Latest / Tech Journeys Podcast / Tech Journeys Podcast with Stuart Milligan
Transcript
- 0:00Welcome to the Tech Journeys podcast.
- 0:03I'm your host, Vincent, and on this podcast,
- 0:05I interview people in tech.
- 0:07As always, this show is sponsored
- 0:09by the Consulting Launch Method,
- 0:11where I help experienced tech professionals
- 0:13build their own consulting business
- 0:16without spending years figuring out the business side.
- 0:19We've helped people go from their first idea
- 0:21to first client in 90 days and six figures in a year.
- 0:24To learn more, connect with me on LinkedIn
- 0:27and send me a message.
- 0:28In today's show, our guest is Stuart Milligan.
- 0:32Yeah.
- 0:33And he's the Solution Architect at Midrange Dynamics,
- 0:37where he invents, designs, and builds tools
- 0:40that are used to develop, manage, analyze,
- 0:43and modernize business applications
- 0:45written on IBM iPlatform.
- 0:47I thought that sounded very exciting.
- 0:49So, Stuart, looking back at your career,
- 0:54what's the decision or project that you are most proud of
- 0:57and what made it stand out for you?
- 1:02Probably, I had a company called Databrow
- 1:06and we had a product called X-Analysis.
- 1:08It was one of the original products that I built.
- 1:12Myself and my colleague, Mark Tregear,
- 1:14we built a technology that essentially
- 1:16forensically understood very, very large complex applications.
- 1:20So, the need for tribal knowledge and guesswork was gone
- 1:24and you could make forensically based decisions
- 1:29to change things, fix things, or replace things.
- 1:33And I'd say that that particular technology,
- 1:38it still currently recognizes the world leader,
- 1:41although subsequently I sold that
- 1:43and I'm building a new variant of it
- 1:45to keep track of all the changes
- 1:48that have changed in technology and applications.
- 1:50But that probably was the biggest one
- 1:52because it unlocked the potential
- 1:55of these large, complicated, legendary applications,
- 1:59as people call them, that run the huge tracks
- 2:02of the world's fiscal and operational and commercial systems.
- 2:06And so, building that was probably my proudest.
- 2:09But more recently, I've been rebuilding new technologies,
- 2:13but that was probably the most significant one that I'd built.
- 2:17And how long ago was this?
- 2:21It started in 1987,
- 2:24and then we sold the company in 2013.
- 2:28So, yeah.
- 2:30And subsequently I've been adding things on.
- 2:32Fantastic.
- 2:33And more recently,
- 2:35since your current role at Midrange Dynamics?
- 2:39Yeah.
- 2:40I mean, what happened with Midrange Dynamics?
- 2:42Because one of the other things that happened
- 2:44is that the legacy applications, as people call them,
- 2:47that run on the mainframe and midrange systems,
- 2:49largely, it was kind of a fortress mentality.
- 2:53It sort of went back to, socially,
- 2:55you might call it the feudal system,
- 2:57where during feudal and medieval times,
- 3:00everything was centered around the castle.
- 3:03So, all the trade happened there,
- 3:04and everybody operated from there.
- 3:06And legacy systems largely work that way,
- 3:09where the platform was designed to work
- 3:10in a sort of castle mode.
- 3:11It was isolated.
- 3:12But with modern application in the information age,
- 3:15you need to integrate.
- 3:17And more recently, after that,
- 3:20I've been building integration technology.
- 3:22So, enabling faster, more effective,
- 3:26more secure mechanisms to integrate
- 3:29large, complicated existing applications.
- 3:32So, systems of record, banking, financial, logistics,
- 3:35other operational systems.
- 3:37And that's what I've been doing since then.
- 3:39And the company that I'm part of now,
- 3:42Midrange Dynamics,
- 3:43the other key focus that we have,
- 3:45probably the primary key focus,
- 3:47is in the DevOps space.
- 3:48So, managing change in an efficient,
- 3:51quality-based, effective manner.
- 3:54Yeah.
- 3:55And this is your own company?
- 3:58I'm a partner.
- 3:59I'm one of the principals in the company,
- 4:01myself and a couple of other guys.
- 4:03It was originally founded
- 4:04by the chap by the name of Michael Morgan
- 4:06back in the 80s.
- 4:08And it sort of crawled its way to the top
- 4:10of being the world leader in this space
- 4:13for that particular platform.
- 4:15Exactly.
- 4:16Sounds very exciting.
- 4:17And the type of clients you serve?
- 4:21On the IBMI platform,
- 4:23it's everything from banks to municipalities,
- 4:27insurance companies, operational logistics, retail.
- 4:32Yeah, a bit of everything.
- 4:33It covers a very broad spectrum.
- 4:35Yeah.
- 4:36Fair enough.
- 4:37And if you were starting your tech career again tomorrow,
- 4:42knowing what you know now,
- 4:43with all the changes and transitions
- 4:47that are happening in this industry,
- 4:49that is never constant, right?
- 4:52Yeah.
- 4:53So, what is the one thing that you would do differently?
- 4:55One thing you would say to your younger self?
- 4:58It's probably learn more about the technologies.
- 5:02I think what we all tend to do
- 5:04is we sort of grasp enough knowledge
- 5:07of a particular technology
- 5:08to solve a specific problem.
- 5:10But rarely do we really absorb enough
- 5:12about the technology to solve a problem really well.
- 5:17And I see that in the industry a lot,
- 5:19where you see the evolution of bad decisions
- 5:23on bad decisions,
- 5:24not really because people are incompetent,
- 5:26but largely because they didn't quite understand
- 5:28the context or the purpose of a set technology.
- 5:31And it kind of just drifts off
- 5:33into what you call architectural evolution
- 5:35and a mess, usually.
- 5:37Yeah.
- 5:38So, learn more about technology before you start.
- 5:41An example of where you say it's sort of
- 5:44not worked out the way it should have?
- 5:46Well, I say it every day.
- 5:48I mean, literally for the last 30 years,
- 5:50I've been forensically analysing
- 5:51large complicated applications
- 5:53that have been around for 40 or 50 years.
- 5:55You know, really household names,
- 5:57large banking institutions, Europe, America,
- 6:00in the Far East, Africa.
- 6:02And I'd say that without exception,
- 6:05when I've run my technologies to forensically analyse,
- 6:09the observation that you make
- 6:13is that every single application
- 6:16looks like it started off
- 6:17with a kind of sound design principle,
- 6:19but then because of business pressures,
- 6:22turnover of staff, evolution of technology,
- 6:25investment in interesting things
- 6:27rather than practical things,
- 6:29just causes a mess to occur.
- 6:31So almost every day I look at the results
- 6:33of decisions that were made for
- 6:36sometimes the wrong reasons over the years.
- 6:38And usually it's about architecture
- 6:40or the application of technology in the right way.
- 6:43And so the disconnect between
- 6:45making a technological choice
- 6:47and what the business requirement
- 6:49is probably something that I see all the time.
- 6:51And when you look at something
- 6:53that's existed for 40 or 50 years,
- 6:55it's a bit like, say, I mean,
- 6:57you know, you live in Southampton,
- 6:58so go down to Southampton
- 7:00and look at the infrastructure
- 7:01and then try and go to see the council
- 7:03and ask them for a blueprint
- 7:04on how the electrical system
- 7:06has worked out in the city.
- 7:07And they'll sort of say,
- 7:08oh, not just because it's sensitive information,
- 7:11because it's difficult to get a hold of.
- 7:13And often decisions are made
- 7:15without that sort of knowledge.
- 7:18So, yeah, that's pretty much
- 7:20what I see all the time,
- 7:21which is, you know,
- 7:22it's just the way that the world works.
- 7:24And so understanding what we tend to think about,
- 7:27and I think it's what people think about
- 7:29and want to do is they want to build new things.
- 7:31They want to change.
- 7:32They understand change.
- 7:33They need to respond to change.
- 7:35But often what they're doing
- 7:36is they're sort of starting at a sort of a zero position
- 7:40rather than accepting
- 7:41that they're actually not starting from zero.
- 7:43They're starting from an inherited level of complication,
- 7:46sometimes because of pressure
- 7:47people build on top of that
- 7:48because they need to solve the problem quickly
- 7:50and they often don't revisit, you know,
- 7:52should it work that way?
- 7:53So you end up with these massively complicated,
- 7:55problematic systems,
- 7:57not unlike, you know, managing a city.
- 7:59And if you think about, you know,
- 8:00trying to fix the underground in London, for example,
- 8:03that'd be a complicated problem.
- 8:04And these large complicated applications
- 8:06that run the world commercially,
- 8:08they're no different.
- 8:09You know, it's people have changed
- 8:12over 30 to 40 years
- 8:13and you've had a massive staff turnover index.
- 8:15You've had an evolution of knowledge
- 8:17and skillset and approach
- 8:19and design principles and preferences.
- 8:21And you end up with this unholy mess that,
- 8:24you know, suddenly someone's charged with saying,
- 8:26well, you know, make it work.
- 8:28Yeah.
- 8:29Yeah.
- 8:30It feels almost inevitable that situation occurs.
- 8:35Yeah.
- 8:36Yeah.
- 8:37There was a piece of work that I did
- 8:39with a chap by the name of Grady Boots,
- 8:41who's the chief scientist at IBM
- 8:43and an incredibly intelligent man.
- 8:45And, you know, he said
- 8:47there were nine things you can do with software.
- 8:49You can leave it as it is.
- 8:51You can kick it down the road.
- 8:52You can throw it away.
- 8:54You can ignore it.
- 8:55You can do it.
- 8:56And there's many things that you can do with it.
- 8:58And I think the reality is,
- 9:00is that I think what most people don't do
- 9:02is they don't accept that such an asset,
- 9:06if you like, actually has an intrinsic value.
- 9:09And they don't do that because they don't see it.
- 9:11They see only what is relevant to them.
- 9:14And I think that is a big inhibitor
- 9:16for a lot of companies and continually,
- 9:19you know, if you think that a modern business
- 9:21is completely dependent upon the information that it has,
- 9:24you know, it used to be that if you had the best product,
- 9:27but these days, you know,
- 9:28a lot of people can make products
- 9:30and a lot of people can make good products,
- 9:32but the people that have control over their information
- 9:35are going to succeed.
- 9:36And so the information system becomes critical.
- 9:38And when the information system becomes so complex
- 9:40that you kind of don't control it any longer,
- 9:43the business really starts to struggle.
- 9:45And that's, you know, a lot of what we see today.
- 9:47And a lot of the, you know, startup companies,
- 9:50you start to see them reach a particular plateau
- 9:53where they get so big and their architecture is so complex
- 9:56and it's suffered from architectural erosion
- 9:58that it really does start to hamper the business.
- 10:01So, yeah.
- 10:03Fantastic.
- 10:04I can see it's happening basically.
- 10:08I've been in tech for 30 years
- 10:10and it's like, you know, code upon code upon complexity.
- 10:14Yeah. And idea upon idea.
- 10:16And it's not that ideas are bad or innovation is bad.
- 10:19It's not that at all, but it's innovation in context.
- 10:22And I think what a lot of people do is
- 10:24because they want their job to be interesting and exciting,
- 10:26they'd rather build something new than fix something old.
- 10:29Yeah.
- 10:30You know, not always because they're irresponsible,
- 10:32but sometimes because something is so old that you can't fix it.
- 10:36Yeah. And now with the different characters,
- 10:39different people are different characters.
- 10:41Some people are finishers.
- 10:42Some people are just simply starting projects.
- 10:44Absolutely.
- 10:45Ideas, people, completers, supporters. Yeah.
- 10:48And also the culture within the company.
- 10:50You know, you get different type of people, don't you?
- 10:52You get type of people that can sustain a business
- 10:54and you get people that can change a business.
- 10:56And the same applies to the slightly smaller microcosm
- 10:59of an application development stack.
- 11:01You know, someone that can change something is different
- 11:04from someone that can build something new.
- 11:06And often they struggle flipping between the two.
- 11:10Yeah. I agree with you.
- 11:13So what does the next chapter of your career look like?
- 11:18Well, I suppose AI was a very interesting injection.
- 11:23And interesting for me,
- 11:25it was kind of not as dramatic as it might have been
- 11:29because what I've been doing is developing algorithms
- 11:33based upon generic abstractions or models
- 11:38of an existing piece of language, which is kind of what AI is.
- 11:42You know, a large language model essentially is an abstraction,
- 11:45one level above the implementation,
- 11:47based upon a prediction of what that particular piece of language,
- 11:52word, you know, token, what it means.
- 11:54And so I've been writing software that does that
- 11:56so that you can actually then come up with a model
- 11:58and then make decisions based upon what you see with that model.
- 12:01But what I see happening is that people feel that AI isn't really that.
- 12:06It's just some way of actually getting another programmer
- 12:09to work faster or more cheaply, not always more cheaply.
- 12:13And my focus, if you like, now is to exploit AI to the maximum,
- 12:20whereby instead of having to run 100 trillion lines of code
- 12:25through, say, Anthropix LLM learning model,
- 12:29which is an impossibility or never happen,
- 12:32is to provide technologies that make LLM more effective
- 12:36and basically like the spec savers of AI.
- 12:40In other words, we give focus.
- 12:42So instead of having a hallucination because they can't see the whole picture
- 12:46and it makes guesswork, you provide snippets of information
- 12:49in a structured way so that AI can then make good decisions
- 12:54based upon known proven architecture as opposed to doing guesswork
- 12:58and then spending forever trying to retrofit
- 13:01and understand what AI did in the first place.
- 13:04And so it's really underpinning the value of legacy applications
- 13:10by using automatically forensically gathered meta models
- 13:14that can be used in context of a context window in AI
- 13:18to make tasks more effective and more efficient.
- 13:21Because what's frustrating a lot of people is they think,
- 13:23well, AI can fix everything.
- 13:25And they chuck AI at a particular problem they've got
- 13:28and they think, wow, that produced something quite good looking
- 13:31and sophisticated quickly, but then they take forever
- 13:34trying to figure out how relevant and how it integrates.
- 13:38And so you can't say, well, every time I want to make a change
- 13:41in my HSBC banking application, I can't read 50 million lines of code
- 13:46because I'll never get anything done, even with LLM.
- 13:48So what I've got to do is provide a sort of a metamap
- 13:51that can say, well, this is your starting point.
- 13:53That's what you're asking.
- 13:55How do the relevant bits count here?
- 13:57And only give me the relevant information.
- 13:59And giving that accurately is a very, very important factor.
- 14:02And that's something that I'm building and have built actually
- 14:05is the ability to map very, very large complicated applications
- 14:09into metamodels that make hallucination a thing of the past for AI.
- 14:15That sounds very exciting.
- 14:17And you're kind of already doing that with mid-range dynamics.
- 14:21And is that the future of the company, that direction you're taking?
- 14:26Well, it's kind of interesting because if you look at companies
- 14:28like Neo4j, Neo4j has been building metamodels
- 14:34kind of for not just software, but for business purposes.
- 14:38So they've kind of got the technology stack that can do it across the board.
- 14:41But I think we'll expand into the mainframe after this.
- 14:44But the IBMI marketplace is big enough and rich enough
- 14:48not to become so ambitious that you end up not achieving anything
- 14:55because your target is too big.
- 14:57So we're focusing on IBMI platforms at the moment.
- 15:01But we will be able to expand beyond that.
- 15:03And I think some of the patterns that we're using
- 15:06is already being used in various bits of technologies
- 15:09because we run over an entire platform where you've got to say
- 15:12complete application stack.
- 15:14It means that our tool set has to cover all bases.
- 15:17Whereas I think what you find in distributed computing
- 15:19in modern distributed applications that have become legacy,
- 15:22I mean, there's legacy Microsoft applications these days
- 15:24that have got hundreds of millions of lines of code.
- 15:28What you'll find is that because it's distributed,
- 15:30people are used to being able to take parts of it
- 15:33and analyzing that and keeping it in context.
- 15:36But large, complicated applications, where does the context end
- 15:40and where does it start?
- 15:42So maybe we can put it on any platform.
- 15:46But I would argue that probably the most benefit is going to be
- 15:49on the legacy applications, mainframe and IBMI.
- 15:53That's right.
- 15:54Well, we appreciate you being on the podcast today, Stuart.
- 15:58What's the best way for people to get in touch with you?
- 16:00They just go to our website, midrangedynamics.com.
- 16:04It's quite a mouthful, but I'm sure we can put that on the link
- 16:07of your interview here.
- 16:09And yeah, we're happy to discuss anything.
- 16:13I recommend people checking it out.
- 16:14Thank you so much for being on the show.
- 16:16And as a reminder to everyone listening, this show is sponsored
- 16:18by the Consultant Launch Method, where I help people
- 16:21experience tech professional, build their own consulting business
- 16:24without spending years figuring out the business side.
- 16:27We've helped clients go from a first idea to a third client
- 16:30in 90 days and six figures within a year.
- 16:32To learn more, connect with me on LinkedIn and send me a message.
- 16:35Until the next episode, thanks, Stuart.
- 16:38Thank you.