How AI Agents Are Accelerating Digital Transformation in Industry

Imagine your competitor increases productivity by 30%, releasing product to market faster than anticipated. Not because they hired more engineers. Not because they outsourced production. But because they deployed software with AI agents. Software embedded with intelligence that automates routine tasks and unlocks insights, enabling their workforce to focus on value-add tasks.

Now, while your teams are manually tracing product relationships and coordinating data collection for a design change, their engineers are focusing on innovation. This isn’t science fiction. AI is reshaping industrial operations and manufacturers that harness AI agents will maintain their competitive advantage. 

Today, software R&D organizations are realizing double-digit productivity gains from AI code assistants like GitHub Copilot (McKinsey, 2023). Service departments are improving time-to-resolution with AI agents that diagnose issues and assist with troubleshooting (Microsoft, 2024). AI agents are transforming supply chains—optimizing inventory levels, managing suppliers, and streamlining logistics to reduce costs (McKinsey, 2024). 

More broadly, generative AI technologies enable users to interact with their industrial data through natural language, improving user experience. These AI systems can summarize vast amounts of data, accelerate analysis, and enhance decision-making—scaling access to insights across the enterprise. Furthermore, AI agents can work together, enabling smart, collaborative AI networks that can interact with data and workflows across software systems.  

In this transformative AI era, manufacturers will accelerate their digital transformation. Building on their journey from traditional methods to digitization, they will adopt intelligent software powered by agents, AI-based software services that can assist, augment and automate core product development processes, empowering people across the value chain to work faster.

PTC is uniquely positioned to enable manufacturers on this next phase of digital transformation. We have an award-winning portfolio of digital thread software that works across engineering, manufacturing and service. We have more than a decade of experience applying AI in our products. To build on this strategic positioning, we partner with hyperscalers like Microsoft and a community of manufacturers who validate product market fit.  

In the sections that follow, we unpack the building blocks of the AI agent tech stack, describe a three-part framework for scaling AI agent capabilities, and provide a conceptual use case in engineering. To conclude, we’ll explore the opportunities and challenges manufacturers will face on this AI journey. 

Digital Intelligence is digitizing product development across the value chain with AI-powered software. The foundation of this transformation is built on enterprise software for industry —such as ALM, PLM, CAD, and FSM—combined with an intelligent software stack consisting of AI agents. 

At the core of this digital transformation is enterprise software. PTC’s portfolio provides world class software across domains - from engineering to manufacturing to service – that manufacturers rely on every day to manage goods and services across industry verticals.

Each product enables critical business operations and manages the quality of data, essential for unlocking insights and driving automation with AI-powered features.

For more than a decade we’ve expanded the value of our software by applying AI. This includes predictive analytics for asset maintenance in ThingWorx, optimizing CAD designs with genetic algorithms in Creo, applying machine learning to manage spare parts supply chains with Servigistics, or using computer vision to detect duplicate parts in Windchill.

Building on this innovation, we are embedding AI agents and associated generative AI technologies into our products. For example, we’ve applied advanced agentic workflows in ServiceMax, are launching new offerings such as agents in Codebeamer, and continue to partner with our customers and partners like Microsoft to expand the value of our software.

To understand how AI agents embed into enterprise software, we break down the core technology stack into four layers—User Engagement, Application Services, Data Management, and Software Ecosystem. Each layer has specific technical considerations and unique implications for manufacturers.

There are three patterns to interact with AI agents:

1. Chat: A user request information or a task to be completed using a chat interface, such as a service technician using AI chat in ServiceMax to generate a work order summary. 
2. Actions: A user can trigger an agent workflow with traditional interfaces, such as a requirement engineer clicking a “Run Analysis” button in Codebeamer to prompt an agent to evaluate a requirement against INCOSE standards.
3. Autonomy: AI agents operate independently in the background, executing tasks without direct user initiation or oversight. For example, an AI agent in Windchill could proactively flag compliance risks in engineering change requests.

A key principle in these interaction approaches is seamless integration—ensuring AI features align with existing workflows, rules, and access controls. Equally important is trust and transparency, providing visibility into AI usage and maintaining a system of record to track AI-driven decisions and actions, especially for auditing autonomous agents.

Now that we see where users interact with agents, let’s dive deeper into how agents work, looking at both basic and advanced models, and how they can work together to create a network of intelligent agents.

Agents are application services that use AI to plan, reason and act. Agents can perform basic or advanced operations and can work collaboratively.

1. Basic Agents: There are basic agents that assist users with access to information, answering questions based on contextual data. An example is the Windchill Document Vault agent that enables engineers to ask questions based on information stored in data sheets, quality documents, test reports and other documents.

Windchill Document Vault agent with AI chat interface

2. Advanced Agents: In addition, there are advanced agents that can augment or automate specific tasks. For example, a service technician can ask scheduling questions using AI chat, and based on the user’s context, the agent can automatically create new calendar events. These advanced agents leverage LLMs to process natural language, infer user intent, and monitor task progress. Additionally, they can use generative capabilities to create a user response or generate code snippets that trigger actions in other tools or agents.

3. Multi Agent Architecture: Agents can work together. There are coordination agents that assign actions and monitor agent activity; they are like a team manager for a group of agents. The other agents are specialists designed to carry out specific jobs based on unique instructions. For example, ServiceMax implemented a Coordinator agent that coordinates several agents specialized for field service management (FSM). One specialist agent is the Service History Agent that can answer questions based on work order data, and another is the Schedule Management agent that can review a technician’s calendar and schedule events based on the user, work order and customer’s context.

ServiceMax AI: multi-agent architecture

Now that we've explored the different types of AI agents and their roles in enterprise software, let's turn to the data that powers their intelligence and enables them to take meaningful action.

AI agents rely on the data stored and managed in enterprise software, ensuring trust and actionability. Three elements –Vector Database, Semantic Layer, and APIs – are essential for implementing agents.

1. Vector Databases: Vector databases store both structured and unstructured data but are particularly valuable for unstructured content like documents and videos. They enable AI agents to search, summarize, and extract insights from files that were previously difficult to analyze, unlocking new ways to interact with enterprise knowledge. For example, Onshape users can ask training or troubleshooting questions without manually browsing documentation.

Use cases leveraging vector databases can be quick wins, as they require minimal specialized programming for Q&A chatbots. This is especially true if indexing mechanisms are already in place to manage documents such as Windchill’s Solr indexing engine. However, since many vector databases are cloud-based, companies must consider IP security and compliance when considering this technology.

2. Semantic Layer: The semantic layer acts as a bridge between complex enterprise data and AI agents (or other tools like reporting dashboards). It translates business-friendly questions into precise database queries. For example, if a Windchill user asks, "What open change requests affect Part X?", the semantic layer:

• Recognizes key terms like "change request" and "Part X." 
• Maps them to the right data objects in the system. 
• Generates a query to fetch accurate results.

By handling this translation, the semantic layer enables AI agents to provide clear, business-friendly answers while working with the complex data structures of enterprise software.

Another key benefit of a semantic layer is improved software integrations. When multiple systems have mapped semantic layers, AI agents can seamlessly navigate data across enterprise systems. For example, mapping the semantic layer between systems like ALM, PLM and FSM unlock new closed loop use cases such as an agent in Codebeamer processing issue reports from ServiceMax to inform a requirement update that in turn initiates a change project in Windchill.

3. APIs: APIs enable AI agents to retrieve structured data from enterprise software, perform semantic searches on vector databases for unstructured insights, and orchestrate actions across systems. More than just a data conduit, APIs allow AI agents to call specialized tools, trigger workflows, and interact seamlessly across the enterprise, ensuring they operate as intelligent digital workers rather than passive responders. As established tools in enterprise software, APIs provide a scalable and secure foundation for agentic workflows. Looking ahead, API monitoring and metering will become increasingly critical as AI agents operate across software ecosystems.

Effective data management is essential for AI agents to operate reliably and provide actionable intelligence. However, AI agents don’t function in isolation—they rely on a broader software ecosystem that provides the infrastructure and integrations needed to scale their capabilities.

AI agents do not operate in isolation—they rely on a broader software ecosystem to access data, perform tasks, and deliver business value. Three key players in this ecosystem are Independent Software Providers (ISVs), hyperscalers, and the manufacturer.

1. Independent Software Vendor (ISV): As AI agents become more prevalent in enterprise software, ISVs will adopt multi-agent architectures within and across software systems. The maturity model for ISVs typically begins with embedding AI agents into individual software systems, such as PTC’s CAD or PLM solutions. After this stage, there are two options. One option is to create agent integrations between software in the ISV’s portfolio, or target integrations between 3rd party solutions. There are cases where integrations within a portfolio are strategic such as leveraging existing integrations between Creo and Windchill to support design engineers. Or prioritization of Windchill PLM integrations with external ERP and MES solutions to support critical downstream operations for manufacturers.

In each integration scenario, high quality APIs, governance frameworks for AI oversight, and business models for agent usage will be essential for both the ISVs and the manufacturers.

2. Hyperscalers: Hyperscalers provide essential AI services and infrastructure for agents - delivering compute power, model hosting, and enterprise offerings at scale. Microsoft, for example, provisions over 1,800 LLMs, offers a comprehensive AI and data platform, Microsoft Fabric, and is building specialized knowledge graphs for industry such as the Manufacturing Data Solution. This is why PTC strategically partners with Microsoft, leveraging their cutting-edge AI infrastructure and expansive set of AI services that operate at scale and securely. Moreover, we see this AI-powered journey as a team sport, where sharing technologies as well as best practices is critical for supporting manufacturers as they adopt and scale AI within their operations.

3. Manufacturer: Manufacturers are central to an AI-powered software ecosystem providing inputs on where to realize business value as well as being a consumer and creator of AI-powered software. For manufacturers, AI adoption is driven by the need to accelerate product development and gain a competitive edge. They are laser-focused on identifying where AI agents can create the most value. By prioritizing high impact use cases, manufacturers not only shape their own AI strategies but also influence the broader market, guiding ISVs and hyperscalers toward the most valuable applications. As consumers, manufacturers aim to accelerate their time to value, leveraging purpose-built solutions. As creators, manufacturers develop their own AI agents, applying the same governance, integration, and scalability principles as ISVs to ensure seamless functionality across their IT landscape.

In sum, a software ecosystem is essential for AI agents, ensuring they seamlessly integrate across software systems, scale with cutting-edge and reliable infrastructure provided by hyperscalers, and deliver business value to manufacturers.

More broadly, we’ve covered the essential elements of a tech stack powered by AI agents. These embedded, smart and collaborative characteristics of AI agents is what positions this technology as a transformative force, leading many thought leaders to declare how agents are going to be ubiquitous in software.

We are in the early phases of applying AI agents in enterprise software. Although market analyst like Gartner predicts substantial AI agent adoption, targeting 33% adoption by 2028, current adoption is less than 1% in the enterprise (Gartner, 2024).

In order to realize this agentic world, manufacturers need to take an incremental approach, starting with quick wins and scaling success. As a guide on this AI journey, we recommend envisioning AI agents based on three key capabilities - Assist, Augment, and Automate - each building on the proceeding one with increasing value to business operations.

AI agents are not limited to one capability. An agent can incorporate multiple capabilities based on context of the use case and associated software environment. For example, an AI-powered engineering agent might retrieve relevant product specifications (Assist) but also suggest alternative materials based on regulatory compliance and cost (Augment). Over time, the same agent could be configured to autonomously apply approved design changes based on integrations to PLM and ERP systems (Automate).

Let’s look at each capability, highlighting their role, unique value, technology considerations, and real-world examples.

• Role: Agent as Intelligent Assistant - provides information, retrieves data, and offers recommendations to help users make better decisions. 
• Value Proposition: Efficiency - reduces time spent searching for data, simplifies workflows, and enhances knowledge accessibility. 
• Human Role: Fully in control - users drive decisions while leveraging AI for faster insights. 
• Technology Considerations: Low investment—requires integration with data sources, search/query capabilities, and often benefits from Retrieval-Augmented Generation (RAG) techniques. 
• Example: A Windchill agent helps engineers quickly locate specifications, compliance documents, or test reports by retrieving relevant information from the Windchill document vault. 

• Role: Agent as Smart Collaborator - Moves beyond assistance by actively executing some tasks or optimizing workflows.
• Value Proposition: Process optimization - reduces errors, improves productivity, and enhances decision-making through AI-driven recommendations. 
• Human Role: Human-led, AI-supported - users review and approve AI-suggested optimizations.
• Technology Considerations: Moderate investment - requires deeper integration with enterprise systems, sophisticated prompt engineering, and AI-driven decision support. 
• Example: A ServiceMax agent retrieves technician availability (Assist) as well as suggest optimized scheduling options based on technician expertise, location, and urgency (Augment). The technician still approves the final selection before it is booked. 

• Role: AI as Autonomous Operator - executes tasks independently, adjusting dynamically based on new data and workflows.
• Value Proposition: Scalability and cost reduction - minimizes manual effort, enables real-time process adaptation, and drives operational efficiencies.
• Human Role: Oversees AI actions, intervening only when necessary to handle exceptions or refine automation rules.
• Technology Considerations: High investment - requires agentic architectures, advanced prompt engineering, and possibly specialized machine learning models for autonomous operations.
• Example: A Windchill agent monitors requirement updates in Codebeamer, managing dependencies, compliance checks, and version control with little manual intervention.

By designing AI agents with progressive intelligence, manufacturers can implement innovation in a controlled, scalable manner, ensuring that AI adoption aligns with business goals and workforce readiness. In the following section we envision this future with an illustrative example in engineering.

Imagine a global manufacturer developing a next-generation software-defined product—an electric vehicle, a smart medical device, or intelligent construction equipment. The engineering complexity is immense, requiring seamless collaboration across requirements, systems engineering, and design teams.

To illustrate how AI agents can transform this process, we’ll explore four key use cases, showing the progression from traditional methods to digitized enterprise software to the integration of AI agents— enhancing enterprise collaboration, unlocking insights, and speeding up workflows.

1. Digital Traceability & Collaboration: Traceability agents monitor changes (Automate) to requirements, designs, and system models within ALM and PLM software, alerting engineers of changes and recommending updates (Assist). If a change is required, the agent automates the linking of new or updated objects based on the data ontology (Automate)

Engineers reduce time on routine data linking and focus on monitoring changes, ensuring quality and increasing collaboration.

2. Enterprise Change Management: Managing engineering change is complex, with modifications in one system affecting others. Change management agents analyze dependencies (Assist) across software, mechanical, and electrical domains, predicting the impact of changes before implementation and generate an impact report (Automate) for engineers to review.

Engineers quickly get insights on change impact, ensuring faster approvals and reduction of downstream issues.

3. Product Line Engineering: AI agents dynamically generate system models based on natural language descriptions (Automate), helping engineers standardize components and configurations across different product lines when changes occur.

Engineers reduce time drafting systems models and increase stakeholder alignment with dynamic updating.

4. Virtual Product Validation: AI automates the generation of test scenarios and validation workflows (Automate), speeding creation of downstream deliverables required for software and hardware components to meet compliance, safety, and performance requirements before physical prototyping.

Engineers improve the quality of product validation and decrease time on document-heavy tasks.

In this AI-powered transformation, AI does not replace engineers—it amplifies their impact, allowing them to focus on innovation while AI unlocks insights, optimizes processes, and automates targeted workflows. These AI capabilities are seamlessly integrated into existing enterprise systems, enhancing the value of established workflows and the data they manage.

In the following section we will close this article by sharing headwinds and tailwinds manufacturers face in this AI journey, and how PTC is a trusted partner, driving AI-powered innovation across the digital thread.

Digital transformation in industry is not a sprint; it’s a marathon. Organizations start with quick wins and incrementally realize greater value over time, scaling success across business operations.

The journey to Digital Intelligence builds on this strategy and like past efforts takes shape as a marathon but the route is constantly evolving. Advances in AI, regulatory changes, and shifting competitive landscapes require manufacturers to adapt and build organizational wherewithal to take advantage of this transformational opportunity.

In the next section, we examine the headwinds and tailwinds shaping AI adoption, highlighting three key challenges and opportunities manufacturers must navigate.

• Data Complexity and Management: AI thrives on accessible, organized data, but manufacturers often manage a patchwork of data maturity—from legacy documents on file drives to fully governed enterprise systems. Unlocking AI’s full potential requires getting one’s data house in order, a journey already underway with digitization efforts like transitioning from document-centric to part-centric PLM or shifting requirements from Excel to traceable systems like Codebeamer. With AI amplifying the need for quality data, ensuring enterprise readiness is more critical than ever. 
• Workforce Readiness: AI adoption is a cultural shift that requires new skills, process changes, and organizational alignment. A key part of change management is framing AI as an accelerator—helping employees work smarter and faster. This means increasing awareness, training, and best practices to ensure AI is used responsibly and effectively. For example, at PTC, we have a structured approach to workforce readiness, including monthly webinars showcasing AI innovations, department-specific AI workstreams that identify, validate, and scale AI-driven productivity improvements, and a library of digital resources—ranging from introductory AI training to specialized learning paths for different teams. 
• Regulatory Changes: AI is evolving faster than regulations, leaving manufacturers navigating complex compliance requirements. AI systems must be transparent, explainable, and aligned with standards (GDPR, ISO, EU AI Act). Legal and regulatory teams play a key role. At PTC we face these headwinds too; we took this challenge head on by building an AI Governance board focused on responsible use of AI, inspired by Microsoft Responsible AI Principles.

• Digital Transformation: The shift toward modern enterprise systems has been a long-established trend, with companies digitizing operations to improve efficiency, collaboration, and decision-making. By replacing siloed, manual processes with connected systems, manufacturers are creating the foundation AI needs—structured, high-quality data, integrated workflows, and enterprise-wide visibility. This digital transformation ensures that AI-powered capabilities, like multi-agent systems, can seamlessly interact with business-critical data and processes. Simply put, AI is only as powerful as the digital infrastructure it builds upon, making digital transformation a necessary precursor to unlocking AI’s full potential.
• AI Technology Innovation: AI has been evolving for decades, from rule-based systems to machine learning, each stage bringing greater automation and insight. However, the recent breakthroughs in generative AI have accelerated innovation at an unprecedented pace—unlocking new ways to interact with enterprise data, generate content, and automate complex tasks. These advancements are not just theoretical; they are actively transforming how manufacturers engineer, build, and service products by embedding intelligence into enterprise software. The rapid progress of multi-modal AI, reasoning agents, and self-improving models ensures that manufacturers will continue to find new, high-value applications for AI across the product lifecycle.
• Cloud / SaaS: The rise of cloud computing and SaaS delivery models has been a game-changer for AI, enabling organizations to access cutting-edge AI capabilities without massive on-premise infrastructure investments. LLMs would not exist without cloud-scale computing, which provides the immense processing power needed to train and deploy these models efficiently. Beyond AI, SaaS solutions accelerate innovation by delivering continuous updates, faster feature deployment, and seamless scalability—ensuring manufacturers always have access to the latest AI-driven capabilities without the long upgrade cycles of on-prem deployments. As AI adoption grows, the combination of cloud infrastructure and SaaS-based enterprise applications will be key to delivering AI’s benefits at scale.

As manufacturers embark on their journey embedding AI in their product development, PTC stands as a trusted partner, uniquely positioned to help organizations realize the value of AI.

With our industry-leading digital thread portfolio spanning engineering, manufacturing, and service, we provide the foundational software needed to integrate AI capabilities seamlessly.

Our deep expertise in AI, based on more than a decade of innovation, ensures that AI is applied responsibly and at scale to drive measurable business outcomes. With the introduction of AI agents in our software, we are bringing intelligence directly into the workflows that manufacturers rely on every day—helping teams work smarter and faster.

And through our strategic partnerships and a global ecosystem of industrial leaders, we enable our customers to harness the full potential of AI while ensuring security, scalability, and enterprise readiness.

The era of Digital Intelligence is here, and PTC is ready to lead the way. 

Join PTC at Hannover Messe

Our joint panel with Microsoft, Implementing Generative AI across the Digital Thread, is featured on the Digital Transformation Stage at Hannover Messe(Hall 17, Booth E44 - also available via Livestream).

Watch the Livestream