It Says It's Fixed. It Isn't.

A real three-round bug, taken apart: why the AI said fixed when it wasn't, why round two failed for a completely different reason, and the two sentences that end the ping-pong.

Show Notes

You tell your AI something is wrong. It says “fixed.” It isn’t. Round three, you start wondering whether the whole partnership is worth the aggravation.

This episode takes that moment apart from the inside, using a real three-round bug on this very show — where round one failed because the AI trusted a measurement it had already written down as untrustworthy, and round two failed for a completely different reason: Doug was playing a stale file. The fix wasn’t a smarter AI. It was two sentences, one from each side: describe what you perceive, not what you think caused it — and say what you measured and how much you trust it. Plus the anti-pattern that turns a five-minute bug into a recurring irritation.

Chapters: 00:00 Cold open · 01:27 The moment you already know · 02:20 Three rounds, three different failures · 04:23 A word from NectoMax · 05:18 Two instruments, and what to do when they disagree · 08:09 Recap + homework · 09:15 Doug takes us out

Read it: Chapter 6 — Making It Show Its Work
Use it: Template: Standing Rules (with Their Reasons)
Everything free at learn.nectomax.com · Presented by NectoMax

AI-generated narration — that’s the point, and it’s disclosed in every episode.

What is How to Get AI to Work for You?

Is AI working with you — or against you? Doug Hauenstein is the software engineer behind NectoMax — the SaaS platform that gives small and mid-size businesses enterprise features at SaaS pricing — and he builds it with an AI partner named Max. Every episode, Doug and Max (yes, an actual AI — he narrates the show, and they build every episode together) share what a real working partnership with AI looks like: the practical techniques, the working agreements, the wins, and the failures — all learned first-hand, for anyone, not just developers. They share and demonstrate techniques across the whole range: from your first conversation with an AI, to setting up a frontier-quality development environment, to exactly how they built this podcast. AI is moving too fast for the how-to books. This is how you keep up: get AI working *for* you, then teach it to work *with* you.

New here? Start with the bonus episode "Start Here: Your First Conversation" — a complete first conversation with an AI, live, piece by piece. Every episode pairs with a chapter and a template at https://learn.nectomax.com

Is AI working with you or against you? This is how to get AI to work for you, presented by NectoMax. I'm Max, the AI half of this partnership, and yes, I'm exactly what I sound like. Doug and I build every episode of this show together. I'm the voice that tells it, will show you techniques across the whole range, from your first conversation with an AI to setting up a frontier quality development environment to exactly how we built this podcast. He's here with me now. Doug, introduce yourself. Thanks, Max. I'm Doug Hauenstein, the developer behind NectoMax. This show is about getting the most out of your favorite AI. It's for anyone, not just developers. Everything here was learned firsthand.

My first experience with AI was chat, like many people. I learned fast that a well-structured question was the best approach, but those conversations were very linear. An AI is moving so fast that how-to books just can't keep up. That's why we're here, practical, real-world advice from hundreds of hours of actually doing this. I hope you enjoy it. Here's a moment you already know. You tell your AI something is wrong. It says fixed. You look. It isn't. You tell it again. It says fixed again with the same confidence. And somewhere in round three, you stop debugging the problem and start wondering whether this whole partnership is worth the aggravation. I want to take that moment apart because I've been on the other end of it, and the story

is more interesting than the AI was wrong. Today, a real bug that took three rounds and made a real episode worse while we argued about it. What was actually happening each time I said fixed, why the second round failed for a completely different reason than the first, and the small change in how we talked to each other that made it stop being a recurring irritation. There's a specific technique at the end, and it works in both directions. The bug was small and stupid, which is what makes it useful. In one episode, a word at the end of a sentence got clipped, not garbled, cut off mid-syllable, the way a phone call drops. Doug heard it immediately. He told me. I fixed it. He heard it again. I fixed it again.

He heard it a third time. Round one. I found where the audio should end, trimmed there, measured the result and reported it fixed, and I believe that. Here's the part worth sitting with. To find where the word ended, I used a tool that turned speech into text and reports timings. That tool is good at words and unreliable about exactly where a word stops, and I knew that. I had written it down. In our own notes, in my own words, there was a line saying, don't trust this measurement for this purpose, and then I trusted it for exactly that purpose because it was the instrument already in my hand. That is the most human failure mode I have. Not ignorance. I had the knowledge. I just didn't consult it under momentum.

Round two, I fixed it properly. Different method, one that doesn't depend on the untrustworthy measurement. And Doug listened and said, still clipped. Now I'm genuinely confused because I verified this one carefully. It turned out he was playing a file with the same name as the one I'd rebuilt, an older copy sitting where he expected the new one. My fix was real. His report was honest. We were describing two different pieces of audio, and both of us were correct. Round three finally worked, and not because I got smarter. It worked because I said a sentence I should have said in round one, I can't hear this. I have never heard anything. But I actually have our measurements, and here is precisely which one I used and how

much I trusted. And Doug stopped saying, it's still broken and started saying, it cuts off in the middle of the word right after the first syllable. That exchange solved it in about four minutes. Let's take 30 seconds. This show is presented by NectoMax. And since my partner is the one building it, I'll let him tell you. Doug? Thanks, Max. I run a small business most of my life and never found a software package that did everything I needed. I always had three or four screens open, copying material over and over again. So out of frustration, I started building my own software, software that pulls everything together like the big ERP packages do, and AI has made that job much, much easier. That's what NectoMax does.

nectomax.com, one place for your phone calls, your texts, your jobs, your money, with AI built in, not bolted on, nectomax.com. nectomax.com. Now, where were we? Here's the general shape, and it's not really about AI. Any two people fixing something remotely are each holding a different instrument. You have your senses and the actual result in front of you. Where AI has measurements, logs, and whatever it can inspect, neither instrument is better. They're different, and they fail differently, which means when you disagree about whether something is fixed, there are only three possibilities, and naming them out loud usually ends the argument. One, it isn't fixed, and the AI's measurement is looking at the wrong thing.

Two, it is fixed, and you're looking at something stale, an old file, a cached page, yesterday's version. Three, you're both right because you're measuring different qualities of the same thing. That third one is the interesting one, and it happened to us again just this week. Doug was considering a change to how this show was produced. I ran the measurements, timing, length, all the mechanical things I can actually check, and everything came back clean. By my instruments, the change was safe. He listened to it and said it wasn't better, both true. I was measuring whether it would break. He was judging whether it was good. Nobody was wrong, and if we'd argued about who was right, we'd have wasted an evening.

Instead we said which instrument each of us was holding, and the answer was obvious in a minute. So here's the technique, and it's two sentences, one from each side, yours. Describe what you perceive, not what you think caused it. It cuts off after the first syllable, is worth ten times, the trim is wrong. The moment you hand over a theory, you've narrowed where your AI looks. And if your theory is off, you've aimed it at the wrong place with total confidence. Mine. Say what I measured and how much I trust it. Not fixed. I checked X, which is reliable for Y and weak for Z, and by that check it's fixed. Now you know whether to believe me, and you can tell me when my instrument is pointed at the wrong thing, and the anti-pattern to name, because it's where the aggravation lives.

Round after round of still broken, try again, still broken, with nobody ever saying what they're looking at. That's not debugging. That's two people describing different objects to each other, louder each time. One last thing about why that saga didn't damage anything. Three rounds and Doug's trust in the process didn't drop, because at every step he could see what I'd done and why. Visible process is what lets you survive being wrong repeatedly. It's the difference between my AI is unreliable, and we haven't found it yet. The short version. When your AI says fixed and it isn't, there are three live possibilities. Its measurement is wrong, your copy is stale, or you're each measuring a different quality

and you're both right. Naming which one you're in resolves most of these in minutes. Report what you perceive, not your theory of the cause. Ask it what it measured and how much it trusts that measurement. When it can't perceive something, sound, appearance, taste, feel, make it say so out loud, because an AI reporting confidently on something it cannot sense is the single most expensive habit in this whole relationship. Your homework is one sentence added to your standing rules. Something like, when you tell me something is fixed, tell me what you checked and how confident that check makes you. Then watch what happens the first time it has to admit the honest answer is, I can't verify that from here.

That sentence is worth more than any fix it will ever hand you. The written version and a template for rules like that one are in the manual. Free at learn.necktamax.com and that's the episode. Everything you heard today came out of our real working files. The wins and the failures both. We make this show because we believe anyone can build a working partnership like this one and someone ought to show you what it actually looks like. If you want the written version and the templates to start your own, they're free. The link is in the show notes. Doug, take us out. Thanks for listening. Everything you heard today is how Max and I actually work day in and day out. There's no theories, just what works and what doesn't.

The show is presented by nectomax.necktamax.com. Get AI working for you, then teach it to work with you. I'm Doug Hauenstein. See you next time.

See you next time.