The AI-powered platform converts 3D models into digital work instructions, reducing creation time from hours to minutes.
At a Glance
- The platform captures retiring workers' knowledge before it leaves the building.
- AI creates work instructions from videos and PDFs in minutes.
- Digital delivery meets the expectations of millennial and Gen Z workers.
Worker shortages are plaguing the manufacturing industry. Boomers are retiring, and millennials and Gen Z are not quick to step in. An additional problem accompanies the dearth of workers: The shortage of manufacturing knowledge. The loss of institutional knowledge that comes when experienced employees retire is exacerbated by the difficulty of teaching young workers who are expecting digital tools.
Canvas Envision is tackling this with an AI-powered platform that changes how manufacturers create and deliver work instructions. "Manufacturers are having all kinds of problems with workforce retention, workforce onboarding, and knowledge leaking out of the building," Garth Coleman, the company’s CEO, told Design News. "We can now solve many of these problems."
From 3D models to digital delivery
Canvas Envision uses a nontraditional AI-based approach to training, one that meets the expectations of younger workers. The company’s platform takes 3D CAD data and packages it into accessible digital containers that can be delivered to workers on the shop floor or saved as traditional documents when needed.
This approach lets manufacturers combine multiple content types within a single delivery package. "I can combine 3D with video, with text, with images, with whatever you need to deliver in whatever language you want, whatever format you want," said Coleman, "And it all stays digitally connected through the platform."
AI eliminates the authoring burden
Canvas Envision integrated artificial intelligence into its platform. AI significantly reduces the time required to create work instructions. "What used to take hours or days is now done in minutes," said Coleman. "AI will propose a sequence. When you agree that's the right sequence, it will create the visual. It will hide or show things at your discretion, move them, explode them, label them, change colors, add text—all that stuff."
An example of Rockwell instructions:
The AI capabilities extend beyond 3D models to video and PDF content. "If you have a training video, throw it in," said Coleman. "How long would it take you to edit a training video? Chop it up into step sequences, put those into some kind of web page, and put some text on it? It would take you several hours. With the AI-based platform, it takes less than 10 minutes."
Coleman noted that with lengthy technical documents, the AI can extract and reorganize relevant information on demand. "If you have a 600-page PDF that explains how to install a pump and maintain it on page 75, but the safety procedure's on page 10, and the necessary drawings are on page 300, you can ask AI to extract the pages you need and rewrite them.”
Solving a decades-old problem
Coleman was surprised by how many manufacturers still rely on static PDFs and screenshots for work instructions, despite massive investments in smart factory technology. 'Why are they still using PDFs? Why are they still using a patchwork of screenshots? It's because authoring is a massive burden," said Coleman. "It takes time to make it easier to understand, and the manufacturing engineer has other work to do.”
This authoring burden has created a gap that was traditionally filled by experienced workers' tribal knowledge. "Those guys are all leaving, and their knowledge hasn’t been captured," said Coleman. "New workers come in, and they have this inadequate document. We solve that problem with AI and reduce the authoring burden."
The solution also addresses generational expectations. "Young digital natives are in college right now, but when they go into the workforce, they're not going to want to do this old-school stuff, reading PDFs and going to page 65. They just want to click and do.”
Multi-agent AI architecture designed for manufacturers
Canvas Envision's AI implementation uses a multi-agent system rather than a simple chatbot interface. The platform operates in three modes: an authoring assistant for manual tasks, an agentic system that autonomously creates work instructions from source materials, and an orchestration layer that coordinates different specialized agents.
The platform focuses on manufacturers with low-volume, complex, discrete manufacturing operations—industries where configurations change frequently, and traditional work instruction approaches struggle to keep pace. "Industrial equipment, med devices, maybe some sub-tiers in automotive and aerospace, drones—that’s where you have lots of configurations," said Coleman. "Those are the ones that will best benefit from the platform, since it's hard for them to constantly create and maintain the work instructions because they're changing so much.”
The platform serves dual purposes: capturing knowledge from departing experienced workers and accelerating onboarding for new employees. "You can record the knowledgeable worker doing something and make a work instruction out of that," said Coleman. "With that done, you can put the new worker to work quickly."
Real-world implementation
One customer, Otis, demonstrated how to build a work instruction application by creating a video of a seven-minute procedure. "They used to train people in a classroom and then go to the floor to do it. On the floor, they would mess up until they figured it out," said Coleman. "Now, the video is timed to do the procedure in the same time it takes on the floor. So it's not watch-then-do; it's watch-and-do. They can take someone who's never been trained, put them in front of the video, and they can do the procedure at the same time.”
Coleman envisions a future where work instructions serve both human workers and robots. "When the robots are doing the work, humans aren't trained on it,” said Coleman. “Yet when the robot goes down, the human has to step in. If you have instructions for a human and you also have the instructions for the robot, you have redundant capability."
Rather than viewing AI as a threat to manufacturing jobs, Coleman sees it as a solution to the industry's persistent labor challenges. "A lot of people are talking about AI putting people out of work," said Coleman. "We're going to use AI to put people to work and solve this manufacturing worker shortage."
Facts Only
* The platform converts 3D models into digital work instructions, reducing creation time from hours to minutes.
* The platform captures retiring workers' knowledge before they leave the building.
* AI creates work instructions from videos and PDFs in minutes.
* The system packages 3D CAD data into accessible digital containers deliverable on the shop floor or as traditional documents.
* The AI proposes sequences, which are then used to create visuals, hide/show elements, move, label, and change colors.
* AI can extract and reorganize relevant information from lengthy technical documents on demand.
* One example involved editing a training video: chopping it into step sequences, putting them into a webpage, and adding text takes less than 10 minutes.
* A customer demonstration showed transforming a seven-minute procedure video into a "watch-and-do" application.
* The platform utilizes a multi-agent system with an authoring assistant, an agentic system for autonomous creation, and an orchestration layer.
Executive Summary
An AI-powered platform named Canvas Envision transforms 3D models into digital work instructions, aiming to solve manufacturing knowledge crises stemming from worker shortages and the loss of institutional knowledge due to retirements. The platform uses AI to create work instructions from various sources like 3D CAD data, videos, and PDFs, drastically reducing creation time from hours or days to minutes. This approach allows for flexible content delivery, enabling manufacturers to combine multiple media types into a single package accessible to younger workers expecting digital interaction.
The company addresses the "authoring burden" felt by engineers, which is often deferred by relying on static documents like PDFs and screenshots despite advanced factory technology. The AI capability automates the creation process by proposing sequences, generating visuals, and extracting relevant information from lengthy documents instantly. The system uses a multi-agent architecture to coordinate tasks, focusing on low-volume, complex discrete manufacturing where configurations change frequently. A real-world example showed that training videos can be transformed into "watch-and-do" procedures, improving the efficiency of onboarding new staff and bridging knowledge gaps between experienced workers and new hires.
Full Take
The narrative presents a shift from knowledge hoarding by experienced employees to systematic, AI-mediated knowledge transfer, positioning the technology as a necessary response to labor market dynamics rather than just a productivity tool. The core tension lies in how this technological solution aligns with the human element of expertise and generational expectations. The system's focus on capturing tacit knowledge via video and 3D models addresses the problem that experienced workers are leaving behind, suggesting a potential mitigation for institutional knowledge leakage.
The reliance on an autonomous, multi-agent architecture implies a critique of traditional, linear instruction methods (like static PDFs) which impose an excessive authoring burden. This moves the discussion beyond mere efficiency gains toward restructuring how expertise is codified and disseminated. The implication is that the solution addresses a systemic failure where documentation lags behind technological advancement, creating a dependency on aging human knowledge for complex, changing systems.
The challenge for future implementation lies in ensuring that the process of AI-driven instruction creation preserves contextual nuance rather than just surface-level procedural accuracy. If knowledge capture prioritizes speed over deep contextual embedding, there is a risk of replacing tribal knowledge with brittle, albeit fast, digital artifacts. Furthermore, framing AI as purely solving workforce shortages requires careful navigation; if the system is adopted primarily to accelerate onboarding and displace manual authoring, it must be scrutinized to ensure it augments human judgment rather than merely streamlining obsolescence.
Bridge Questions:
What frameworks can be developed to audit the contextual fidelity of knowledge extracted from non-textual sources like video and 3D models? How should organizations balance the efficiency gain from autonomous creation against the risk associated with delegating complex, high-stakes procedural instruction generation to agentic systems? What structural changes are necessary for manufacturing environments to integrate dynamic knowledge systems that satisfy both legacy expertise and digital nativism simultaneously?
Sentinel — Human
The text appears to be a human-authored business feature that successfully synthesizes technical details with broader socio-economic commentary on the manufacturing workforce.