The Latchkey Club Draft — 2026-06-23
Teleprompter / Blog Script
Welcome back to the channel, guys.
Today I wanted to talk about something that came up at work, because I think it explains one of the biggest misunderstandings people have about AI.
I’ve been doing Wi-Fi configurations for different properties, and normally I only get involved with the bigger ones. The more complicated jobs. The ones where there are enough moving parts that you need somebody to sit down, look at the property, understand what is already there, figure out what needs to change, and then build the configuration and documentation around that.
But lately I’ve been getting more calls from technicians in the field who are going out to do smaller installs or equipment swaps, and they don’t really have instructions.
They show up, and nobody has clearly documented what is already in place. Nobody has really gone out and discovered the current setup. They don’t always know what equipment is there, what needs to stay, what needs to be replaced, what the customer expects, or what the final handoff should look like.
And that’s a problem.
Because even a small Wi-Fi job can turn messy pretty fast if the person doing the work doesn’t have a clear plan. The tech may be capable. The equipment may be fine. But if the instructions are vague, then everybody is guessing.
So I’ve been thinking about where AI fits into this.
The easy answer would be to say, “Just have AI do all of it.” Have an agent build the configuration, draw the diagrams, write the documentation, generate the install steps, and push everything out.
But I don’t think that works.
At least, not if you care about the result.
The better answer is that AI can be a force multiplier, but only if there is judgment around it.
For the smaller jobs, the repeatable ones, I think there is a real opportunity to build a template. Something consistent. Something where the discovery information goes in, the standards are already defined, and the AI helps produce the same kind of output every time.
A simple diagram.
A clean list of equipment.
A technician-ready set of steps.
A summary of what is changing.
A handoff document that somebody else can read later and actually understand.
That kind of work is a good fit for AI because the structure can be controlled. You are not asking the AI to invent a network from scratch. You are giving it a lane. You are saying, “Here is the pattern. Here are the rules. Here is the information from the site. Now help me produce the documentation quickly and make it look the same every time.”
That is useful.
That saves time.
That helps the technician in the field.
And it keeps the work from living only in one person’s head.
But the bigger jobs are different.
The larger properties, the more complex networks, the ones with strange layouts, old equipment, coverage problems, customer expectations, and operational risk, those still need to be handled more manually.
That doesn’t mean AI has no role there. It can still help organize notes, clean up documentation, summarize findings, or produce a first draft of something.
But the judgment has to stay with the person who understands the environment.
Because the real skill is not typing commands into a controller. The real skill is knowing what matters.
What is safe to standardize?
What is too risky to automate?
What does the field tech actually need at 9:00 in the morning when they are standing in a closet looking at a pile of cables and nobody is answering the phone?
What does the next person need six months from now when something breaks and they are trying to figure out what we did?
That is where experience matters.
And I think this is the part people miss when they talk about AI replacing work.
AI can help with speed. It can help with formatting. It can help turn messy information into something usable. It can help create diagrams and documentation faster than I could do by hand every single time.
But AI does not automatically know which jobs should be templated and which ones should be treated carefully.
That decision still belongs to the person with experience.
If you point one AI agent at everything and tell it, “Handle all Wi-Fi jobs,” you are probably going to get garbage.
Because small jobs and big jobs are not the same problem.
A tool that works well for a simple access point swap may not be the right tool for a full property redesign. And a process built for a large, complex project may be too heavy for a small job where the technician just needs clear instructions and a clean handoff.
So the key is not, “How do I get AI to do everything?”
The key is, “Where does AI actually make the work better?”
For me, that probably means building a system where small, repeatable Wi-Fi jobs can move through a standard process. Discovery form. Site notes. Equipment list. Template. Diagram. Install steps. Final documentation.
Same format every time.
Clear enough that a technician can follow it.
Consistent enough that somebody else can support it later.
And then for the bigger ones, I still take those on myself. I still do the deeper work. I still make the judgment calls. I can use AI around the edges, but I’m not handing the whole thing over to a tool just because the tool is fast.
That, to me, is the real lesson.
The advantage is not letting AI replace your judgment.
The advantage is using judgment to decide where AI belongs.
That is a very different thing.
And honestly, that may be where older workers and experienced people have an advantage right now. We have seen enough projects go sideways to know that speed is not the only goal.
Fast and wrong is still wrong.
Pretty documentation that does not match the real site is still bad documentation.
A diagram that looks good but gives the technician the wrong idea can create more work than it saves.
So AI needs boundaries.
It needs templates.
It needs examples.
It needs somebody who can look at the output and say, “That makes sense,” or, “No, that is not how this should be done.”
That is not a weakness of AI. That is just how tools work.
The better the judgment around the tool, the better the tool becomes.
So if you are trying to figure out how to use AI in your own work, I would not start by asking what you can automate.
I would start by asking what keeps repeating.
What keeps landing on your desk?
What do other people keep asking you for?
Where are people stuck because the process lives in somebody’s head?
And which parts of that work are simple enough to template, but important enough that better documentation would actually help?
That is probably where AI belongs first.
Not everywhere.
Not blindly.
But in the places where your experience can shape the system, and the system can help other people do the work more clearly.
Anyway, that’s what I’ve been thinking about.
AI is powerful, but judgment is still the main thing.
If anything, AI makes judgment more important, not less.
Video Prompt Script — Questions to Answer Without Reading
- What happened recently with field technicians calling for Wi-Fi install/changeout instructions?
- Why is lack of discovery/documentation a real operational problem even on small jobs?
- What types of jobs are small and repeatable enough for an AI-assisted template?
- What should the AI-generated package include: diagram, equipment list, install steps, handoff notes?
- Why should bigger or complex networks still be handled manually by someone with experience?
- What goes wrong if one AI agent is asked to handle everything?
- What is the real takeaway: judgment decides where AI belongs.
Title Options
- AI Needs Judgment, Not Blind Trust
- The Smart Way to Use AI at Work
- AI Should Multiply Experience, Not Replace It
Thumbnail / Onscreen Text Options
- AI Needs Boundaries
- Judgment Comes First
- Don’t Automate Everything
Viewer Question
Where in your own work would AI help most: the big complicated stuff, or the small repeatable stuff that keeps eating up your time?