Get your digital house in order in preparation for a future with AI in Manufacturing

Hello, I'm Brett Hopkins, Senior Product Manager, Manufacturing Solutions at Autodesk. I have worked in manufacturing for over 30 years, focused mostly on implementation of emerging technologies and process automation. This work has given me a deep understanding of the effort required and the challenges that exist to successfully implement new technology and capabilities into a functioning system. This expertise serves me well today as I work on issues with digital transformation that is required to prepare the manufacturing industry for the challenges and opportunities that lay ahead.

Today, I want to talk to you about the current state of manufacturing, and what the future holds, and how we can make sure we are prepared to take advantage of the changes coming. First, let's take a brief walk down memory lane.

If we go back to the beginning of industrial manufacturing, we can start in the 1700s with the advent of mechanization, when we first moved from entirely manual methods to processes driven by external energy sources. A century or so later, we converted from water wheels and steam to systems driven by electricity and methods of mass production. These dramatically increased our production capacity and speed.

Fifty to 75 years later, we introduced automation, which further expanded our ability to produce at scale. Within 25 years, we developed the capability to drive these processes through digital means rather than mechanical ones. This improved precision, and efficiency, and gave us more flexibility to change our production systems more quickly to keep up with shifting market demand. Those digital systems created the ability to automate processes at a dramatically new scale and leverage the documented results in the form of data for significant improvements in using the past results to optimize future performance.

These disruptions have continuously transformed manufacturing, driving increased productivity and innovation. What jumps out to me with these key moments is that each of these disruptions have come faster than the previous one. In my view, the coming wave of AI is just the next iteration of this continuous disruption in our industry.

Since we find ourselves here in 2024, let's dig into the topics of digitization and automation, as these most directly impact us today. The first big move in the digital world was implementing tools that moved activity from paper, and physical models to digital representations, and allowed for more efficiency, collaboration, and synchronization of activities. Early on, we moved from the drafting board to CAD. Then we expanded these capabilities to ever-expanding areas of design, planning, and execution.

With our designs and processes digitized, we were able to build systems that could simulate and validate our work in the digital world before spending the time and money on costly tooling and production systems, all while eliminating many errors along the way. These systems, simulated forces, pressures, and manufacturing processes like machining and injection molding.

These systems are built on specific and validated mathematical algorithms. This makes them precise, predictable, and trustworthy. But because they run on specific mathematical algorithms, they do not evolve unless more advanced algorithms are created to improve their results.

More recently, generative design has been introduced. It dramatically expanded our capabilities to explore a greater range of options, expanding our ability to consider more innovative ideas. Generative design is still precise and validated design as it runs on mathematical algorithms. It also requires precise inputs and produces results that meet specific parameters. It just takes advantage of cloud computing to run a massive series of precise processes to arrive at a set of optimized results.

For most of its existence, generative design has been purely computational. We are now seeing AI capabilities being incorporated more deeply into many generative workflows, and the two capabilities are definitely converging.

Generative AI is the form of AI creating all of the excitement today. It is ChatGPT, Midjourney, Stable Diffusion and the wide array of other tools making the news. AI generates results based on having been trained on very large data sets. Considering ChatGPT, for example, the data set was the entire internet.

The way AI works is by identifying patterns within the data set it has been trained on. It then uses that pattern recognition to reassemble unique responses based on those patterns. Some argue that AI isn't intelligent because it doesn't really know what it is doing in the way a human does. But some question whether we as humans do any more than use our past experience to reassemble knowledge into unique responses. I'll leave you to judge the relative intelligence of humans and AI.

Regardless of that debate, AI's strength is its ability to quickly generate solutions from a loose set of instructions. This drives ideation at a scale we've previously been unable to resource. This makes it a powerful tool for exploration and brainstorming.

While often correct, the results it generates, are not validated and cannot be automatically trusted. We can't rely on those outputs alone. We need to validate those results using other means.

Machine learning is the AI I believe matters most for industrial applications. This is predictive AI. Machine learning typically works within a narrow focus and with specific parameters, using smaller, more specific data sets. Ideally, our own data sets that document our past performance. When you want to improve the existing processes, this is the AI that will have the greatest impact since it can explore past performance data to improve current processes, optimize future processes, and identify risks, as well as opportunities for greater efficiency.

Here's a few common examples of machine learning. It is used in health care today to read MRIs, identify potential areas of concern, and flag that for the professionals to focus their review on that specific area. This relieves much of the tedium of searching MRI scans all day long and relieves stress on the technician. In turn, this reduces errors and improves patient outcomes.

Machine learning is also used to review maintenance logs in a factory, identify common failure points, automatically order spare parts, and increase the frequency of routine maintenance in that area to reduce unplanned failures. It can also monitor production processes, identify bottlenecks and failure points, and modify workflows to optimize the efficiency. From these examples, you can imagine how many opportunities might exist for this capability to drive dramatic improvements in performance across manufacturing operations.

I believe the future will blend these capabilities to maximum performance and dramatically shift what is possible. AI will be used for the fast ideation and novel concepts that it can generate. Generative design might take those results as inputs and add further mathematical rigor, sorting out the concepts that fail to meet engineering and market requirements and refining the initial concepts into a more validated result.

Simulation can use those outputs and optimize the processes before we ever invest in the production hardware or run the process, increasing confidence. And machine learning will study the production processes in real time, optimizing efficiency and mitigating risks on a daily basis.

But there is an underlying requirement to make all of that magic happen and that is data. The more data, the better. And it needs to be current, organized and accessible. Manufacturing faces more pressures than ever before. We want to make better products, do so with less waste and environmental impact, and improve profitability.

But we need to understand every aspect of the product design and the production system in order to optimize results. And we likely need to do this with fewer resources than ever before. Supply chain disruptions, economic volatility, an aging and shrinking workforce, and environmental pressures are all creating challenges requiring a different approach. And they are all problems that data can help us solve.

We need access to current, organized data so that we can connect activity across product design, factory planning, production, and all other aspects of the business. Connected data drives collaboration, visibility, and good decision making today. And it's the foundation of every AI capability we've just talked about. Without it, we can't make the best decisions today. We can't gain insights into how we might improve. And little, if any, of the future potential of AI can be realized.

Despite those challenges and opportunities, 80 percent of business data goes unused today. Think of all the insights that are locked inside that data, all of the business opportunities that are blocked from being discovered. So why is data not used? In my view, it's due to a lack of an effective digital frame.

The reality too often today is still disconnected workflows between teams and stages of the product lifecycle. And the quality of the digital thread typically degrades as we go from design, through engineering, and onto production. By the time we get to the shop floor, we're reduced to paper documentation and whiteboards managing business-critical operations. We're left with an analog world with little to no data that can be used to drive optimization today and those future AI capabilities that are coming. Why is this?

Simply put, it's because it requires a lot of work and infrastructure. Connecting the data is a challenge. This structure shows an advanced connected data environment in the manufacturing space. It looks complex and it is. We talk about all the functions and the products that support each activity across the lifecycle. But we rarely talk about all those areas.

It's the arrows where all the problems are. The arrows are where data translation causes headaches. Separate departments have access or don't to different software tools. And there is often significant automation required to make all those arrows work effectively. And that's a challenge.

But with the pressures we've just talked about, we can't put off this digital transformation any longer. All of those industry trends mentioned earlier require a revolution in our data management capabilities so that we can leverage the insights locked within to competitive advantage.

We simply must implement a modern, collaborative, connected data environment to be competitive today and remain so in the future. For our part, Autodesk has assembled a team of industry experts, of which I am a part, to develop outcome-based solutions that address this complexity, provide clarity of the challenges present, and provide a roadmap for how we get the task done in a structured, logical, organized approach.

Here we see the entire life cycle of a factory. The white text around the outside represents key activities that occur throughout the life cycle, from design, plan, and buildout of the factory shell, to the design, planning, and operations that happen within it. The circular arrows in the center along with the blue loop, represent key outcome-based solutions that my team is developing to guide you on this journey. With that, let's dive in and see how each solution can help transform your manufacturing capabilities.

Let's first look at integrated factory modeling. Integrated factory modeling connects teams across the factory building and operations, enabling collaboration that can optimize utilization, efficiency, and resilience. In this solution, we're focused on connecting everyone from the architecture, engineering and construction activities of the building itself, as well as those working in the production and operations of the factory.

By connecting these groups and allowing visibility across these activities, we open significant opportunities to optimize the factory design and layout and improve efficiency of the operation, while also expanding the ability to quickly change based on external pressures.

Typical activities within integrated factory modeling are site planning, layout and configuration of the factory and production lines, construction, and other related activities. Tied closely to integrated factory modeling, but focused intensely on the production activities within the factory, is manufacturing capacity planning.

Manufacturing capacity planning helps optimize production capacity, resulting in improved asset utilization, reduced waste, and increased efficiency. Our goal in manufacturing planning is to improve decision making and optimize factory performance, identifying opportunities to improve efficiency and resilience in the production system. This phase of planning has a significant impact on cost controls, time to market, and competitive advantage.

Our focus is on optimizing utilization and efficiency of operations, as well as maximizing flexibility to adjust to changing market disruptions. Understanding production needs and where the unexpected challenges lie is critical to getting this right. And those insights are sitting in our data from past performance, if we have it.

Once all that planning, preparation, and construction is done, it's time to produce, which we can do on time and on budget if we get the last two areas right. Digital manufacturing operations integrates production activities into the data environment and uses that access to improve operational efficiency, decision making, and resilience.

Here we are in the day-to-day operation of a factory, and our goal is to make that go as smoothly as possible. We're focused on production management and risk mitigation, as well as remaining flexible so that we can react to unexpected change. As we've seen, utilization, efficiency and risk mitigation is the focus here and it is relentless. We need to work these issues every day to optimize our system and react to the unexpected.

The data generated here is also the pot of gold for everyone else across the organization. We can use it to optimize our operations today and improve them tomorrow. It needs to be shared with facilities teams so that they can optimize plans for future factories or the next major retooling. And the capacity planners can't optimize much without the data generated here to inform their activities. Otherwise, they're flying blind.

And that is where connected manufacturing comes in. This is foundational to everything else we've talked about today. Connected manufacturing provides the data environment that connects teams, enabling collaboration across the business to improve insights, optimize decision making, and increase predictability. Connected manufacturing connects users of data across the product design, factory planning, and production and maintenance. It provides access to accurate, current data for each user across the ecosystem, increasing transparency, which drives speed and reduces errors in decision making.

Connected manufacturing ties all aspects of the business together and provides each team with all the insights they need to make the best decisions possible. And to do so with speed and the confidence that the data they used was the most accurate and current available. This is also the key to unlocking the power of AI, as those tools become increasingly available.

Without a connected data environment in place, you'll be left using commonly available generic data sets to support AI. This is better than nothing, but it won't set you apart from the competition. The brilliance will be in leveraging the intelligence of those generally available data sets, then layering in the competitively advantageous data from your own operations. Think of all the insights you can unlock inside that data, all of the business opportunities that will be open for your taking.

The future of AI is full of opportunity, and we at Autodesk are determined to be your partner in understanding what is needed and how to take full advantage on your road to future success. With that in mind, let's look more closely at the structure of one of these solutions. I'll use connected manufacturing so that you can better understand how outcome-based solutions can meet your complex needs.

The solution content is always comprised of four key segments. The first is an introduction, which frames the issues and creates a common understanding, explains the challenges that we intend to address and how to best position your business to take advantage of the opportunity. It also explains why acting now is valuable and necessary.

This section will feel very similar to the presentation I've given here to this point. And in the interest of time, I'm going to skip over this section as we dive deeper into the solution content. Technology landscape is the next section, and it provides a high-level overview of the core technology involved in the solution and shows a simplified view of the solution architecture so that we can start to understand all those arrows in the slide we saw earlier, the arrows we rarely talk about because they are hard and potentially require a lot of work.

Next, we dive into the stages and workflows. This is where the details live. This is where we show how we solve all those errors. This area details what is needed to happen so that you can implement the capabilities and drive efficiency. Finally, we provide templates to help quickly and easily organize the planning and execution of the solution.

As I said, outcome-based solutions begin with an introduction covering the current state, the challenges faced, and what the future likely holds. This looks and feels much like the presentation you've seen to this point, just focused on the particular solution at hand. So I'm skipping this section at this moment.

Once we level set and position the content, we then take a high level look at the technology landscape. This section opens with a view of the core products involved. This list isn't exhaustive, but it helps to better understand the products that are generally involved and helps to explain what capability each is providing in this situation to deliver on the solution.

In this instance, we see that various components of Fusion, from operations and team, through manage and manufacturing extension, along with Vault, deliver the essential capabilities to support a connected manufacturing environment. We then dive a bit deeper to better understand what connections and workflows will be involved.

Here, we can better understand that we will have various teams working in their key software tools across the engineering, planning, and operations functions. The product design teams will be using tools such as Inventor. Separately, our manufacturing team will be using Factory Design Utilities for layout and planning, while Fusion, along with PowerMill and PowerInspect, will be actively used in the production teams.

Despite their very different work activities, these teams can all be kept informed and connected because they will all be collaborating through a connected product data management system to ensure product design data is shared and everyone is working in the latest version of that design. Those same teams will also be able to maintain alignment across the design and manufacturing lifecycle, using a product lifecycle management system to coordinate change orders, production schedules, and other operational activities.

Moving on to stages and workflows, this is where the details live. Here we start by showing what implementation might look like and the areas of the business and workflows that may be involved. These are laid out in a logical set of stages, but not all of these are necessarily needed for every situation. Sometimes you will already have some of these areas under control or maybe some don't apply to your business. But this is the roadmap that needs to be considered during the planning stages.

For connected manufacturing, we start with design data management. We consider collaboration capabilities with external parties, such as third-party engineering services and supply chain partners. Moving along from design, we now see the manufacturing stages come into the picture, managing all of the data generated here and maintaining efficient operational operations with connectivity between our data environment and the shop floor equipment producing the product.

We also consider the data connectivity that is needed in the manufacturing phase, where we may be connecting with the supply chain to ensure they have the latest design data, as well as maintaining visibility into their production and shipping details so we can be assured our operations remain on time and on budget, without unexpected delays or errors.

Additional stages detailed out are the quality control requirements, as well as formal processes for manufacturing change orders, including all of the requirements that go into managing these processes efficiently, while maintaining production without interruption. While this roadmap view provides an overview of the solution, each of these bullet points gets further detailed. So let's follow the first stage, design data management, as one example.

In each stage, we provide a more detailed review of what is involved, key points to consider and benefits it provides. To overview, each stage typically then has a video associated with it. This video shows the workflow in the software and is intended to give you more detail about each stage, while also giving you more clarity and understanding of all that is involved, and how it will feel to put this capability into use within your work environment.

In this example, the designer is working on a design update. Once complete, they check that updated design in the Vault. With the latest design-- with the latest design in Vault, all parties now have access to the current design iteration. Others can now access the design and pull details like the complete bill of materials from the current design and execute on their responsibilities with confidence that they are working with the most current data. Additionally, everyone can instantly understand the status of designs, whether it is a work in progress, in review, the current release, or obsolete.

Having reviewed each stage to understand how all of the components of the solution come together to deliver various capabilities, each stage then gets further detailed, building out the workflows involved. In the case of the stage design data management, there are three workflows that are further detailed, showing how to execute that stage effectively. These workflows are define the vault environment, design data management with Vault for on-premise instances, and design data management with Fusion team for cloud connected capability.

Here, we see the workflow for defining the Vault environment as one example. And you can see the key steps across this workflow from understanding what the current state of the data environment is through setup, testing, and on to the production rollout. After detailing all of the stages and workflows and gaining a deep understanding of the solution, we wrap up with-- we wrap up the content with commonly-used templates to help you more quickly and easily plan and track the work from initial planning through all of the execution activities.

These templates are for you to work with and fill out with the details of your specific project. They just give you a jump start, with templates like timelines, and implementation milestones, risk registers that provide insight and visibility to the common risks, and stakeholders for the project, so everyone knows who is involved in responsible. There are many other templates to further support the planning and execution of the project.

I hope that gives you a sense of the approach and content of outcome-based solutions. I think it is important to say we recognize that despite all of us being in the same industry, we are all experiencing unique challenges and our businesses are highly complex and very different. These solutions are not a product. You don't buy one and plug it in. Rather, they're a roadmap showing the path to being prepared for the future.

They provide an informed perspective on the pressures the industry is facing and provide some understanding of what to expect in the future. They then lay out a path to implement new workflows and gain new capabilities that will address those issues and prepare you to benefit from the disruption that is occurring.

Our intent is that with these insights, outcome-based solutions can provide the context to help you better understand the pressures you face, reduce the stress you feel, and provide a clear and organized plan to help you design and make a better world. Thank you.