Why AI told them to delete every row

Thanks for the invite, Micholas.

All right, so I I could probably spend half an hour just answering that one question, so I'll try to be brief. short version, I started programming computers when I was 12, so that's over 50 years ago. Over the course of a time, I went to university, got my degree in computer science back in the early 80s. While doing that, I also realized that folks in the computer field had, let's just be polite and say, a poor reputation for being able to communicate and write. So I took a concentration in writing courses as well. As time progressed, I went from programming to the database arena. I went from programming, I went from working as an employee into consulting, went from consulting for a company to starting my own consulting company. my wife and I are founders. So I'm a co-founder and co-owner, just to get the record straight. We built the consulting company into something which very rapidly became a public company. And that happened because we were right in the internet startup area and people were throwing money at any sort of tech at that point. It it just had to sound like tech and you could get funding. we very rapidly found that dealing with lawyers and accountants and venture capitalists and SEC and shareholders was not as much fun as it sounded like. But we liked running a small company and we liked the challenges that the databases and database consulting in particular brought. So we left that company and started Soaring Eagle. Soaring Eagle started out with the idea that we would be doing training, consulting, and a little bit of software. Over the course of time, what we found was that we really liked the idea of managed services because of the recurring revenue model. So in consulting, you're selling yourself over and over and over. With the managed services, you're selling yourself once, and your customers, as long as you're making your customers happy, they stick around pretty much forever. And it provides a better growth approach. So We started moving into managed services several years ago. While doing that, we discovered a number of things, not the least of which was that automation, and you mentioned AI in our pre-talk, is important. We built a lot of things that automate what junior mid-level DBAs would do, which allow us to A have a higher price point for a lower cost than most of our competitors, but we're able to do things onshore and simply make ourselves look smart by never missing anything because the little things are all picked up and we know how to identify the things that are going to spiral out of control. So our customers tend to be very happy that we are proactive, which is very hard to find in the IT arena. You're probably aware of the 80-20 rule that you're 80% reactive and 20% proactive and everybody spends their lives trying to reverse that. I I think we've come very close.

gosh, that's another one I could talk about for hours. So let's start with some of the biggies. one of the biggest problems that people are simply unaware of is that small tuning mistakes snowball. So you've you've heard of all of the companies that start almost literally in somebody's garage and they get a bunch of customers, and then they sort of plateau. one of the reasons they plateau is they'll get 10 or 20 or 30 customers, all of small to medium size. Then they find that big customer, that whale, and they realize that 30,000 rows of data does not tune the same as 30 million rows of data or 30 billion rows of data. They don't know how to scale or escalate, and before they solve that problem, they're out of luck. another problem they run into is that. You know, Microsoft puts these ads up in Super Bowls where you float a balloon around in the server room and everything takes care of itself, even in the cloud, which is far from the truth. There's a lot of preventive maintenance that needs to go on. one of our Current customers found us because they discovered that they had a corruption in their database, which it had snowballed, and they had bad backups dating back six months. This was an insurance company that had to furlough over a thousand employees before they brought us in to solve that problem. And it's all something that should have been identified early and fixed early and never become a serious problem. So problem one, understanding scaling and understanding the database is an important part of the environment. second major problem, not understanding that the database profession is something that can't be passed on to a network administrator because they simply don't have the training to spot problems before they occur. Those are probably the two biggest. We can get into Data sprawl where you know companies have 50 or 60 or 100 SQL servers all over the place and don't know where they are and as a result don't control their data and as a result haven't everything there. we have a current customer that told us they had twelve SQL Server databases. we've discovered over sixty in their environment already and are are still discovering more. So a lot of those things going on.

Okay, so I have to preface this with my thoughts about what AI is are probably not the same as what the rest of the industries are. So I'm going to qualify and define things before I start getting into it. I grew up, I've been reading science fiction since I was five years old. So to me, AI is something that I can have a conversation with, that can make decisions without me and comes up with creative thought. We don't have that today. AI today really is more machine learning and the ability to aggregate data in a reasonable manner. That's the closest that AI comes to it. One of the I'll give you a couple of funny stories with AI in my field. We have a number of customers that have tried to explore using Claude instead of us as a service. And what we find is that the AI makes sophist responses. So they are absolutely correct but incredibly inappropriate. For example, we were all asked to do some tuning and they said, well, we asked Claude how to tune our 10 billion row table, and Claude said delete all the rows in the table. That will actually make the queries faster because the a query on a zero row table is very, very fast. But it doesn't solve the actual problem. So that's what I mean by sophistry, right? So the things to remember are that you know right now AI is a tool. It does not solve problems, it helps you identify problems, it helps you aggregate problems. But anything it recommends you have to look at very closely and test before implementing. from a practical standpoint, the things that we've built into our own software for managing multiple clients' environments meets The current definitions of AI. We've been using them for 10 years. But we're not using CLOG, we're not using anything else. It's all developed in-house. we do regression analysis, we do predictive analysis, we analyze trends and patterns across industries and across SQL environments. But that to me is not AI. Right now, what is AI being used for? It's a tool. It consolidates your emails, it listens to conversations like this and summarizes them. And it does it pretty well. but as far as the creative thought, as far as replacing people, not there. Flip side, one of the things that a Claude example has been able to do well, a good use case is it seems to be able to develop things like phone apps pretty efficiently. And I have a very close friend who developed a phone app just using Claude who hadn't written a program since she was in college 40 years ago. So there are some use cases for it. There's some validation for it. But I don't think it's gonna replace people in my field for some period of time. And I caution everybody to understand that AI is more of a marketing term than it is a facility. And again, reiterating, it is a tool and not a solution.

I do. So as I mentioned earlier, IT people have a reputation for not being able to have a conversation or do anything more than sit in their corner and do their job. I did not want that. So I started with creative writing in college. And after I got out, at that point, the whole idea of client server was fairly new. SQL still server was still running on OS two and really was not ready for. significant production environments. So as I was growing up in that field, I was able to be in situations where I was on I'll call it the bleeding edge rather than the leading edge of the technology. And people were hungry for that information. So I started writing some articles, which were well received. At that time, I was also doing a lot of technical training. We had written our own course material, which was far more popular than the vendor materials, because the vendor materials teach you what to do it and how to do it, but not why to do it. And we always brought in a very practical approach to doing our training. Well, back in the day, and we're going back to very early 90s here. I was on a message board. the in fact at the time I think it was CompuServe, and that's dating it way back, right? And somebody posted, Hey, we're looking for somebody who can write a book called the the DBA survival guide. And I replied with, Well, I have the background, the training, and I've actually written courseware to do exactly this. I could write the book. And they said, Send us an outline, and I did, and within a week I had a contract. It was the craziest thing. So I wrote that first book. It sold like hotcakes. we wrote a few more for the same publisher. And over the course of time, the the publishers are really happy with authors that actually sell books because a lot of people write books and nobody hears about them. But the vendors themselves were promoting it because That just made their products more effective, right? And I became popular along those lines because my name became very well known and the fact that we solved the problems became very well known. I have a memory of a call from Microsoft, directly from Microsoft. One of the product managers for SQL Server said, contacted me and said, Jeff, and this was through the publisher because he didn't know how to reach me directly, which was a sad thing. He said, Jeff, I've just read read your eleven hundred page book. I four found four tiny mistakes. Here they are for your next edition. And that kind of thing is just absolutely amazing from feedback, from hey, this is valid. Things along those lines. So the books have helped a lot because it's free credibility. If I walk into an environment, I have a a fun memory of walking into a technical environment. They said, Well, your reputation is good, but we have an obligation to do a tech interview. And I said, That's fine. And they sent me to the tech person who had one of my books on her desk, bookmarked to places with questions she wanted to ask me and I asked, Do you realize I wrote that book? And she looked at me and she looked at the the the author of the book and she just shook her said, I'm sorry, I have to do this anyway. the

It was a fun interaction, yeah. So I've got lots of stories like that, but the what the books did was add credibility. it reduces the clients need to go out and find somebody else to do it. But I I think the fact that much of my work is word of mouth. so I I don't need it a lot anymore. I've also trained about two or three thousand DBAs, so my reputation significantly exceeds what I actually do. I've been speaking at conferences for 30 years and and that does it as well. I'm glad I wrote the books. today I'm really not likely to write another tech book unless somebody writes a really big check for me, which has happened. But a lot of people these days get their information from Google, from Bing, from whatever, and they just don't read a lot of tech books anymore.

So you have to run a business which is about something you love. There's no substitute for that. If you don't, it's going to be mechanical and you might be marginally successful, but you won't be overwhelmingly successful. The second thing to understand, and most tech people do not understand this in their bones, is that sales and marketing is an absolutely critical component to building any business. when I started building mine, it was all by word of mouth, and I built a nice small sustainable business, fed my family very well, had a lot of fun. But until you start getting serious about sales and marketing, you don't grow it to the point where it is a living thing, it will live beyond me, it will continue to grow when I'm not giving it input. Those are all things which are absolutely critical to making a business grow. I have another memory of going to a prospect, a very, very large insurance company. And to put it into perspective, the CIO had a budget of a billion dollars a year. I mean that's a big IT budget. And he was fond of telling me that He was interviewing me for a critical position, but my cost was a rounding error in his budget. So that was the environment I was walking into. But the environment at the time, it was a Cybase environment, which very few had the expertise to manage. We had already demonstrated that we could solve his problems, which just shocked everybody because they've been trying to solve the problems for some period of time. And we were talking about bringing managed services for that environment. And I was told in advance that his first reaction was, no, not another mom, mom and pop company. So his questions were along the lines of succession. What happens if you and I and your wife are hitting hit in a or are involved in a car accident and are no longer available? am I out of business? Or do I no longer have support? So the fact that we had a succession plan was very impressive to him. And once again, books that I've read. That I r highly recommend. One of them is a book called Built to Last. I read that when I was in my mid-20s. And what it teaches you is that innovation is important, but it also teaches you that you have to look to tomorrow and the next day to make sure that what you're doing is sustainable. And that whereas you are going to be critical to the immediate success of your company, you need to make sure that you're not critical to long-term success. And that you need other people who can do what you've done. So have I answered your question completely? What have I done differently? yeah. the only thing I would have done really differently was found good sales and marketing much, much earlier.

No. Yeah, from a marketing perspective, I'm not an expert. I bring that bring in experts, I hire experts, and messaging is very important. what is important in the messaging changes periodically, so you have to be cognizant of that. understand that marketing sales are different skills. but messaging clearly is critical, and above and beyond that, I let the marketing people do it the marketing people. are trained for.

I wanted to be owning a very large building in New York, among other places.

I am very easily found on Google. my website, Soaring Eagle, is easily searchable. the website specifically is currently Sedatasolutions.io. My email is Jeff.garbus at secatasolutions.io. And if you don't want to type that much, you can also send it to Jeff at Sed.io. We've abbreviated it for folks who don't like to type, Jeff at SCDS.io, or just Google my name. I'm not that hard to find.

My pleasure. Thanks for inviting me, Nicholas. Have a wonderful day.