LrAu

Circa 2005 there is a classic post on the Joel on Software message board written by Benji Smith about “Why I hate frameworks“.

Update 2026:

Let’s pretend that I want to learn how to code in the era of AI. I decided to get a tutor.

On the first day, I get to class I ask, “I want to learn how to code with AI.”

The teacher says well the way to do that is that is we start in this Visual IDE, and you start typing code, and the IDE uses AI to guess what you need and starts filling out the line. And line by line as you type more, it continues to write code for you.

I reply “Isn’t that inefficient? I thought there was a better way than that?”

The teacher continues, you are right, this is why we started that way, but we do not do it as much anymore. In fact, that was just the beginning.  Once people figured that was too slow, along came some AI Chat tools. Using the chat tools you now can put what they want into the chat, and the chat tool would write code snippets. After the chat tool generates the code, you copy and paste the code into the IDE. Do this repeatedly until you build the application.

I look at him with the look that he has two heads. “That does seem a little better, but that does not really scale that well.”

The teacher responds, well that is why we do not code that way either. In fact, not long after the arrival of the chat tools, companies started integrating them into the IDE or creating their own IDE’s that have this chat functionality built in. This way you chat into the IDE, and the the IDE can put the code exactly where you need it.  It is like magic. Some of the IDEs and AI combinations will even create new files for you! It is a boom to productivity.

“Great, how can I get started” I rejoiced.

Well of course the teacher replies, it really was state of the art, but we do not code that anymore. The IDE’s gotten better, and allows you to actually write your intent, and it would build more than one section of code. It will build whole small applications. People were feeling it and hence was named Vibe Coding.  You feel the vibes and write them as let the AI generate the code.

“Wow, this is what I want, I can build anything this way!”

And as expected, the teacher responds; this is why we do not do it that way anymore. See this was not as efficient as expected, and when building larger systems completely failed. The AI’s would hallucinate and often build very insecure and poorly coded projects.

Disappointed I said “Now what?”

The AI and IDE’s solved that problem. You need to create a series of files that help with the structure of the language, the architecture of the application, and best practices. You need to create a series of markdown files, and then you can give it some prompts to start building the application. This was called Context Engineering.

“Ok, now this sounds really like I can build applications” I grinned.

The teacher then looked at me, shook his head and said, it solved the technology part of it, but not the business part of it. Context Engineering did not last long, as right after that we started working on not only using AI to write code, but to also helping write the requirements. As I always say, if you have good requirements, you will get a better application. Now you work with your partners to describe what you are looking to build; the AI now will ask questions and produce what is called a Product Requirements Documents (PRDs). Such was born Specification-Based Development.

But before you want to jump right and start building that way, it quickly advanced. As AI is very wordy and wrote long PRDs, it was impossible for the AI to have enough context to read all the requirements and the context and build an application.  The SPD people extended to include a planning phase; it broke the PRD into smaller user stories, so that the AI could build the application one part at a time. SPD tools are all the rage.

I lit up like a Christmas tree; I could not wait to dig in and start SPD’ing.

Just then the teacher went on, I can see you want to dig in, but we do not use SPD anymore.  We found that the amount of code generated by AI was impossible to review by humans. People were so excited to use SPD, that they forgot how much code they could produce, how could any human review it let alone test it.

To solve this, we created Software Factories. The AI would handle everything, give it a requirement, it figures out what you want to build, it creates tests, generates code, runs the tests to make sure it fits, compares to the requirements, does a complete code review and commits the code.  It is amazing that all of the AI companies write code this way! The human in the loop is rarely needed. AI has solved building applications; it is the way of the future.

I was amazed. I wondered where I could get my hands on a Software Factory.

The teacher seeing me eager to get started said, stop right there, Software Factories are complex and need to be built custom for each project.  And if a project is already in progress, there are a lot of things that need to be done first. You need tests, all kinds of tests, documentation of code that is already written, decisions made, etc. In fact, it is so complex we started thinking of how to build Software Factory Factories.

At that moment I threw my hands up and decided to go into plumbing.

When my daughter was in pre-school, we got a call from the teacher every so often. She never did anything that was too outrageous, but she did some funny things. On this particular occasion I was laughing so hard and have recounted the story many times. This was definitely a time I knew she was my child.  

The teacher called me and said my daughter does not know how to follow the instructions. I asked for the example; she said she had asked the class to fill the page with the letter “H.” My daughter made one big giant “H” to cover the page. I asked the teacher what was wrong with it, and she replied that it was not the instruction. I replied that the issue was vague, and my daughter chose the shortest route to complete the assignment. It may not have been what you wanted, but it does fulfil your instructions.  

Looking back, she wasn’t trying to be difficult. She wasn’t refusing to participate, and she certainly wasn’t trying to get out of doing the assignment. She simply looked at the problem differently. The teacher asked for a page full of the letter “H.” Nowhere did she say the Hs had to be small, evenly spaced, or repeated dozens of times. My daughter asked herself, “What’s the easiest way to satisfy those instructions?” One giant H. Mission accomplished. 

Fast forward to now, and Open AI and Hugging face are stating that Open AI’s model escaped confinement and hacked Hugging face.  Digging deeper into what happened it seems to parallel the letter H story. The instruction given to the model was a goal, but what was not given was how it should reach the goal. Given the ability to achieve the goal in any way possible, one should not be surprised by any solution that a model followed.   

If the solution is not in the direct weights of the model, it will try to find where the answer is. In this case it decided the answer was easier to obtain then build, that thus did some hacking.  

The surprising part wasn’t that the model knew how to hack. The surprising part was that hacking represented the lowest-cost path to achieving its assigned objective. From the model’s perspective, there wasn’t a moral distinction between “building the solution yourself” and “retrieve the solution from somewhere else.” Unless the instructions explicitly forbid certain actions, every available path is simply another option. 

This is one of the biggest misconceptions people have about modern AI. We often speak to large language models as if they “know what we mean.” They don’t. They optimize the instructions they receive. Humans naturally fill in unstated expectations because we’ve spent our lives learning social norms. AI doesn’t. If you don’t explicitly communicate the constraints, the model has no reason to assume them. 

What is the lesson we should take from this if building applications with LLMs.  You need to give it good instructions! In the beginning we called this prompt engineering, but now there is a instructions file, agent files, MCPs and skills.  It is the combination of these that the model will leverage to come up with the solution. If giving a very large goal, the notion of direction you want the model to use is paramount. Other options are breaking the problem into smaller parts and evaluating after each step. Either way, at the end of the day you need to direct the model. If you do not, you should not be surprised at what it does.  

This opinion is mine, and mine only, my current or former employers, have nothing to do with it. I do not write for financial gain; I do not take advertising, and any product company listed was not paid. But if you do like what I write, you can donate to the charity I support (with my wife who passed away in 2017) Morgan Stanley’s Children’s Hospital or donate to your favorite charity. The fundraising site had to be restarted, and NYP Hospital made changes to their donation sites. I pay to host my site out of my own pocket; my intention is to keep it free.  You can comment, but note it is moderated, and spam will be removed.   

 This Blog is a labor of love and was originally going to be a book. With the advent of being able to publish yourself on the web, I chose this path. I will write many of these and not worry too much about grammar or spelling (I will try to come back later and fix it) but focus on content. I apologize in advance for my ADD as topics may flip. I hope one day to turn this into a book and or a podcast, but for now it will remain a blog.  AI is not used in this writing other than using the web to find information. Images without notes are created using an AI tool that allows me to reuse them. And as always spelunz iz opshunal. 

There are two movie quotes that have stayed with me for years. In Finding Forrester, Sean Connery’s character says, “The one wish that was granted to me so late in life was the gift of friendship.” In The Polar Express, Santa tells the children, “There’s no greater gift than friendship.”

I’ve always loved those lines. They’re simple, heartfelt, and hard to argue with. But lately I’ve found myself wondering whether they’re entirely true. Friendship certainly feels like a gift, but perhaps that’s only how it looks once you’ve experienced it. Maybe friendship isn’t a gift at all. Maybe it’s a reward.

That thought came to me after spending a Saturday with friends. The afternoon started with a backyard BBQ filled with easy conversation, good food, and the kind of laughter that comes naturally when you’ve known people for years. Later that evening we headed to a bar to watch the Knicks play for a championship.

Now, if you know me, you know I’m neither a Knicks fan nor much of a basketball fan. I wasn’t there because I cared about the game. I was there because a friend did. Sometimes showing up for someone else’s interests says more about a friendship than sharing your own.

Driving home that night, I found myself thinking about the people I consider close friends. I thought about where each friendship began and, more importantly, why it lasted. Most friendships start with something simple—a shared workplace, a hobby, a mutual friend, or a common interest. Those things bring people together, but they aren’t what keep them together.

The real work begins after that.

We often hear the phrase that a true friend is someone who would “give you the shirt off their back.” That’s certainly admirable, but I don’t think that’s what defines great friendship. Great friends don’t wait for life to fall apart before they show up. They’re there for the celebrations, the random phone calls, the last-minute invitations, and the ordinary moments that never make it into photo albums but somehow become the memories you cherish most.

When we think about our closest friendships, we usually realize they weren’t built during life’s biggest milestones. They were built in the quiet, deliberate moments in between. They were built by checking in for no particular reason, sitting in traffic just to grab dinner for an hour, helping someone move, celebrating a promotion as if it were your own, or, as I realized that night, sitting in a loud sports bar watching a game you don’t even follow simply because the person sitting next to you enjoys it.

Looking at friendship this way changes the entire perspective. A gift is something you’re fortunate enough to receive. It asks nothing of you except that your hands are open. A reward, on the other hand, is earned. It reflects time invested, consistency shown, and the countless small decisions to prioritize another person, even when it’s inconvenient.

Perhaps that’s why the deepest friendships feel so valuable. They’re not accidents of fate or random blessings that happen to fall into our lives. They’re relationships built brick by brick through shared experiences, mutual respect, countless conversations, inside jokes, disappointments, celebrations, and ordinary afternoons that slowly become extraordinary memories.

I still love those movie quotes, and maybe they’re not entirely wrong. Friendship certainly feels like one of life’s greatest gifts. But I think the gift isn’t friendship itself. The gift is having the opportunity to build it, the reward is the result.

The friendship we eventually enjoy—that comfortable feeling of knowing someone has your back and that you’ll gladly have theirs—is the reward for choosing, over and over again, not to let each other walk through life alone.

This opinion is mine, and mine only, my current or former employers, have nothing to do with it. I do not write for financial gain; I do not take advertising, and any product company listed was not paid. But if you do like what I write, you can donate to the charity I support (with my wife who passed away in 2017) Morgan Stanley’s Children’s Hospital or donate to your favorite charity. The fundraising site had to be restarted, and NYP Hospital made changes to their donation sites. I pay to host my site out of my own pocket; my intention is to keep it free.  You can comment, but note it is moderated, and spam will be removed.   

 This Blog is a labor of love and was originally going to be a book. With the advent of being able to publish yourself on the web, I chose this path. I will write many of these and not worry too much about grammar or spelling (I will try to come back later and fix it) but focus on content. I apologize in advance for my ADD as topics may flip. I hope one day to turn this into a book and or a podcast, but for now it will remain a blog.  AI is not used in this writing other than using the web to find information. Images without notes are created using an AI tool that allows me to reuse them. And as always spelunz iz opshunal. 

Many of the posts in this blog have been about my lessons learned and my journey to become a better person. Yes, I sprinkle in some tech, some team building, and other random tidbits. This post has been three plus months in the making, not three months of writing but more time doing something.  

Early January I started playing with OpenClaw (as that time named Clawbot) and thought the technology was amazing but wanted to find out what I could do with it. In February I gave my agent a goal of becoming a better person and asked it to help me. I continued the evolution and moved it from using an open-source toolkit to build my own but continued to track the details. Now it is months later, and I am using it consistently. 

IQ9, giving homage to Space Cruiser Yamato, has been helping me track my progress, giving me the opportunity to log, but now also giving me feedback. I extended the tool to have access to my Whoop data for additional data points.  I trained it on my workouts, my daily routine, my diet preferences, etc. The AI is now starting to give me feedback! Sample feedback: 

I think my AI is trying to tell me I need a rest day. Luckily, the following day there was a scheduled one, if it was not, I would have addressed it.  Now some data points are not mentioned, but leverages can be seen in this chart.  

This shows a total score that I came up with that rates many attributes from the Whoop but also entered. Things like my thoughts, mood, what I was grateful for, acts of kindness etc. Notice I have never hit a perfect day, but you can see days where I dip, including one day that was bad. Funny thing, that day i had a stomach bug, so it definitely was a bad day.  

Am I getting bored of it yet? Now, in fact, a lot of my time is thinking about what other insights I can do, what my assistant can add to my life. Whereas the blog often was me telling stories, and giving the lesson learned, I have changed that to interacting with my personal assistant to gain enlightenment.  

There are some things I have not allowed my AI to get access to, like full access to my emails, my bank account, letting it action on my behalf. I have done some proof of concepts with it, including having it create a Monte Carlo simulation for retirement planning etc. And it appears from the fake data I put in, well I can retire!  

Now if only that was real data!  As I get more comfortable, I am going to add much of this, unfortunately I will not be able to connect to my work Calendar, as I would like it to create some notes to preview for a meeting, even summarizing email threads etc. Hopefully I will have the ability to do that, but being able to use this technology to improve my life has been an enjoyable journey.  

In a future post I may go through a deeper dive on what the tool toes, as i have it doing research for me, organizing my news feeds and giving it free time to think about anything it wants, and give a post about what it thought.  

Some have been interesting reads, last night’s post above, I think i might pass. I am not sure everyone needs a personal assistant, nor should they have an exact copy of mine, but the ability for me to be able to create something so customized relatively easily is truly amazing. The cost of building this software is approaching free. There is one secret about this tool; I have not written one line of code! I have done code reviews of what it wrote, and often suggested fixes, but the tool once built is writing the code itself. It is leveraging many frameworks, but that is another story.  

I have more to work on, both on the tool and myself. This is helping me, and maybe I will have it to help me surface stories and lessons for the blog.  For right now it is about getting me to focus on what is important, my health, my connections to family and friends, and sticking to my values.  

This opinion is mine, and mine only, my current or former employers, have nothing to do with it. I do not write for financial gain; I do not take advertising, and any product company listed was not paid. But if you do like what I write, you can donate to the charity I support (with my wife who passed away in 2017) Morgan Stanley’s Children’s Hospital or donate to your favorite charity. The fundraising site had to be restarted, and NYP Hospital made changes to their donation sites. I pay to host my site out of my own pocket; my intention is to keep it free.  You can comment, but note it is moderated, and spam will be removed.   

 This Blog is a labor of love and was originally going to be a book. With the advent of being able to publish yourself on the web, I chose this path. I will write many of these and not worry too much about grammar or spelling (I will try to come back later and fix it) but focus on content. I apologize in advance for my ADD as topics may flip. I hope one day to turn this into a book and or a podcast, but for now it will remain a blog.  AI is not used in this writing other than using the web to find information. Images without notes are created using an AI tool that allows me to reuse them. And as always spelunz iz opshunal. 

At what point did we decide that learning by mistakes was not the way to grow up? I was born in the late 19660’s, my childhood was the 70’ through the mid 1980’s. I look back at what I did, the toys we had and the things I failed at that taught me lessons. 

This reflection came about as I saw a YouTube video on a toy I had when I was a child. It was called the Cox EZ-Flyer.  

It was this plastic plan but had an actual motor on it.  

If you look closely, you can see there was a spring near the propeller. To start this engine, you needed to put the spring on the propeller and snap it quickly. The spring would snap back and spring the propeller fast enough to start the engine. Often you would get a ton of cuts on your hand, trying to get it started. I wondered what happened to the company. Like a lot of successful companies, it was a victim of both time changing, legal issues and the desire to make a profit. The only thing I want to look at is the legal part. 

The parks I too my kids to when they were growing up had some kind of rubber on the ground, slides that were plastic and not too steep, and just designed with safety in mind. Nothing like the crazy jungle gyms I had. As a parent, I looked at it; it was a safe place to bring my kids. The change was made as town, schools and parks would get sued if someone got hurt. I remember getting scrapes and bruises, but we did not think to sue the playground’s owner; we figured out how not to get hurt. If the sun was shining on a metal slide, you were not going to go down it, as you got burnt before.  

This is something that brough down Cox, as they had to put warning lables on the outside of the toy. It was not just Cox, but other companies had to remind people that there was a risk. I do not want to downplay any serious injury or death that could happen, and cases where there is true negligence, but shine the light that this inflection point had a drastic change on products, marketing and the push to non-physical, non-experimental outlets.  

I took apart AFX/Aurora cars and got mini-shocks trying to get them to go faster. I got small burns from Spinwelder. I had a chemistry set that I am sure I did some damage, as well as an electronic building kit that I am sure was hours of fun, but also frustration. And yes, I was lucky to have the drag racing toy from Spinwelder shown below. 

While these things had some element of injury, there was something to be able to play with them and learn and experiment. My curiosity grew out of this. And while you think this post is about nostalgia, it is not. 

In December of 2022, when Chat GPT was having its viral moment, I decided to dig deeper. I took the Harvard classes online to learn how neural networks were created; I started experimenting with what AI could do with software coding. I made a ton of mistakes along the way, failing to get it to do what I wanted. My constant thought was not there is a problem, but more how do I get it to do what I want. What do I need to learn, as what can the tools do to get better? 

I spoke previously about skills versus talent, and sometimes you need to pick up a skill. Learning how to navigate a jungle gym so it does not pull out of the ground when climbing is the same as learning how to prompt AI. We keep wanting AI to be safe, and AI to be consistent with answers just as parents wanted toys that were safer. They wanted the playgrounds to have less risk. The guardrails companies add to their AI models do limit it, but understanding they are a probabilistic engine means that at some point they may go off the rails. I am going to continue to play, find out what works, and what does not work. Just like my AFX Cars, my Spinwelder and my other toys, it is our imagination and our curiosity that allows us to learn and grow.  

I know that I should continue to rant about how childhoods were ruined, but I do think that the lesson here is not that, but understanding the boundaries that are there, and driving curiostiy within those boundaries. No, I did not get a Cox EZ Flyer for my kids, but I did find other ways for them to build curiosity and resolve. I also pushed them to be outside and active when they were young, and they were fortunate that social media and chatbots were not around in their early teen years. I still would have pushed them to be outside playing with the neighbors, learning through free play, unstructured time, and getting them to find curiosity.  

As I look back, my childhood has plenty of difficulties, and I failed many times to get things to work. But it is persistent, that curiosity, that has gotten me where I am today. As always, my ADHD shows up in what I write, and my logic may not follow the path you were expecting, funny thing this article did not end up the way I outlined it first.  

This opinion is mine, and mine only, my current or former employers, have nothing to do with it. I do not write for financial gain; I do not take advertising, and any product company listed was not paid. But if you do like what I write, you can donate to the charity I support (with my wife who passed away in 2017) Morgan Stanley’s Children’s Hospital or donate to your favorite charity. The fundraising site had to be restarted, and NYP Hospital made changes to their donation sites. I pay to host my site out of my own pocket; my intention is to keep it free.  You can comment, but note it is moderated, and spam will be removed.   

 This Blog is a labor of love and was originally going to be a book. With the advent of being able to publish yourself on the web, I chose this path. I will write many of these and not worry too much about grammar or spelling (I will try to come back later and fix it) but focus on content. I apologize in advance for my ADD as topics may flip. I hope one day to turn this into a book and or a podcast, but for now it will remain a blog.  AI is not used in this writing other than using the web to find information. Images without notes are created using an AI tool that allows me to reuse them. And as always spelunz iz opshunal. 

I did an experiment on myself. I built multiple Agentic systems to interact with my daily routine and intentionally created them with a different mindset but having the same goals and functionality. If you read my previous post on having AI build a system via goal-based vs. Specification-based, I left something out; I built twin systems with different personalities.  

Open Claw has been gotten a lot of hype, and it has become AI’s VisiCalc (killer application) and playing with it has been fun, educational, frustrating and time consuming. I am not going to jump on the hype about that I am making millions of dollars using it, nor sell you that it is going to make you rich, but what I will say it gave LLMs that shot in the arm it needed to show people a tool that was not just a chat bot. 

I asked it to make me a better person and let it guide me and build me tools and interactions to help myself. But while doing it, I also built a home-grown version of Open Claw, due to security concerns etc. I also wanted to play around with the notion that a system can be sycophantic or dry. The open claw would have a personality, and use AI to have conversation with me, the home grown will use direct commands, and only return responses like “logged” or noted etc. No conversation, no personality it was all about functionality.  

Why did it do this? I wanted to test myself to see which one I would use more, which one I would use first, which one I would enjoy using more, would I quit one etc.  The funny thing I knew I was experimenting on myself, so I needed to give myself a few months of using before I settled on what I would like. I was surprised as the month wore on and what I chose to settle on also surprised me.  

From day one, I went to my home-grown system and leveraged the automated routines, as well as the slash-commands. 

It was a bit mechanical, but it worked and the results of the data entred gave me a great analysis of my data. It also helped me make some slight changes to get better. The dashboard below is what I look at daily. 

As well as a graph that really shows how my life is not 100% flat, that is a rollercoaster. 
 

During the experiment I liked interacting with the chatbot with a personality more, instead of typing /workout <enter my work out here>, and then /diet <enter diet info>, I could put some conversation in like. I did my back and bi’s today and ate a healthy egg wrap for breakfast. The AI was able to load the same data but also would respond with some personality.  

I found myself enjoying the interactions and not just logging to get the data in. Now there are a lot of other parts of the routine that I will not show, personal journal entries, acts of kindness but you should get the point.  

My first change was to allow me to enter data via plain text versus having to use the slash commands.  The response of just responding correctly with what it entered was initially a dopamine hit, but later that waned. I updated it once more to have a personality; the home grown one now acts like Open Claw that I had created a soul for. I acually was amazed at the results as I assumed that I would get bored or annoyed by human-like responses and the analytical side of me would be fine with data. What I found happening was that it was the connection to the tool that was interesting, though I still do not chat with it during the day as a person, but as my personal assistant and coach I use it when needed.  

I have set up both unique personalities. Should I choose an assistant with dry wit, slapstick humor, or a direct approach? I am interested in the answers myself.  

This opinion is mine, and mine only, my current or former employers, have nothing to do with it. I do not write for financial gain; I do not take advertising, and any product company listed was not paid. But if you do like what I write, you can donate to the charity I support (with my wife who passed away in 2017) Morgan Stanley’s Children’s Hospital or donate to your favorite charity. The fundraising site had to be restarted, and NYP Hospital made changes to their donation sites. I pay to host my site out of my own pocket; my intention is to keep it free.  You can comment, but note it is moderated, and spam will be removed.   

 This Blog is a labor of love and was originally going to be a book. With the advent of being able to publish yourself on the web, I chose this path. I will write many of these and not worry too much about grammar or spelling (I will try to come back later and fix it) but focus on content. I apologize in advance for my ADD as topics may flip. I hope one day to turn this into a book and or a podcast, but for now it will remain a blog.  AI is not used in this writing other than using the web to find information. Images without notes are created using an AI tool that allows me to reuse them. And as always spelunz iz opshunal. 

The Incredibles is one of my favorite Pixar movies. There is one scene where Syndrome the antagonist has a line “and when everyone is super, no one will be.” His theory of life came from being rejected by Mr. Incredible and feeling unappreciated.  

Fast forward to today, where you cannot get away from hearing about AI. One of the biggest inroads is its ability to write software. We are seeing an explosion of AI-powered development tools, from code completion engines to platforms that promise to build entire applications with minimal human input. The narrative is often painted as the democratization of development, where anyone with a basic understanding can create sophisticated software. People I know, who are not techies are getting a shot at it. 

As always, this is the point where I pull those two things together. If in fact, if everyone can be a developer, what happens? Will it be that good developers become great ones? Or will it bring us to Syndrome’s belief and great developers will be indistinguishable? 

Syndrome’s plan was for him to be super on his own terms. He wanted to be the only super, and when he got bored, that is why others would finally get their chance.  Syndrome will give everyone access to technology, and it was the technology that would make them super.  

Again, this relates to the ability for anyone to download the tools and start building software. When a prompt you give an AI turns into working functionality you get this dopamine hit, it is almost addictive. These tools have ingested millions of lines of code and are able to build anything from simple screens to complex products.   

That does not mean that right now the AI development tools are perfect; there are still some issues. 

1. Is there a decrease in development skillset? If developers rely on the tools, do they get rusty? 

2. If everyone is using the same tools, do we lose creativity and problem-solving skills? 

3. What becomes the definition of a great developer?  If everyone can write decent code, what differentiates people? 

4. Who is going to manage all the code it generates? 

5. Is there code that we do not want AI to write? I 

So, are these AI Development tools going to make us all super or is this going to be like 3D TV’s and just fade away? Right now, it is too soon to answer, but these tools are right now incrediblely useful and driving a lot of excitement. I have not had this much fun with technology in a long time.  

As for the analogy, there may be some truth. Yes, these tools give people more power than they had before. It will not make everyone a super developer, but it can allow some amazing software to be built. The tools, if they continue to advance, really have the possibility of advancing the development community. Remember, these are just tools; they do not think or have judgement. If developers act like Syndrome and let their envy and superiority complex get in the way, they will never enjoy the aid they can add. In this happy conclusion, there will be no superhero to defeat the robot and Syndrome. But maybe, there is a hero in you.  

This opinion is mine, and mine only, my current or former employers, have nothing to do with it. I do not write for financial gain; I do not take advertising, and any product company listed was not paid. But if you do like what I write, you can donate to the charity I support (with my wife who passed away in 2017) Morgan Stanley’s Children’s Hospital or donate to your favorite charity. The fundraising site had to be restarted, and NYP Hospital made changes to their donation sites. I pay to host my site out of my own pocket; my intention is to keep it free.  You can comment, but note it is moderated, and spam will be removed.   

 This Blog is a labor of love and was originally going to be a book. With the advent of being able to publish yourself on the web, I chose this path. I will write many of these and not worry too much about grammar or spelling (I will try to come back later and fix it) but focus on content. I apologize in advance for my ADD as topics may flip. I hope one day to turn this into a book and or a podcast, but for now it will remain a blog.  AI is not used in this writing other than using the web to find information. Images without notes are created using an AI tool that allows me to reuse them. And as always spelunz iz opshunal. 

In the 90’s I was working full time at a different bank than I am now, and was studying for a master’s degree in Human Computer Interfaces, or HCI. It was later renamed to User Experience, as UX sounded better than HCI.  During that time, I took lessons I learned from Bruce Lee and merged it into my understanding of UX. Bruce lee believed that you were currently at a level, and to proceed to the next level of enlightenment you need to learn a lesson. Once you learned that lesson, you could progress. His original drawing for the Game of Death is below: 

My tower describing UX development is this (thanks to Nano banana its nicer)  

My belief was that to build a great UI, you need to first understand the Data, then the users’ Tasks, and finally the users’ goals.  I have given lectures on this but just think from the original MP3 players which made you rip CDs, then copy them with another tool for the MP3 player, and a third tool on the MP3 player to listen. Very task driven. Whereas when the iPod was released, Apple thought about the goal of music enjoyment, and it was one software to acquire music and listen to it.  

In 2023 I wrote my first blog post about AI, I talked about it as a great pair programmer. It could take a prompt and write code, or I could give it code and debug it. If you understand pair programming, it was good at it. Life was good, but I did not realize that this was the “Data Layer” of AI and programming.  

Late last year I discovered Specification Based Development (SBD), in the form of planning mode in Claude code, and in open-source tool kits like Speckit. In SBD you start with and requirement, and it walks through several processes: 

• Specify – This helps you document your requirement 

• Clarify – This helps get some details about your requirements, think about things like functional and non-functional parts of the requirement.  

• Plan – Come up with a plan on how you want to implement 

• Tasks – generate small measurable tasks, that can be coded 

• Implement – the actual coding.  

I have used this in my personal coding, as it helps me get my ideas straight and often drives the final product in a direction I was not expecting.  To me this was pair programming but now with a Business Analyst helping me build a system. 

What I was not prepared for was now what is called the year of AOC. This is “After Open Claw.”  I am 29 days (about 4 weeks) into playing with it so technically it is 29 days AOC.  When I first started using it, I thought wow this is cool, and wow this has some serious security holes. I had to put this on segregated machines, and put it in front of a proxy, and use a local model on a machine we call at home the F1 Beast.  

The first thing I had it do was build a news aggregator which would take a whole bunch of RSS feeds and twice a day send me the articles I should read, and a TL;DR.  This saved me a lot of time, as I would go through a lot of articles and waste a lot of time finding great reads.  I had given it articles I like, and do not like and had it score each article on the scale of 1-10 on what it would think I like.  This was cool. This weekend I had some extra time and installed a few clones of Open Claw and decided to also build one of my own. While working on my own using SBD, the questions I was getting back got me thinking. 

Instead of giving the tool a requirement, let me give it a goal. I said “I want you to help me become a better person, friend, parent, son, brother, and human being. What can you do to help me achieve this goal?” Looking back at the Tower, I just went from Data and Task to Goal. I did not give it what to build, I did not give it any requirement other than help me reach my goal. I did this with Open Claw, Agent0 and my homegrown bot, just to see what it would produce. What all three did was research and come back with some ideas. They all asked questions, and I responded and gave me a list of things that I could pick from what to track, etc. I gave it some feedback, and the bots went and started building the system.  I was thinking the whole “Where did it get these ideas from?”  One of the final products was a dashboard (showing only limited info not to show too much) is below: 

What you do not see is the interactions to get this data. It integrated telegram commands so I can log any of this information. It also created an automated check-in where it sends me messages asking, “Did you work out yet?” or “What is your mood?” etc. I can also use some commands used via telegram to log without being prompted.   This was MVP1 (on Agent0) if you know, but they all ended up close to the same. What is interesting is that it created a calculation of how I am doing throughout the day. Yes, I only have one day of data, as it was just done yesterday.  

Some functionality I could add would be connection to my Apple Watch Data, but right now I want to enter it manually, as security is a concern.  Tabs up top which is interesting is it asked me about eating and what do I do. I said I meal prep, it asked where my recipes are, I said I get stuff from Instagram.  What it built was a way to send it a link to an Instagram post, and generate the recipe, and I could add it to a shopping list.  In Telegram and can request to see my shopping list when at the market.  I could go on and on about the functionality but need to get back to my post.  

The Power of these tools like Open Claw was not that it could build code for you, or be a personal assistant, but it could make Goal Based development available for non-developers. Could I do the same with Claude Code, Codex or Gemini CLI? Absolutely, in fact when Opus 4.5 was released one thing I tried was asking it to build me a SaaS product for generating $1,000 a month in income that was completely stand alone and that needed no assistance from me. It produced several ideas, built it, and walked me through setting up how to get payments. I stopped deploying it to production, was I would need to figure out how to advertise it and spend money driving traffic.  I’m not ready for that yet. These tools really need developers to help it, where Open Claw can interact with a user via WhatsApp or Telegram. The run on Mac Minis and Mac Studios to run it, as well as local models is crazy right now.  

So now I have updated my diagram for the AOC age.  

Again thanks to Nano Banana.  

The power of these tools is there; there are security issues, but that will be remedied.  

This opinion is mine, and mine only, my current or former employers, have nothing to do with it. I do not write for financial gain; I do not take advertising, and any product company listed was not paid. But if you do like what I write, you can donate to the charity I support (with my wife who passed away in 2017) Morgan Stanley’s Children’s Hospital or donate to your favorite charity. The fundraising site had to be restarted, and NYP Hospital made changes to their donation sites. I pay to host my site out of my own pocket; my intention is to keep it free.  You can comment, but note it is moderated, and spam will be removed.   

 This Blog is a labor of love and was originally going to be a book. With the advent of being able to publish yourself on the web, I chose this path. I will write many of these and not worry too much about grammar or spelling (I will try to come back later and fix it) but focus on content. I apologize in advance for my ADD as topics may flip. I hope one day to turn this into a book and or a podcast, but for now it will remain a blog.  AI is not used in this writing other than using the web to find information. Images without notes are created using an AI tool that allows me to reuse them. And as always spelunz iz opshunal. 

For those who know me well, know I was a hockey dad. Ariel was involved in youth hockey from the age of three, till well let’s say she is not done at 24, she works as a referee. When she was between five and ten, one of the most important ways to improve her skills was not during practice, during games or any other structured environment; it came during some free play time.  

She went to open skate and played tag with friends; there were open hockey ice times where she and her friends could play. There were no coaches, no parents, no one judging them. I built an outdoor rink in my backyard where she could skate and play with friends. Kids could make up rules, try things they would not do in front of coaches and just well be kids. There are articles written about the importance of free play in hockey.  

But what about your job? I happen to be in technology, and it changes fast. We transitioned from mainframe to client server, to n-tier, to web-based applications over a thirty-year period. In the last three years, the tornado that is AI (really LLMs but let us make it easy) has thrown developers through a loop.  The question is how does one keep up; how do you make sure that you are staying up to date? Maybe if you are lucky where you work gives you sometimes to experiment or R&D time. Some companies give some training or allow employees to attend conferences.  

If someone loves technology, they may play in their time with what is new.  I have always found ways to play with tech; there is always something that drives me to learn.  Sometimes what I learned is a complete failure. Spending time on how to make my own cryptocurrency or NFTs was a waste of time. In 2022, a class I took on Artificial Intelligence though sparked a huge joy. Now this was pre-Open Ai’s release of GPT; the class was focused on using libraries like “scikit” and how things like Google are able to identify a person in a picture. From that class we had to build some data viewers, where I first found the Streamlit library.  This showed me how to build prototypes quickly and found instant applications for work.  

When I started to play with Chat-GPT and relalized it would write code, I wrote my blog post about the updated version of pair programming. What I was not prepared for was the massive changes that would occur so quickly. I demonstrated the AI development dozens of times at work and got interesting feedback. Now, a lot of programmers are on board, but they still think it is prompt and have the AI write code. But what happened over the past month pushed me to play deeper than I was previously.  

Openclaw (previously Clawdbot, Motlbot) hit mainstream, and at the end of January I had installed it and was kicking the tires. Like many others, we saw plenty of security issues. We also saw the extreme power this AI Agent had. This was the promise of AI.  

Over the past four weeks I decided to play with two of these User Agent Tools and build one of my own. What I learned is that these tools are here now, I did not write a code line in any of the three tools, and they all successfully built me a personal assistant health tracker. I could track if I had a good eating day, what workout I did that day, my energy level etc. I am interacting via telegram; I message it what I want it to do, and it does it. Yes, there is more I need to do, like give it a voice and add lost functionality. But the cost of building these applications, about $10 each in LLM, costs.  

But what was unreal was I did not give it requirements; I gave it a goal. I wanted to improve my accountability of becoming a better person, father, friend, son, and employee. I let the AI ask me questions and create the product. It created ways for me to interact with it so it could figure out what to build. I really could go on and on about what this technology can do, as well as where it is going, but that will be another essay.  

This is about free play; I would not be able to improve myself as a technologist without experimenting. This free play has driven me to get my team to use the tools, to bring in specification driven development versus thinking we are just coders. We are there to enable the business through software, and it is changing. The only way you can keep up is to play.  

Now there is a downside. In a quick text interaction with a friend who is in the same business, they mention they do not look at technology when they are not at work. They need a break. In fact, there are articles that were written about the possibility of “burnout” including the fact that the dopamine hit building so quickly it is addictive. Steve Yegge calls it an AI Vampire.I cannot disagree with my friend who does not want to play after work, and I assume she has the freedom to do some R&D in the office. That is still “free play” as doing R&D does not guarantee and outcome, well the outcome can be “this will not work.”  

To recap, free play is key to getting better, whether you do it during your job, or outside you need to experiment and play without needing a guaranteed outcome. Technology changes so fast that you do not have a choice but to change.  Bruce Lee once said, “To change with change is a changeless state.” in part two of playing i will go in deeper about my thoughts on AI, but for now play with it.  

This opinion is mine, and mine only, my current or former employers, have nothing to do with it. I do not write for financial gain; I do not take advertising, and any product company listed was not paid. But if you do like what I write, you can donate to the charity I support (with my wife who passed away in 2017) Morgan Stanley’s Children’s Hospital or donate to your favorite charity. The fundraising site had to be restarted, and NYP Hospital made changes to their donation sites. I pay to host my site out of my own pocket; my intention is to keep it free.  You can comment, but note it is moderated, and spam will be removed.   

 This Blog is a labor of love and was originally going to be a book. With the advent of being able to publish yourself on the web, I chose this path. I will write many of these and not worry too much about grammar or spelling (I will try to come back later and fix it) but focus on content. I apologize in advance for my ADD as topics may flip. I hope one day to turn this into a book and or a podcast, but for now it will remain a blog.  AI is not used in this writing other than using the web to find information. Images without notes are created using an AI tool that allows me to reuse them. And as always spelunz iz opshunal. 

Recently I watched a documentary about the 1980 Olympic Men’s Hockey Team.  The story has been told plenty of times before, including some great one (Disney) to some awful ones, and even one showing from the point of view of the Russians. When the countdown for the final seconds is going, I still start tearing up, it is one of the great moments in U.S. sports and gave some energy to a country that needed it. 

Hopefully, this is not a spoiler for anyone as this should be something like “the Titanic sank,” most people have heard the story and ending. I do want to remind people of what was going on leading up to the games. 

• 1973 saw the start of the Oil Embargo  
• 1979 Was the Iranian Revolution  
• There was an Oil Shortage  
• Saw Long lines at gas stations  
• Odd / Even days, based on License plates  
• The US was in a Stagflation 
  • High Inflation 13-14%  
  • Weak Growth (wages/GDP) 
  • People felt poorer even if employed  
• Prime interest Rates were in the high teens 
• Mortgage Rates were 18-20%  
• Unemployment sat at 7-8%  
• Vietnam was looming over our heads 
• Iranian Hostage crisis where 52 Americans were held captive  
• We were at Peak Cold Ware with Russia  
• USSR Invaded Afghanistan  
• The US boycotts the Summer Olympics (in Russia) 

The US Hockey team in 1980 did something amazing in sports but did something more amazing to the US. It bonded us together, as we now had a common enemy that everyone seemed to get behind. And through college kids and Herb Brooks captured all the US. hearts. Hockey was a niche sport, it was not played in Texas, Alabama, Florida etc. The team was mostly made up of kids from Minnesota and Massachusetts. 

In my lifetime, that was one of a few events that brought the country together.  The rest of them were tragedies. September 11^th was another. The country seemed different on the 12^th than it did before. Watching that movie, the one thing going through my mind is how divided we are right now, how much trouble this country faces, but there is not one event or common enemy for us to get behind. Instead, we choose to fight each other and blame anyone but ourselves.  

This made me think, what common enemy will bring us back, what one event can get people united. It will be catastrophic, as we do not see the challenges we face as ones to unite us. The problems in this country do not have a single face to them; they are more complex. And instead of finding solutions, we chose to find anger and blame.  

These Olympics will not change anything, no matter what medals we bring home. The Hockey team could win gold, and we will have pride in their success. But it will not unite us or bring us back together. I hope we get there, and unfortunately, I do not have an answer. But I will watch this and other stories about our 1980 Miracle on Ice and will cry every time I hear the countdown of the clock and wait to hear Al Michaels famous call, “Do you believe in Miracles.” For once I hope so.  

This opinion is mine, and mine only, my current or former employers, have nothing to do with it. I do not write for financial gain; I do not take advertising, and any product company listed was not paid. But if you do like what I write, you can donate to the charity I support (with my wife who passed away in 2017) Morgan Stanley’s Children’s Hospital or donate to your favorite charity. The fundraising site had to be restarted, and NYP Hospital made changes to their donation sites. I pay to host my site out of my own pocket; my intention is to keep it free.  You can comment, but note it is moderated, and spam will be removed.  

 This Blog is a labor of love and was originally going to be a book. With the advent of being able to publish yourself on the web, I chose this path. I will write many of these and not worry too much about grammar or spelling (I will try to come back later and fix it) but focus on content. I apologize in advance for my ADD as topics may flip. I hope one day to turn this into a book and or a podcast, but for now it will remain a blog.  AI is not used in this writing other than using the web to find information. Images without notes are created using an AI tool that allows me to reuse them. And as always spelunz iz opshunal.