by Gary Mintchell | Jun 21, 2026 | Generative AI, Manufacturing IT, Podcast, Services, Software
Nothing in industrial technology news annoys me more than the hype around artificial intelligence—AI. I recorded this podcast on the eve of the Automate trade show and conference in June 2026.
Looking to for realistic use of Industrial AI, I’m bringing in an interview with a practitioner. Bryan DeBois is Director of Industrial AI at RoviSys, one of the largest independent system integrators. He has 20 years in MES, historians, and plant floor software. He leads teams that operationalize AI and data infrastructure in live plants, working with the C suite and ops to turn goals into running systems.
We look at definitions of AI. Then turn to the technology development from when RoviSys developed its AI practice in 2019 pre-LLMs. RoviSys took autonomous AI beyond predictive applications. Hiring deep manufacturing expertise, they can use AI to assist the human in the loop to make constrained decisions. DeBois discusses real-world applications. He then leads us through the beginning of a project.
Watch on YouTube.
Listen on your favorite podcast app or on my Website.
by Gary Mintchell | Jun 5, 2026 | Software
Deepgram continues to roll out extensions to its real-time voice AI platform.
- Deepgram and Fortanix are the first to bring confidential computing to real-time voice AI, allowing organizations to protect sensitive conversations and AI models even while they are actively being used.
- Until now, you could protect data when it was stored and when it was moving. The hard part was protecting it while the AI was actually using it. That’s what we’re solving here for real-time voice AI.
- Sensitive enterprise data and proprietary model IP remain private during active inference, with no exposure to underlying infrastructure
From the press release:
Deepgram and Fortanix announced a partnership that will enable enterprises to run voice AI in their own environment on their own terms while ensuring their most sensitive data is securely protected. Under terms of the agreement, Deepgram can leverage Fortanix Confidential AI and NVIDIA Confidential Computing to add an additional layer of advanced security to self-hosted environments to ensure that its proprietary model weights, built on business-critical intellectual property, can be deployed while protecting against model theft or inappropriate use.
For enterprises, especially those in highly regulated industries, security requirements continue to tighten. Organizations handling patient conversations, financial transactions, or classified information increasingly require that sensitive audio and AI model weights remain protected not only at rest and in transit, but also during active processing in their own environments. This level of protection enables organizations to build highly-secure real-time voice applications without sacrificing on performance.
The on-premises solution runs Deepgram’s voice AI models with Fortanix Confidential AI on NVIDIA Confidential Computing-enabled GPUs, creating a hardware-isolated environment where both audio data and model weights remain encrypted and protected throughout active use. NVIDIA GPUs with Confidential Computing enable AI workloads to process sensitive data inside a trusted execution environment — a capability traditional infrastructure cannot provide. By bringing together best-in-class voice AI models, hardware-rooted isolation, and a jointly engineered, pre-integrated stack, the partnership delivers a level of in-use data protection that, until now, has not been practical to deploy at enterprise scale.
by Gary Mintchell | Jun 4, 2026 | Manufacturing IT, Software
Here is a new service—operational decision intelligence. Also a company new to me—SteelTree. They define an operational decision intelligence service as something designed to help industrial teams improve awareness, reduce friction, and coordinate action across fast-moving operations.
The information I could gather combined was very sparse. Not sure why it’s better or worth than anything out there already (unless the “.ai” means something new in AI. It could be worth checking just in case.
The company seeks to help teams move beyond disconnected dashboards, spreadsheets, reports, and silos to improve visibility, coordination, and execution across day-to-day operations. SteelTree enables teams to quickly identify changes, recurring issues, performance drift, coordination gaps, and priorities requiring attention.
The model consists of:
See → Decide → Execute → Learn
SteelTree helps industrial teams:
- See what’s happening
- Decide what matters most
- Coordinate and execute actions faster
- Continuously learn before small issues become larger problems
“Most teams are not lacking systems or data,” said Kanwar Arora, Founder of SteelTree. “What they’re lacking is continuous operational awareness across fast-moving environments. Teams still spend too much time moving between dashboards, spreadsheets, reports, and silos just to understand what requires attention. SteelTree reduces the friction between operational signals, decisions, and action.”
Unlike traditional BI, dashboarding, and reporting tools that often depend on analysts, dashboard development, and delayed reporting cycles, SteelTree is focused on helping teams maintain awareness and coordination without adding overhead.
The company believes many organizations still struggle with operational visibility and coordination despite significant investments in business systems and reporting tools.
“As the software industry races to embed AI across enterprise applications, many teams still struggle with a more fundamental challenge: maintaining awareness across fast-moving operations and coordinating action effectively,” said Peter Price, Founder of SteelTree. “SteelTree starts by helping teams see clearly, but visibility alone is not enough. The real value comes from helping teams decide faster, execute more effectively, and continuously learn across day-to-day operations.”
SteelTree’s launch is focused on industrial teams looking to improve operational awareness, decision-making, and coordination without the complexity typically associated with traditional enterprise analytics and reporting tools.
The service is available immediately with free access.
by Gary Mintchell | Jun 2, 2026 | Robots, Software, Technology
Manufacturers have lots of data. Every day brings new technologies for gathering and storing it. The right question probes into what specific problem can be solved. I sat in the world’s shortest press conference (not complaining, even though they blew off my question) with a company I don’t know who asked the question—how can we better integrate the myriad details required to build the best robotic workcell.
The company is called Robotiq. Based in Quebec City, Canada, introduced IQ, an AI-enabled platform designed to make robotic Workcell integration faster, more predictable, and easier to scale. IQ captures unstructured automation project data, coordinates engineering workflows, and helps partners generate validated Workcell designs based on real customer inputs and historical deployment data from thousands of previous factory installations.
“AI” can be a generic marketing buzzword. I asked for a definition, but the press conference closed before they got to it. Reading through the press release, the definition apparently involves machine learning algorithms. Fair enough.
“Automation does not scale when integration remains manual,” said Samuel Bouchard, CEO of Robotiq. “With IQ, we are moving from manually engineering robotic systems one project at a time to automatically generating Workcells from real customer inputs, Robotiq components, AI, and proven know-how from thousands of past projects. For manufacturers, this means a clearer path to automation: fewer surprises, faster decisions, more predictable performance, and better financial justification, including in many 1-shift operations.”
Robotic Workcell integration depends on thousands of small details. Customer requirements, production constraints, factory floor layouts, site measurements, throughput targets, product variants, and local installation realities all affect whether a project succeeds. When that data is incomplete, fragmented or siloed, engineering teams experience project delays during the discovery and design revision phases.
The IQ solution includes:
- Automated data capture: Extract technical requirements via voice notes, legacy file uploads, and 3D site scanning.
- AI-enabled project coordination: Machine-learning models align manufacturer specifications, partner capabilities, and Robotiq application engineering expertise.
- Simulation and design validation: 3D environment scans are converted into digital twin models, matching customer cycle times and application data against standardized engineering rules to validate Workcell performance before physical deployment.
IQ is available today for robotic palletizing applications, where Robotiq has already standardized the hardware components, software workflows, and deployment knowledge needed to generate validated Workcell designs. Over time, Robotiq plans to extend the same Automatic Integration model to additional robotic applications.
To commission robotic systems successfully, manufacturers need local system integration support, application expertise, and reliable service. Robotiq partners play that role. IQ provides partners with a repeatable digital workflow to capture project information, apply Robotiq deployment expertise, collaborate with customers and Robotiq experts, and support Workcells more consistently after installation.
“IQ does not replace partner expertise,” Bouchard added. “It amplifies this expertise to accelerate and scale projects. Manufacturers need local partners who understand their production reality and can provide the installation capacity and support needed to keep lines running. IQ gives those partners better information, better coordination, and a clearer path from opportunity to running system.”
by Gary Mintchell | May 14, 2026 | Generative AI, Manufacturing IT, Operations Management, Software
I devoted three days in April to attend the Aras Community Event (ACE 2026) in Miami, FL. Even though I am not a specialized market analyst in that market, I’ve been involved with the application of product lifecycle management ever since I was “The Kid in Engineering” at a manufacturing company back when, well, I was just a bit older than a “kid.”
Our company (another company that designed and built automated assembly equipment) transitioned to computer-aided design (CAD) while I was in management. Later, I became involved with AutoCAD.
So, there are memories of the great advances in the technology and capabilities.
My first summary of my three days with the Aras community in Miami was recorded on my podcast and YouTube channels. As I wrote at the time, “These PLM events always return me to the time when I did this sort of work–manually. Then my first taste of computers digitizing the bill of materials as a first step in our data management journey.”
Aras product managers showed how LLMs trained on the data within the app along with proper governance worked with agents to perform a number of tasks. Tasks in many cases that would require days of pain-staking work from a human.
While I heard from an analyst in the market that they thought this was all painfully slow, I’d offer the thought that a company does not want to outpace its customers. Most will not want to jump into the deep end immediately.
Chatting with CTO Rob McAveney, I heard how the company is taking a balanced approach to introducing these new technologies assuring that they are bringing their customer base along laying out the progression of “agentification of PLM.” The vision includes turning Aras Innovator into an “enterprise nervous system.”
The pressure of digitalization and the so-called digital transformation of companies drives these developers and suppliers into trying to find solutions to the immense data problems they face. Aras’ core technology lies in the digital thread, a topic often referred to.
Ironically, my discussions with Aras and some customers and prospects during the conference revealed an unhealthy fact that I’ve often heard in another software application market—MES. It seems that few users use the full complement of solutions offered by the vendors. This means that what could be a mature market is actually open for new solutions—meaning an innovative upstart like Aras has opportunity for market growth.
I researched the market using my favorite search engine—Claude.ai. The global PLM & Engineering software market reached $31.1 billion in 2024, growing 9.7% year-over-year, and is projected to hit $41.6 billion by 2029 at a ~6% CAGR. The top 10 vendors account for roughly 85% of the total market.
The leading suppliers include Siemens Digital Industries, Dassault Systèms, PTC, and Autodesk. Analysts report Aras Innovator is built for adaptability, offering a platform designed to evolve quickly with a low-code development environment and strong Digital Thread capabilities.
The four key development points for Aras agentic AI and LLMs, which were repeated often are:
- Trust
- Governance
- Observability
- Explainability
Shortly following the Aras event, I attended virtually the Siemens press conference from Hannover Fair.
Further research between the two revealed these thoughts from a variety of analysts.
Siemens Teamcenter Copilot is powerful but bounded. Siemens’ approach includes Teamcenter Copilot and AI Chat for natural language queries, RapidMiner for spotting quality issues, and AI extraction of procedures from static PDFs. Siemens describes it as “training AI in the language of engineering and manufacturing” — embedding domain-specific intelligence aligned with physics, lifecycle context, and operational constraints.
However, what Siemens is doing is focused, practical, and grounded in helping users navigate data Siemens already manages well. The copilots do not attempt to extend beyond Teamcenter — they do not ingest data from other PLM tools or external systems that influence product decisions, and the improvements remain confined to the boundaries of one platform.
Aras’s approach is architecturally more open. InnovatorEdge is designed so that product data, processes, and digital thread remain governed inside the core platform, while Edge services make them consumable everywhere else — enabling agents to link data across PLM, ERP, IoT, and documents.
One independent analyst commentary summarized the broader landscape bluntly: all four major PLM vendors — Siemens, Dassault, PTC, and Aras — are adding AI inside their products, but none of them are rethinking PLM architecture for an agent-native future. They are embedding assistants inside old systems rather than redesigning systems around the needs of agents. That said, Aras’s open, low-code, API-first architecture puts it structurally closer to an agent-ready foundation than Siemens’s more monolithic platform.
ACE attendees noted that while AI’s transformative potential was clear, discussions also centered on the need for human oversight, data governance, and addressing concerns about traceability and the dynamic nature of LLMs — suggesting customers are excited but appropriately cautious about full autonomy.
by Gary Mintchell | May 4, 2026 | Operations Management, Software, Workforce
Digital transformation initiatives are all the rage—at least in the marketing release system. I remain amazed that after all the released products and articles I’ve written the software layer of PLM and MES remain under utilized. One recent concern discussed in two recent interviews focuses on frontline workers and their supervisors.
The same situation exists that I confronted 50 years ago in an early role as data manager for a manufacturing company—no appropriate work is accomplished without reliable, easily assimilate-able, and clear instructions make it to the people doing the work.
In the standard words of reporting, I caught up with Garth Coleman, CEO of Canvas Envision, at the recent Aras Community Event in Miami, FL. The was the first of my two conversations on the topic.
He told me that while over the last few years, many industries have become dynamic, data-rich, and modernized, factory floor instructions are still largely outdated with PDFs, screenshots, and text-heavy documents that are now increasingly stale.
Just as part of my job years ago, manufacturers are still struggling to align as-built with as-designed.
He argues the shift here demands interactive, model-based instructions where teams adopt systems in real-time, creating a continuous loop for operations, rather than the other way around.
Canvas Envision features these cutting-edge technologies:
- No-Code Workflows: Allowing users to build and modify instructions without the need for IT involvement.
- CAD Fidelity: Ensuring that instructions are always up-to-date with design changes through native CAD visualization.
- AI Assistance: Automating the generation of complex views and lists with Evie, the integrated AI assistant.
- Gadgets: Providing ready-to-use components like checklists and data capture.
- Integration and Flexibility: Seamlessly connecting with enterprise systems (PLM, MES) and offering flexible deployment options (SaaS or self-hosted).
Until you close that final loop aligning as-built with as-designed in a 360-degree loop, everything is only data.
