Making Room for AI Integration

Artificial intelligence (AI) is making a lasting mark on additive manufacturing (AM), accelerating automation, enhancing quality control measures, opening new doors to customization, and more. AI’s addition to the AM process is also benefiting companies by helping slash operational costs and reducing material waste. 

Looking at the AM market as a whole, some market reports put the global AM market size at $67 billion-plus in 2026, with expectations of growth to exceed $441 billion by 2035 with an estimated CAGR [compound annual growth rate] of ~23% from 2026 to 2035, according to Business Research Insights. In instances where AI/IoT is integrated with AM, the results to date translate to improved production, enhanced process automation and quality control, among other factors, the market report adds.

We talked to several vendors in the software and hardware space for their perspectives.

Authentise knows AI—its software is driven by it. The company, doing business since 2012, develops tools for data-powered workflow management and process automation in additive manufacturing. Its tools, Threads, Flows, and Digital Design Warehouse, were created with a goal to optimize the 3D printing production lifecycle, according to company founder and CEO, Andre Wagner. 

“We build software for additive manufacturing that uses AI to solve practical, high-value problems rather than AI for its own sake,” Wagner explains.

For example, in reverse engineering, he explains that the Authentise Threads software uses AI to “help stitch together fragmented drawings, generate bills of materials, investigate missing information, and draft Technical Data Packages” The Flows software also integrates AI to “extract structured data from documents such as test results, which reduces manual entry and improves traceability.”

At Materialise, a 3D printing software and services developer, Bart Van der Schueren, chief technology officer, says the company uses AI for production quality assurance and to improve engineering productivity and automation.

“On the production side, in CO-AM Quality & Process Control we apply AI to analyze in-situ process monitoring datasets (layer images and other sensor inputs),” Van der Schueren says. “This makes it feasible to interpret complex, high-volume build data consistently, helping move ISPM [in-situ process monitoring] from a research capability into advanced in-process quality control and potentially reducing reliance on post-print inspection.

Materialise CO-AM quality and process control applying AI to analyze in-situ process monitoring datasets. Image courtesy: Materialise

He further explains how Materialise views large language models as becoming “a practical ‘copilot’ layer, especially in tools that support scripting and workflow automation.” For example, for Materialise solutions like CO-AM Brix, natural language can allow for users to develop or change workflows by “driving the right API calls, with appropriate guardrails.”

Arvind Rangarajan is global head of Product and Strategy at HP Additive Manufacturing Solutions, a developer of hardware, materials, and services. He shares how at HP, AI already is built into several stages of the AM workflow: design, build preparation, production optimization, and quality assurance. 

“Our AI Text-to-3D technology lets users turn simple written prompts into ready-to-print models, cutting design time for complex geometries. HP 3D Build Optimizer also analyzes build data to refine packing, anticipate risk areas and improve both throughput and unit cost,” Rangarajan explains.

AI and Adaptive Manufacturing

As use of AI in AM becomes practically second nature, there may be a shift emerging, with AM moving toward a more intelligent, adaptive manufacturing process.

Materialise’s Van der Schueren notes that AI is already moving AM from mostly a manual, expert-driven workflow to a more data-driven, adaptive production approach.

“In production, AI enables more objective interpretation of in-process signals and faster decisions about anomalies and defect criticality,” Van der Schueren explains. “In engineering operations, LLM copilots can reduce friction in building automation: instead of searching through menus or writing boilerplate code, users can express intent in natural language and have the system generate executable actions through APIs [application programming interfaces].”

His prediction is that traditional user interfaces won’t disappear, but he suspects the future will have more of a hybrid approach. “Copilots help users reach a first workable result quickly, and then engineers refine, still applying domain knowledge and adjusting parameters as needed,” Van der Schueren adds. 

HP Additive Manufacturing Solutions has three focus areas when talking about use of AI: ease of use, productivity and manageability. 

“By analyzing sensor data and print-process variables, AI enables more accurate parameter tuning, faster root-cause detection, inventory management, and seamless support,” he shares. “This reduces TCO [total cost of ownership], improves repeatability, and reduces the need for manual intervention. Over time, these capabilities support more predictable production output, smoother scaling and better integration with wider digital manufacturing systems.”

At Authentise, Wagner suggests current use of AI is only the beginning of the road toward adaptive manufacturing, “Today, AI is mostly being applied to specific point problems, such as visual inspection, data extraction, or documentation.

“Over time, the bigger shift will be AI managing context across the full additive workflow, connecting data and decisions from design through production, post-processing, inspection, and documentation,” Wagner adds. “That will make the process more intelligent overall and allow downstream steps to adapt based on upstream insights. For example, what happens in printing should increasingly inform how a part is finished, inspected, or approved later in the process.”

Trending Factors

To date, certain trends surrounding use of AI in additive manufacturing may prove more beneficial than others. The experts rank their list of trends.

Reverse engineering and design intent capture (or the embedding of design requirements and manufacturing constraints into the digital 3D model) top Wagner’s list of potential trends regarding AI. 

He expounds, “If we can use AI to understand not just the geometry of a part but the intent behind it, we can begin to connect that intent to other AI systems in design automation, process selection, documentation, and ultimately qualification. That opens the door to much faster and more scalable redevelopment of parts across different manufacturing routes. It will take many systems working together, and strong human oversight, but the direction of travel is clear.”

HP’s Rangarajan sees the combination of AI-powered design generation with connected digital manufacturing platforms as a long-lasting influential trend. “AI-enhanced design tools can help engineers create optimized geometries more rapidly, including simulation-driven iterations that previously required niche expertise. 

“Another trend is the rise of scalable digital manufacturing networks in which AI supports distributed production, quality assurance and fleet-level consistency. This will improve the acceptance of additive as a true manufacturing modality,” he adds.

A hardware piece from Materialise Control Platform, an open, machine-embedded, hardware- and software-driven platform that gives real-time management over laser-based 3D printing systems. Image courtesy: Materialise

Van der Schueren of Materialise details several specific trends to watch: 

• Multi-sensor, multi-source correlation for higher confidence in decision making, especially for critical parts.

• LLM copilots that can take actions (not just answer questions). He gives an example of generating scripts or creating workflows via application programming interfaces in automation-driven solutions (such as CO-AM Brix).

• Orchestration/guardrails layers (model context protocol-style patterns) that “provide context, constrain actions, and ensure traceability, critical for manufacturing use where ‘free-form’ AI can otherwise skip important constraints.”

• Better validation and standardization, “especially where AI results need to stand alongside or reduce traditional inspection,” he adds.

Challenges to Overcome

To date, there are at least a few obstacles surrounding AI use in AM that are worth mentioning—and tackling, according to the experts. 

For HP, Rangarajan says a big challenge is determining the optimal use cases and assessing where there is true value add. “Not every workflow benefits from AI-based automation, so organizations need to evaluate where AI delivers measurable value rather than applying it by default. In an industry just recovering from inflated expectations, it is important that we build confidence in this technology by delivering value as you scale its relevance in digital workflows and physical hardware.”

Rangarajan also acknowledges technical hurdles relevant to AI, such as “early-stage AI-physics modeling, data privacy needs and the cost of collecting and managing datasets large enough to train reliable models, especially models that generalize well. Beyond technology, manufacturers must upskill teams, build trust in new processes and navigate intellectual property questions related to AI-generated designs.”

Wagner of Authentise says the greatest challenge remains a human-oriented one—trust.

“If AI is going to influence engineering or manufacturing decisions, operators need to understand why it is making a recommendation and be able to judge whether that recommendation is valid,” Wagner explains. “That explainability is essential, both for user adoption and for compliance with quality standards and operating practices. Without it, AI may generate interest, but it won’t be trusted enough to deliver real operational value.”

At Materialise, Van der Schueren says while the novelty of AI may pose a challenge, far greater are the issues of reliability and industrialization, and, as Wagner mentioned, that five-letter word again—trust.

“Manufacturers, especially in aerospace/medical, need evidence that AI-supported decisions are as reliable as established inspection methods, and they need standards,” Van der Schueren mentions. He also addresses issues surrounding natural language: “Natural language makes powerful changes easy, but it also increases the risk of asking for something that violates process constraints (e.g., minimum wall thickness). Users still need domain expertise, and the software needs an orchestration layer to provide context and safety.”

Lastly he cites a need for data and model governance at scale. “AI needs to work across different machines, materials, and environments, with controlled deployment (for example, research vs. production models) and traceability.”