangerously inaccurate but it did awesome on actually explaining electrical theory and how equipment works and so on… due to the fact that it probably knows everything an electrical engineer would know.
The next thing was that one user in particular, an apprentice I work with, used it to ask about specific procedures and how to do them basically in preparation for how to attack a job we were about to do in the near future which I thought was interesting. Of course AI right now has no grounding in reality of how to perform any physical task of working on energized high voltage lines so the answers were terrible and resorted to the safest possible response. Of course no job is the same and there are so many variables that could change the thinking pattern of someone who’s about to do something like say replacing a pole mounted transformer.
They had access to it for a couple weeks and not everyone used it but there were a select few who reached out to me separately saying how “cool” they thought it was and how I was “onto something” but I know that doesn’t mean anything. These few also used it several times unprompted. I think that the trust in the initial project was killed pretty quickly because it clearly just didn’t sound like someone that performs the work and it would say things that are physically impossible.
After the initial test I took the fact that it was generally used to prepare for specific procedures, walk through a past trouble call, or to learn a thought process that only comes from years of experience in the field.
The original app was bad at this and it would regularly hallucinate even though I had RAG content made specifically for certain questions. There was higher friction in my eyes because they had to go to the app with a question vs just opening it and having stuff right in front of them so I pivoted.
Instead I decided to make basically a content library of how to do certain procedures, how to troubleshoot different service calls, specific things I knew I struggled with understanding as an apprentice and so on. The idea was to essentially have a sh*t load of specific stuff that usually only lives in guys heads written down and super simple to understand. I made templates for different doc types and basically just talked with Claude about a topic or a situation and we’d extract all my knowledge into these templates that became user facing content. These docs would also serve as RAG layers in the chat function. Which I haven’t exactly nailed down yet but the hope was to open it and immediately have something useful in front of you.
I’d record voice memos of my own day or trouble calls I was dispatched to talking about what I found how I fixed it and my thought process and it was honestly pretty cool to do because I finally found a use for all my experience.
Built it on nights and weekends with every spare minute I have which sometimes isn’t much with my current job and family but I was locked in, maybe too much so that I got caught in the builder loop of let me make this perfect for 4 months and avoid putting it out there.
Long story short I said screw it and gave the link to the same guys that used v1 and had the analytics on it’s been about 5 days now and no one has really used it much past the first two days or really said anything about it to me which I’m taking so far as it might be more of a nice to have and not necessarily a genuine product.
I’d be curious to see what anyone has to say about any of this regarding a better route to validation or pivoting to something I’m not fully seeing.
I know there is opportunity within the utility industry even just due to the fact that the demand for electricity is going to increase 10 fold in the coming years as well as data centers. Current infrastructure just can’t handle it. There also seems to be a pretty big gap from guys retiring to guys coming into the trade.
One obvious struggle for what Ive been working on is the fact that for a physical trade half of the work is knowing how to do something and the other half is actually doing it. Only so much can be taught and comprehended through text on a screen so that’s definitely where i fall short with this.
I’ve been apprehensive to give this out to people I don’t know because much of the information is safety critical and I don’t have anything in place currently to protect myself if they misunderstand what’s written or the chat hallucinates.
I’ll answer any clarifying questions anyone has on this whole thing and thanks in advance for anyone who has some insight!
Constructive criticism is welcomed..
Thanks,
J
P.s. the real problem that brought this project to mind was the “engineering” tone current text books are written in as well as being subject to how well you understand them and how much the journeyman training you knows. You’d be surprised how many guys do this job that don’t understand how stuff works.
Current models can explain theory like the back of their hand but they don’t have the specifics of how to actually do stuff or the seasoned journeyman thought process… of course someday sooner than I think they probably will though.
Dislike ads? Become a Fastlane member:
Subscribe today and surround yourself with winners and millionaire mentors, not those broke friends who only want to drink beer and play video games. :-)
Access Restricted: Unlock 1,000,000+ Posts
Stop Peeking Through the Keyhole.
Open the Door.
You hit a wall because the best advice isn't free—it's earned. Since 2007, MJ DeMarco has been active here daily (99.9% active rate), building a war room for entrepreneurs who refuse to settle for mediocrity.
- Active Daily: Life-changing content posted dozens of times every day. No dead threads.
- MJ's Inner Circle: Direct access to the author of The Millionaire Fastlane.
- Vetted Network: Connect with founders scaling to 7 and 8 figures.
- Proven Roadmaps: Strategy over motivation. Execution over "hustle porn."
"SCALED TO 6-FIGURE MONTHS"
"Tips and advice here saved me from huge mistakes, others allowed me to scale my business to 6-figure months."— User MitchC
"INSANE QUALITY OF PEOPLE"
"The number of high quality people I've met from this forum has been insane. All of it predicated on building real value."— User Richard Peck
"You are the average of the five people you surround yourself with."
Are you surrounding yourself with success?
