AI and the Talking Wall
From Whiteboards to Chatbots: Rediscovering the Talking Wall
by Allen Edwards. CEO of L-36.com
Table of Contents
- The Talking Wall
- Finding New Walls
- When the Walls Were Gone
- The Promise of AI
- The Disillusionment
- The Realization & Closing Thoughts
The Talking Wall
Back in my days at Hewlett-Packard, there was an unofficial tradition whenever someone got stuck on a problem. You'd push your chair back, wander over to a colleague, and start explaining what was giving you trouble.
What we discovered was almost comical: the other person didn't even need to know anything about the problem. Half the time, their expertise didn't matter. What mattered was the simple act of talking it through. The brain, when forced to put tangled thoughts into words, often untangles itself.
We started joking about it. "You don't need an engineer — you just need a wall." Pretty soon, it became part of the culture. When someone solved their problem mid-sentence, they'd laugh and say: "Thanks for being my talking wall."
Of course, sometimes the wall had a name. Paul, for example.
One afternoon, I remember walking into Paul's office, frustrated with a design that wasn't behaving the way it should. I started explaining my setup, sketching little diagrams on his whiteboard. Paul nodded politely, occasionally asking a question. Halfway through my explanation, the solution clicked in my head — it had been obvious all along, but invisible until I heard myself say it out loud. I stopped, laughed, and said, "Never mind, Paul, I've got it."
He grinned and said, "Glad I could help."
Other times, Paul wasn't completely in the dark. He'd know just enough to nudge me in the right direction. Maybe he'd remind me of an obscure test procedure, or point out that a certain assumption I was making might not be valid. He wasn't always the expert, but his tiny contributions could make the difference between hours of spinning my wheels and moving forward in five minutes.
That was the beauty of the talking wall. Whether it was a real colleague with some insight, or just someone lending their ears, the process worked. The wall didn't need to solve the problem — it only needed to listen.
Finding New Walls
When I retired, the talking walls of HP vanished almost overnight. No more colleagues to lean on, no more casual office visits to hash things out on a whiteboard. I suddenly found myself working on projects alone, with no Paul down the hall.
To my surprise, though, the internet often stepped in. Google, Reddit, and YouTube became a new kind of community. Sometimes a search would land me on exactly the explanation I needed, a Reddit thread full of people wrestling with the same oddball issue, or a YouTube tutorial that patiently walked me through a process. They deserve more credit than they get. For countless problems, those three together became my replacement walls.
Every now and then, I'd stumble into something even better — real people willing to help.
I remember when I was first writing my Android app StartLine. A fellow developer took the time to introduce me to Stack Overflow and gave me some basic guidance on how to ask good questions and get solid answers. That was a turning point. Suddenly I wasn't just fumbling in the dark — I had access to a whole new ecosystem of problem-solving.
Another time, one of the early users of StartLine — a sailor in Australia who also happened to be a developer at IBM — reached out and offered me working code for UDP data transmission. That gift was the foundation of my onboard WiFi network, which connected the instruments on my boat. It wasn't just helpful — it was transformative. One person's generosity of time and expertise opened up a whole new capability for me.
So it wasn't that I was completely alone. I did find new walls — digital walls, scattered across forums, tutorials, and the occasional generous stranger. And often, those walls were remarkably solid.
When the Walls Were Gone
But for every success story, there were just as many dead ends.
The truth is, not every problem can be solved with a clever search or a helpful stranger. Sometimes the issue is just too specific, too wrapped up in the quirks of my code, my project, or my boat. When I tried talking it out with friends, the answer was usually the same: "I don't know anything about that." And they meant it kindly, but there was nowhere else to go from there.
Even the people who might have been able to help were often too busy. Everyone has their own jobs, their own deadlines, their own lives. My questions — obscure, technical, and usually one-of-a-kind — weren't high on anyone's priority list. And so, I was left staring at my own screen, talking to no one.
Sometimes Google, Reddit, or YouTube could bail me out, and I'll give them credit — they often did. But there were plenty of times when the only results I found were my own posts from a year earlier, asking the exact same question. That was always a low moment. Nothing makes you feel more stuck than realizing you're your own only source.
It was frustrating, and at times isolating. I missed the days at HP when you could just pop into Paul's office, sketch something on the whiteboard, and find the answer spilling out halfway through your explanation. Without those talking walls, problems lingered longer. Some even sat unsolved, waiting for weeks or months until I finally circled back to them with fresh eyes.
The Promise of AI
The early days with AI felt a lot like discovering a new, brilliant colleague.
One time, I handed over a twenty-line block of code buried in nested loops. It had been running tens of millions of times without error, but I needed to understand it better in order to fix a rare failure. AI explained my own logic back to me more clearly than I had ever framed it in my own head. That clarity gave me exactly the foothold I needed to solve the problem.
Another time, I faced a security challenge on my website: blocking parameters that ended with a single quote, like &tide=brisbane', while still allowing legitimate ones like &tide='brisbane'. I could have burned hours trying to craft the perfect regular expression. Instead, I described the problem, and AI produced this in seconds: preg_match("/(?:^|&)[^&=]+=[^']*'(&|$)/"). It worked perfectly. In that moment, AI wasn't just a tool; it felt like having a colleague at my side who actually liked writing regex.
Sometimes, the help went far beyond coding. When a Medicare claim was denied, I asked AI to explain the billing codes. It pointed out that the clinic had submitted a combination — -25 and G2211 — that wasn't allowed together. I contacted the clinic, they corrected the submission, and Medicare paid the bill in full. That wasn't just convenience; that was real money saved.
And then there were the times AI showed me what I didn't know I didn't know. Programming languages evolve, and instructions exist today that weren't around when I first learned them. For example, in PHP, instead of writing long isset() checks to avoid undefined index warnings, AI reminded me of the null coalescing operator (??). With it, I could replace lines of defensive code with something as simple as: $value = $_GET['key'] ?? '0'; It was shorter, cleaner, and exactly the kind of quiet improvement that makes code both easier to read and more reliable. Those little nudges have made my work cleaner and my life easier.
Moments like these made AI feel revolutionary. Not flawless, not magic — but undeniably powerful.
The Disillusionment
Of course, the honeymoon couldn't last forever.
After a while, I began to feel the cracks. Many of the helpful-looking links AI suggested turned out to be dead ends — literally, 404 pages. And when I asked for a small change to my code, I often got a full rewrite instead. It didn't seem to matter that the code in question had already run successfully tens of millions of times; AI was eager to throw it all out and start fresh, even when stability was more important than elegance.
When I really did want new code, things weren't much better. AI would often try to solve everything in one big sweep, producing a wall of code riddled with errors that took ages to sort out. The best results came not from letting it "do it all," but from breaking the work into smaller steps and guiding it carefully, piece by piece. With enough back-and-forth, I usually got there — but it took patience.
And then there were the times it was simply wrong. No malice, just confident mistakes. I once asked how to deal with a stuck valve in my boat engine. With absolute certainty, AI told me I could just pull the valves out without removing the head. Anyone who's ever held a wrench knows how ridiculous that is. I laughed out loud — but it was a good reminder that "ChatGPT can make mistakes. Check important info," as the disclaimer says at the bottom of every screen.
It wasn't that the tool stopped being useful — far from it. But the hype had promised too much. This wasn't a magic box that could replace developers or solve problems instantly. It was more like an overeager assistant who needed a lot of direction, and who occasionally made things harder before they got easier.
The Realization & Closing Thoughts
Over time, my expectations shifted. AI wasn't the flawless replacement for developers that the hype had promised. It wasn't going to hand me perfect solutions or write production-ready code without supervision.
But then I realized: maybe it didn't have to.
What I had really been missing all these years wasn't an all-knowing expert — it was a talking wall. Back at HP, I didn't need Paul to solve my problems for me. I needed him to listen, to reflect my thoughts back, and occasionally to nudge me in the right direction. That was enough to unlock the answer in my own head.
AI turned out to be exactly that. It's always available, never too busy, and never embarrassed to ask the obvious question. Sometimes it's wrong, sometimes it overcomplicates things, but it's always there to talk it through. And just like with Paul, half the time I find the solution in the middle of explaining the problem.
I've since learned that I'm not alone in this discovery. I came across an article describing how others were finding the same thing — AI as the modern-day talking wall. It's not a replacement for people, or for expertise. It's a companion for thinking out loud, a partner in untangling problems that are too messy to sort through alone.
That's the real promise. Not the hype, not the magic — but the simple, steady presence of a wall that talks back.
Closing Thoughts
In the end, I've come to appreciate AI for what it really is. It may not be flawless, but it has given me back something I had lost since my HP days — a talking wall, always ready to help me think through a problem, explore an idea, or polish a solution.
And beyond that, it has become a partner in creation. This very story is proof: I brought the memories and the fragments, and AI helped shape them into something whole. It doesn't just answer questions; it helps me turn thoughts into reality.
I hope you enjoyed it.
NOTICE: Some pages have affiliate links to Amazon. As an Amazon Associate, I earn from qualifying purchases. Please read website Cookie, Privacy, and Disclamers by clicking HERE. To contact me click HERE. For my YouTube page click HERE