From Automation Leader to Solution Architect at Autodesk with Adam Sheather

Andrea Ciaffi

Hi, Adam. Welcome to the Digital Construction podcast.

Adam Sheather

Thank you, Andrea. Thank you for having me.

Andrea Ciaffi

Today we are joined by Adam Sheetal with a solution architect at Autodesk. I'm sure everybody knows Autodesk. We will learn more about his role and we will explore our automation and computational workflows are transforming the construction industry. I'm very excited for this conversation with Adam.

Andrea Ciaffi

And Adam, can you start telling us a bit about your career journey and what led you to work in the digital technology space for the construction industry?

Adam Sheather

Yeah, so my journey started off probably a little bit unconventional. I started off on site in scaffolding fresh out of high school. So I joined a scaffolding company and started erecting scaffold on the Gold Coast in Australia. Actually was in there for about three or four months and then realised.

Adam Sheather

I really suck at actually doing the work compared to a lot of people a bit bigger than me. I'm a very short guy, so anyway I moved in. I got the opportunity to move in the design and drafting team there, so I started drawing a lot of the temporary works and scaffolding for the projects over the Gold Coast and that was interesting.

Adam Sheather

Using some 3D technology built on top of audio. I had so that that kind of was a very different paradigm because I was drawing sort of 3D structures, you know, around 2000 and one 2002 doing quantity takeoffs and all that moved into structural engineering, picked up Rabbit, moved into a reseller, spent a few years there.

Adam Sheather

Sort of around that O5 to O7 upscaling a lot of architects, engineers and then moved into MEP as well, and then went to GHD for a few years over in Malaysia. So I managed A-Team there, more on the BIM consulting side, worked for a builder for a few years in Malaysia.

Adam Sheather

So a turnkey contractor. So again, very different experience, but helping that.

Adam Sheather

Me migrate from CAD to BIM processors was involved a bit more in the technology side, then moved into Acom where I LED a lot of the application pieces before then going out on my own and sort of building a technology consultancy which is still going today, which is autonomation and then BIM Bates, the software that I've developed with Conrad Sobin, is also still going.

Adam Sheather

And then. Moved on to Autodesk so that's been my journey. So a lot of varied roles, but still very centric around BIM and two D 3D workforce.

Andrea Ciaffi

Yeah, you've seen it all. You've seen all the aspects of the digital automation construction. That's very exciting. But at the moment you are at the desk and you're working as a solution arcade. So solution arcade in preparation for this episode, we were discussing that solution.

Andrea Ciaffi

Arcade could be different things. On where they operate, I can explain to us what a solution, what solution architects actually do and and particularly what do they do at Autodesk.

Adam Sheather

Yeah. So if you actually talk about your solution architect, what that what that higher purpose role is is it's really about bringing business and operations and IT strategy kind of get together where where you see a lot of disconnects in business is, yeah, and you see this a lot, a lot where people.

Adam Sheather

People comment on our people who don't understand what I do and you know I'm building all these. I don't know scripts or automation and no one gets it. And a lot of that is that's that's in the operational realm when they're speaking, the world of operations.

Adam Sheather

So and you hear this a bit, whereas in in the business side, this is about finance upcoming, you know targets and KPIs. And we've got. We're in work and we've got to pay a lot of people, so the the idea of a solution architect from an industry point is how you kind of bring those two together and create alignment between the business and operations within that.

Adam Sheather

There's a vast spectrum. So we have solution architects that are doing software development we. So we have some WHO plan a lot of the software development, so that will more look at an approach, say a customer has an integration problem. So this will be specked out with the customer.

Adam Sheather

We need to integrate with their particular technology stacks with our technology as well in order to achieve an outcome. And then at the higher level on the advisory side, it is about creating these sort of Rd maps to help business to understand operations and kind of target initiatives that are important in the business.

Adam Sheather

So that when you move into operations and you want to get an automation or or a solution developed, it's aligned with the business goals. So funding is sort of already mapped out. Out and there's these pieces.

Adam Sheather

A lot of the time you see a bottom up approach where they kind of hit the people kind of hit a ceiling, so to speak, when they're trying to get these things approved. And a lot of time, it's because they haven't really aligned with the business, if that makes sense.

Adam Sheather

So it's like, well, there's no funding on this. And why isn't there no funding because the work wasn't done on the business side to kind of allocate funding in those spaces. For improvement. So.

Adam Sheather

So yeah, so that it's a vast spectrum and depending on the seniority of the solution architect, I would say that's where they sit on their journey from a developer all the way up to that sort of advisory role.

Andrea Ciaffi

Yeah, it must be very interesting. And you get to see all of the different products and features that other desk is developing or planning to develop. And an interesting one is of course, we are not talking about anything confidential at this stage. But I think many people are wondering what is going on with AI and other desk.

Andrea Ciaffi

Is others doing anything interesting on the AI side?

Adam Sheather

Yeah. So it it's, it's interesting there because again, if you, if you look at just the A you just gone and I don't know if you watched the keynote or anything on that, there was a lot of AI technology actually released there. So there's a lot of buzz and chatter around what's what's happened.

Adam Sheather

And and I guess if you if you actually, if you haven't watched it go watch the Autodesk keynote and then there's also a session on AI as well for those who are interested that kind of gives you a better insight but to break it down, if you actually look behind the scenes like our our research team, I think the research team's been running about eight or nine years now.

Adam Sheather

I think that originally started way back in 2008 as like a computer. National. Team have now turned into this AI lab, but there's been a lot of papers that have been released, so 90 odd papers. Plus these are papers, are normally released to the AI industry.

Adam Sheather

So even though it might be an AEC centric sort of research topic on AI or machine learning, those are actually shared with other research institutes. And it's a very what would you say, computer science heavy read. So a lot of people just don't get involved with them because the underlying technology seemed very disconnected from the outcome in AEC.

Adam Sheather

So that's one part. Also, if you're familiar with hugging space, I don't know if you've heard of this. Autodesk has a channel there. So hugging for those who aren't familiar. It's an AI space. It's kind of like GitHub for AI for people who are aware and people put models up there, they put data sets up there.

Adam Sheather

So Autodesk has actually released like some of the largest constraint databases. When you're constraining, say, a model with a geometric system, things like that, a lot of those have been released by by Autodesk for others to train their. Our models on as well.

Adam Sheather

So again, if you're not in the AI scene at a really technical level, you're probably just not aware of these things happening. So there's a lot of having there. What that translates to in today, though, and we've been quite transparent around this because obviously there's a lot of worry around AI, some uncertainty around what it's doing, who's. Doing what?

Adam Sheather

So there is a website. If you go to our Autodesk trusted AI website, there is actually a breakdown of every feature we've created, like a machine learning model or an AI approach and map that to various features in the software so that people can get some real understanding of where it's used.

Adam Sheather

So it's not this all-encompassing thing. I mean it normally hides itself as little features and buttons and things like that that most people might not even associate is AAI, or if we use more technical language like a machine learning approach to predicting an outcome.

Adam Sheather

So constraining a stretch rather than using mathematical functions, we use a prediction function to do that.

Andrea Ciaffi

And expanding on these AI topic, we have seen the rise of AI agents. So many people may be familiar with that and I know you are involved in what is

Adam Sheather

Yeah.

Andrea Ciaffi

called this is another acronym, MCPS and others can explain what they are.

Adam Sheather

Yeah. So I think hopefully most people have used your LLMS that are out there. So and even that's probably the wrong word ChatGPT, Google Gemini, Claude, whatever it is. So people I assume have used those where they probably haven't done too much is other than chat.

Adam Sheather

Starting with them, you can integrate technologies with them, so e-mail or something like that so you can actually start it to getting to take actions on your behalf. So go create a meeting for me and add these five people to it. You know, go summarise this e-mail and put it in a Word document.

Adam Sheather

So those technologies that actually perform those actions, those are those are MCP tools. So essentially like it's like building a normal API or any sort of product solution that you might build on top of Revit or on top of Autodesk Construction Cloud or any solution.

Adam Sheather

But then. Instead of you calling those functions directly, what you do is you put a description around that function like hey go and get me all the projects on this hub for example and you put that description so whenever the user's talking to the chat interface, it knows that that tool is available.

Adam Sheather

So you could say oh hey, could you go give me a list of projects and tell me. All the ones. In Brisbane it would then query that that would then return that list. The LLM would be able to read that list of projects.

Adam Sheather

Hopefully the the locations are there and then return back in plain language text. Oh hey, here are the six projects we've currently got in Brisbane so when you when you start bringing down these are like controlled interfaces into other. The technologies so you can recall information or you could tell it to take action and that that's when you start thinking about the term magentic.

Adam Sheather

It really is just taking actions on your behalf.

Andrea Ciaffi

Yeah, that's very exciting. It's exciting to see how this evolves in the future. It's very powerful and you are, I want to say, an expert in automation. Many people suggested to speak to you

Adam Sheather

Yeah.

Andrea Ciaffi

about automation and I know you know a lot. And one of the things that I found always interesting is to. Find somebody who knows a lot about topic and say can you explain what automation in construction is to someone who is new to this topic?

Andrea Ciaffi

How would you explain that?

Adam Sheather

Yeah, to to explain it properly, I think I think the easiest way when you're talking to someone new because there's various levels of automation and there's complexities in that and it's just getting away from the magic. A lot of people, you know, AI is touted a lot, is a magic way to.

Adam Sheather

Do automation and that's not really it, it's just reducing reducing the low value tasks and not having to do them yourself anymore through a a computerised process or another process. So the easiest way is along with my very first job as I was saying with scaffolding.

Adam Sheather

Design I used to have to extract a Bill Bill of materials essentially that then needed to be transferred into a spreadsheet which then had all our cost items for each one, and then you could eventually create a a tender or a a total cost for that project.

Adam Sheather

Now that was quite a laborious task and the reason is is. The previous people had set up that process, hadn't thought of a way of just extracting that data, putting it into Excel, and then having sort of another Excel reference it, if that makes sense.

Adam Sheather

So it reads all the quantities automatically, so they were hand typing all that. So it was a very simple automation in. I didn't have to do any code, I just did some fancy excel. But what it actually did do was reduce the time of me doing that work from three hours to like 30 minutes per project, right?

Adam Sheather

So massive, massive efficiency. But that is that is an automation that is reducing my time. So that's an interesting part too, because a lot of businesses are time and materials, businesses, architecture and engineering you charge. Time. So it's also understanding low value time that you can't really charge back to the customer as well.

Adam Sheather

So trying to figure out a tender is an exercise you really want to automate because you generally can't charge for a tender trying to do a simulation of some time is usually a high value exercise. So if you actually think it through. Do you?

Adam Sheather

Do you really want to automate that? If the customer's finding it really valuable, or do you want to explore alternatives and and expand on that value to the customer rather than trying to shrink that value? If that makes sense?

Adam Sheather

So I know I've answered it easy, but it is just about reducing those those low value tasks, whether it's excel, whether it's a fancy coding or system, it doesn't matter. You're taking this much time and reducing it to this much time and getting the same outcome or better.

Andrea Ciaffi

And what are some of the common myths that you encounter when you discussing automation with people?

Adam Sheather

Yeah. So the biggest Smith I think is usually the one that it's only required for big projects or big initiatives that that's usually the biggest one that you see. So people worry that developing the automation is going to take a lot of time.

Adam Sheather

So that's that's probably one of the really big ones. And the other one is people seem really focused. I don't even know if it's a myth, but it's an interesting people seem really focused on automating really cool or interesting stuff all the time.

Adam Sheather

So like, I don't know, you've got some big fancy system and doing that but. I don't know if it's a myth, a myth, but I kind of say it's a missed opportunity. Like we've developed automations where you automatically download, say, a drawing package.

Adam Sheather

It's a simple tool that goes through all the PDS, figure out what drawings need to be superseded, what don't move the drawings that need to be superseded into another folder, rename the other ones with the right version. So. Very simple file tool, but it's astonishing to me how many people still do that manually because no one's seen the value in automating such a laborious process.

Adam Sheather

That kind of gets left. The contract admin does that, if that makes sense. So I don't know if it's a myth, but more of a missed opportunity that people seem very focused a lot of time on. Making really cool or interesting things where you really should be focused on some of those boring tasks that are out there.

Adam Sheather

And how do you solve those?

Andrea Ciaffi

And one of the challenges that people have, of course, is to develop computational workflows on projects, and particularly they're spun across multiple disciplines and project stages. But how do you, how do you approach the development of computational workflows for projects?

Adam Sheather

Yeah. So with any sort of computational workflow and and I guess, computational workflows are really just automations, but the the understanding is they kind of sit within rhino or grasshopper or or whatever solution you know sort of automating a process there and it's a visual programming.

Adam Sheather

But it's essentially still in automation. I guess the key thing to me is always understanding what is the purpose of any computational design Wi-Fi, and then what is the value. So a really good example is when Dynamo came around, which is a visual programming package for Revit, we saw.

Adam Sheather

A lot of people going and spending a lot of time creating, I would say routines that automate something that Revit already does, and because it's Revit, such a sophisticated programme, you kind of had users who weren't, you know, really, really knowledgeable with the software going.

Adam Sheather

Oh, I don't know how to do that, but I'll go away and. And programme this thing to kind of solve a problem that Revit can kind of do. So I think the first thing when you when you sort of approach any computational workflow is understanding.

Adam Sheather

Do you really understand what the tool is capable of that you've got right now and then understanding how you tweak that a lot. So you see a lot of people rewriting geometry systems or something when there's. There's a geometry kernel that you can take advantage of, so I've seen people like add in a third party library and do calculations on booleans.

Adam Sheather

When Revit has that all built in in house and so they build these complexities without realising it. So that's that's probably one thing is really making sure that you understand.

Adam Sheather

Not only the problem and the value of the problem that you got, but that the actual technology that you're building on because you can take a lot of share cuts by leveraging the underlying system and then with the development I think it comes down to if it's, it's the, the complexity is relative to.

Adam Sheather

The tasks right? So if you're doing a very simple task, so a good example is we we did run a few years ago, a scaffolding, ironically enough, a scaffolding tool where you could draw detail lines in Revit and kind of define your edge protection.

Adam Sheather

And then we would assemble the scaffold based on this customer's components to automatically assemble that. And they could then. And do a tender or something like that in 3D within Revit and that was pretty useful, but it solves a task that one user has.

Adam Sheather

So this is the problem again of where you're taking one task that one person has and just shrinking that that time to delivery. So that's quite a simple thing to do. Where it gets complex is when you start solving end to end workflows.

Adam Sheather

So. A good example might be structural integration with a drafting model or something, or whatever you want to call it. So you've got this process now where you've got a decision to be made of who's generating the initial model originally. So let's pretend it's E tabs model.

Adam Sheather

So first question is who who generates the model initially? When does that move back into Revit and then when do any changes that are either up made updated in Revit or in Etabs flow back either way and who's responsible for that? So those are the tasks that become quite challenging, not just because it's. Jumping between software or anything, those totals can be had.

Adam Sheather

It's really then understanding like how does this engineer work? How does the the technician work? Who is doing what, when? And then is that consistent across that business. So it's very easy and I find it all the time people will go and solve that for a project.

Adam Sheather

And you'll see this a lot. People will go create a computational workflow that's end to end and it works great. And you see this with a lot of complex learning and all of a sudden it falls down on the next project and it's it's not because there's anything wrong with the flow.

Adam Sheather

It's because the people have now been changed out. The process has now been changed, even though the project topology is probably the same. These create these inherent things that destroy these, so it's about taking that step back and really understanding.

Adam Sheather

Like are we trying to build this on a project level, which is fine, there's still value to be had there, but those are we trying to build this across multiple projects creates a whole nother level of complexity. I had a project with a bunch of engineers a couple of years back, some senior engineers doing some early calculations of structures.

Adam Sheather

I think it took us about nine months to agree on the spreadsheet that they were all happy to use, and then even after that, I think 30 to 40% of them maybe actually used it. And the other guys just refused. So there's a lot of complexity in the people piece rather than the computational piece, if that makes sense.

Andrea Ciaffi

Yeah, yeah, totally. And I'm thinking what you just said, make me think about my out of days when people wanted to develop tools every time. And you touch on this. You said sometimes it already exists. It's already available. Somebody may have developed that already.

Andrea Ciaffi

Potentially it's already in private. And we have this. Philosophy. That said, adapt first, then sorry, adopt first, then adapt. If you couldn't adopt something and then only then invent so it was. Adopt, adopt, invent. That was the philosophy to make it very simple for people to understand, explaining that like you have you work

Adam Sheather

Yeah.

Andrea Ciaffi

with different types of roles. We work with builders, we work with designer with contractors, I imagine. Working with them could be potentially slightly different or very different. So how do you tailor your automation solutions to meet the specific needs of these different parties?

Adam Sheather

Yeah, the the the difference probably between the main difference I think between say contracts and contractors and architects and engineers, when you're working for them is down to the business model, right? Like it sounds pretty fundamental and maybe it's too abstract.

Adam Sheather

But if you actually think about it like your average builder is working on a fixed on a fixed price scenario or some sort of so they've come up with a number, they've given it and they're working to that number, right. So in those cases, any opportunity that you have to do, a workload that saves them money is.

Adam Sheather

Automatically money back in their pocket, right? So a good example I was involved in a project in Malaysia a few years ago and we got the formwork design in these horrible drawings. One of those automated systems that spat out, that didn't make any sense.

Adam Sheather

We got that back. And we we started looking at me like, oh, this is a mess. Let's redraw it in Revit so we can actually make sense of it. So we spent about two weeks redrawing this, and then we realised through it that the actual formwork that they proposed had 85% custom components, right?

Adam Sheather

So with a lot of these formworking systems and there's like a lot of the standards. Out-of-the-box components that are quite cheap and then you have the stuff that's, you know, custom made that has a fee on it.

Adam Sheather

Now you know not a signing blame of how that design happened, but we were able to then take that just re redesign that ourselves, send it back to that particular subcontractor and go hang on. This is what we're thinking brought that customizer. Down by about to about 15% and we save basically the project $4 million in two weeks of work, right?

Adam Sheather

That is such an easy tangible outcome that when your boss goes back renegotiates with that builder, the money saved like everyone gets Pats, right? So it's actually quite elusive in. The design land because you're focusing on like you are selling time and materials, you might have a project point or you know you might come in with a contract value or something.

Adam Sheather

But in a lot of cases you are selling time and materials so you don't get that opportunity. You might do an incredible design or something like that. That solves a problem or achieves a problem, but you don't get that direct, tangible outcome if that makes sense.

Adam Sheather

Like you don't see that immediate impact to a solution that you create. So I actually find it's sometimes harder doing that computational design, except at the early feasibility stages, because that's a very generative process. Process where you're trying to take minimal inputs and like maximise a maximum output to come come up with a solution or generate a result so and the other thing that's intentional.

Adam Sheather

I think with designers is there's a focus on reducing time. So you hear this all the you know, imagine if we save 25% of drafting or 30% of drafting. Or whatever, but if you're if you're charging time and materials and hours your business does not really want to charge the client 30% less.

Adam Sheather

So you need and even when you do these hours don't really get captured either, so they get lost, right? Like someone. If someone saves 25% on drafting, it's not like the projects save money, though. That's fine. It gets reallocated somewhere else, and if you're not intentional with that, times getting reallocated maybe more into engineering, or maybe into something of value for the client.

Adam Sheather

That's where you might be more productive, but the project isn't actually seeing any value because that time wasn't intentionally recaptured. So I find that's the interesting part because. Lot of people. Gravitate to saving themselves time and what you should be sort of gravitating to and that's valuable for low value tasks, right?

Adam Sheather

Like if you can have less low value tasks and more high value tasks to a client but to a client up to yourself, that's where the value is. But I don't know if that makes sense. It might sound a bit abstract, but that's where I find.

Adam Sheather

Intentional kind of Peters out sometimes because people get so focused on whatever tasks they're trying to automate, they don't think about that in terms of value to the client. So they sometimes get upset when things get cancelled or it doesn't get used or it doesn't get the gravity that it needs through the business and they spend a lot of time on it.

Adam Sheather

So it becomes meaningful to them. So that's where you see some of these. Things Peter out, if that makes sense so.

Andrea Ciaffi

Yeah. And your experience? What barrier exists when trying to introduce automation into different teams and how do you overcome these barriers that you find?

Adam Sheather

I think when you're trying to get this stuff into different teams, there's a difference between servicing the project and servicing the users. So whether it's an automation or just a process change, a lot of people. Come in with something to automate something. Let me let me try and come up with a good actual.

Adam Sheather

OK, so. So a lot of a lot of projects right now require some sort of data assigning to all the elements, right. So we have a lot of government entities and asset owners now saying, hey, you need to attach all these attributes. To every single object in the project, right?

Adam Sheather

And that's cool. That's a client outcome that might be valuable to the client, but when you're forcing a user to now do all this work, that doesn't benefit them in any meaningful way. It's not helping the design, it's not helping them. They're not getting anything out of it.

Adam Sheather

If it's not getting. They checked properly. They don't even they get critiqued on it. It becomes very hard for those users to really care about making sure that process is done properly, if that makes sense. So a lot of the times, even with new technology solutions, so as a bin manager for example.

Adam Sheather

We did a lot of BIM management work for contractors, so we would come in that, you know the contractor might be naive to BIM and so they would bring us in to help support that delivery and the the focus is always on the client.

Adam Sheather

The client wants these things. The client wants this and the client wants that. Our customer is the contractor, so if we're not producing things of value for them during the process, if we're only focusing like, oh, let's make sure all the data's added to the thing or you know, now all of this data's added for the owner, the owner's kind of going to get that benefit.

Adam Sheather

But our customer, the builder's not going to get anything out of it. So they're less interested to engage. So we always tried to make sure that there were value. There was value for the customer, not just the end client of that.

Adam Sheather

But if that makes sense and if they if you can generate a value there even if they're not the end result through a process or an automation or something, they engage in it way more like if you had some data that I don't know.

Adam Sheather

To great. In to help them do a 4D process or, you know, assign some values for the concrete calculations. You'd be surprised how many like little simple things can be added as part of that experience to give the direct customer value too. So it's not just about ticking boxes in the bin management plan for the final outcome.

Adam Sheather

Then the customer's more happy to invest. And with and with the customer being the contractor to invest in automation initiatives because they're seeing the value as well. Because if that makes sense. So yeah.

Andrea Ciaffi

Yeah, I really like your very practical approach of the file. I can see that you've tried to embed this automation and digital workflows in multiple places, and you've seen it all.

Adam Sheather

Yeah.

Andrea Ciaffi

And one of the challenges is all the inefficiencies that we have when we try to develop new digital tools for. Projects and where do you see the big in biggest inefficiencies are and and and how can you help the teams to overcome them?

Adam Sheather

Yeah, tough question. One of the biggest inefficiencies. I think I think a lot of the times people try and build, how would I say it. People try and build solutions to tasks that they have problems with so. A really good example I think is this was like an early Dynamo script.

Adam Sheather

I think almost everyone built one back in the day. So when Dynamo first came out with rabbit, you had this ability where people would come in and you could build A level generator, right? So you had a little slider.

Adam Sheather

It slide it up and down, it would add, you know 10 or 20 or 50 levels or something like that automatically to a Revit project because if you'd ever done it before, you're copying and pasting and grouping and and all of this sort of stuff.

Adam Sheather

But as a result of that, you would end up with fifty of these in a firm. All of them slightly different, and none of them really achieving a real outcome, if that makes sense.

Adam Sheather

So I think I think one of the biggest inefficiencies when I when I'm what I'm trying to get to is like taking a step back and and it's hard when you're new because when you're new, you're still learning the tools. And and there's this difference between trying to master the tool and then trying to actually master an automation.

Adam Sheather

And and one of the things that you'll see that becomes an inefficiency a lot is people will start down, say, say they're they're doing something is almost a learning exercise. I'm trying to learn the tool. So I'm building this particular script or efficiency tool.

Adam Sheather

Learn the tool and they get to the end and they don't want to throw it away. So what happens? Oh, I've built this amazing thing for the company. Now it's got to go into the company if this makes sense and what you've actually taken is a learning exercise and then trying to jammed it into a production workflow.

Adam Sheather

Whereas being very intentional at the beginning is. It's about like, OK, what is it? We are actually changing. What is it we are automating? Are we adding a new capability? It's not always about efficiency, sometimes about creating new capabilities and what is the purpose and the value and then constructing from that and all too often the biggest inefficiency is people go, oh, that sounds cool.

Adam Sheather

As soon as I always hear the word. Oh, that sounds cool. It kind of triggers me as to like because I love doing cool stuff. I stuck doom in rabbit a few months ago. It was a bit of fun, but I'm all about cool stuff too.

Adam Sheather

But in terms of production workflows I think cool all too often leads to inefficiency if that makes sense.

Andrea Ciaffi

Yeah. And I'd like to touch about to, to, to look about the future and it's hard to predict the future, especially the future of innovation. It's very hard because everything changes so quickly.

Adam Sheather

Aye.

Andrea Ciaffi

But where do you see the next big opportunities for automation for for the industry?

Adam Sheather

Yeah. Well, I will definitely just preface this by saying I am not a good predictor of future technologies. Sometimes you know I got into rivis to a dynamite and things like that early, but on the counterpart to that, I would have sworn to you black and blue, that PDS would die many, many years ago.

Adam Sheather

And Bluebeam doesn't make any sense. So again, I am not a good predictor of these things like everyone else, but where I see the big next opportunities, I think it's gonna be in the AI adjacent space. So these are technologies that work with.

Adam Sheather

The large language models and these existing, you know, ML and AI technologies out there, but you're not necessarily building your own. So I don't think there's going to be a lot of value in building like another anthropic or another open AI or something like that.

Adam Sheather

But I think there's a lot of opportunities around this MCP workflow doing vision like you know these. Systems. Are getting better at reading drawings, interpreting drawings, understanding that. So there's this piece of can they get to the point where you can feed a drawings?

Adam Sheather

And this isn't about checking for every single, you know, Australian standard and did all the compliance but you can ask things like right now, which I've seen in people's QA cheques. Do you have the North arrow on every drawing that is?

Adam Sheather

Something that you can create an AI work play right now with open AI, feed it through and just go hey, do all the GAS have AI drawings. It'll cost you a few cents and it's done a lot more of the digitization I'm seeing.

Adam Sheather

So we did a project a while ago where we were reading particular values of engineering drawings. This is for over rail. Lines. Putting those values into Excel spreadsheets to calculate something and those will then drive a strand model.

Adam Sheather

So looking at those opportunities that have been typically much harder for us to do and AI's allowed that ability to like, read documents, read drawings, read read images, read photos, all of. This stuff and then translate a meaningful outcome. So I think I think there's a lot of value in this sort of AI adjacency space initially rather than the ******** building.

Adam Sheather

You know, your own ML models from scratch. So I think we've got a lot, lot, we're still just getting started. There's so much white space here. Yeah.

Andrea Ciaffi

Yeah, plenty of opportunity in that space and

Adam Sheather

Hmm.

Andrea Ciaffi

we are close to the end, but just one more question. One of the things that there are companies that already have established automation teams or digital teams, but there are always like new leaders, new managers that emerge and they try to figure out this space.

Andrea Ciaffi

But if you could give one piece of advice to a company starting now, their automation journey, or maybe they already have something that already exists, but they won't improve it. What? What advice would you give to them?

Adam Sheather

I would probably say you're in the best space that you could ever be. Like all the technology stacks that are out there right now are. You know, you can jump into AI, have a conversation about stuff, having it, having that first sketching of that, you know approach like whereas previously what you would have to do is a lot of research.

Adam Sheather

You'd have to go back through, so I think there's a huge opportunity now literally to go to ChatGPT or anything say, hey, I'm doing this. What's this approach? What does this look like? So I think he can get so far just going through that and defining a part and defining approach to your your automation journey, whatever that looks like.

Adam Sheather

But I would. But always bring it back to it. Is these core fundamental concepts we have a business and that has objectives around, you know, making money, being sustainable, looking after people. And then we have the operations and like the key is making sure those are always In Sync because that that leads to making sure that there's funding available for automation, making sure that.

Adam Sheather

There's funding for people and budgets without someone bridging that gap. That's where you run into problems. So the first thing I would almost always solve is how In Sync is your business and your operations and that's them both understanding at a you know at an interface level and where that interface can sometimes be tricky.

Adam Sheather

What is the other? Team is doing because when one doesn't understand the other, then that's where you start getting the disconnects and the friction we're trying to do various automation initiatives and it needs to be governed and managed the other way as well.

Adam Sheather

Like all too often people are afraid to fail and and in a lot of this space you need to embrace failure. Failure does mean that you cost the business. Some money I have done phase where investments were quite large that didn't turn out and this is part of it.

Adam Sheather

So you need to embrace both sides of this and be honest with each other and all too often one gets a bit skittish with the other, so that that's what I would probably say is a. Advice a really long way very first point is make sure business and operations are aligned and you can achieve much much bigger things.

Adam Sheather

If you're just operations doing it by yourself, you'll get you'll get to a position, but it'll be a very hard slog because you just don't have that the backing to unlock the resources and finances that you need to tackle the big problems. Homes.

Andrea Ciaffi

Adam, this was a fantastic episode. Thank you very much for your time. But if people want to reach out to you, where can they find you?

Adam Sheather

LinkedIn is my usual go to, I think, as with most people in this space in our industry, so feel free to reach out there and connect. I'm sharing usually random things. I haven't built anything random that I could share in a while, so I do need to get back into it.

Adam Sheather

But yeah, I do like to play around there and and message on that. So yeah.

Andrea Ciaffi

Thank you, Adam. Thanks.

Adam Sheather

Not a problem, Andre. Thank you very much for your time. Thanks everybody.