Your SolidWorks CAD data describes the product accurately and in detail, and it still isn't a work instruction. The model shows the product finished, with every part in place. On its own it doesn't explain how to build that product or how to service it. Nothing in the model tells an operator which part goes in first, which way round that part faces, how tight to make it or what to check before moving on.
Creating work instructions from SolidWorks CAD data means using the design to build those instructions, then delivering them in a form and format the operator can work from. The design already carries the geometry, the part names and the assembly structure, so the author's job is determining the sequence and building the instructions around it. Both halves of that job, the authoring and the delivery, often run on a patchwork of manual tools and processes, and Canvas Envision is built to radically improve both.
You can't hand an operator the CAD files. They expose engineering data the floor has no reason to see, and even opened in a viewer the model still shows only the finished product. So someone opens the assembly and takes out what they can, saving a partial view, then an exploded state, then a screenshot of a subassembly with the housing hidden, each one as a separate file.
The sequence isn't in the model either, so the same person works it out and writes it down by hand, stepping through the assembly to decide what goes in first, why it goes in that order and which check catches the mistake people actually make. Every step then needs a view to match, and every view has to be set up, captured, saved and placed beside the right text. That capture work alone consumes a large amount of time and effort.
The pictures and the words then get compiled into a PDF or a slide deck, because the instruction has to reach the station as something an operator can use, and until recently a document was the only thing that qualified. None of the files that went into it has any connection to the others or to the model they came from. When the design changes, the author has to work out which pictures are now wrong, open the assembly again, retake the views and rebuild the document.
Engineering moved to 3D decades ago, and the bench still gets a PDF. A static document can carry only flat pictures and text. It has nowhere to put an interactive model, the part structure or the connection back to engineering, so none of that travels past the screenshot.
So a better instruction, one with the views, detail and checks an operator actually needs, means more of the same manual work. Every extra angle is another capture, every callout is drawn by hand, and every page has to be kept in step with the design. That time and cost fall on the same engineers the release schedule is waiting on, so speed and quality become a trade-off, and the release date decides which one gives.
Thin steps, skipped views and screenshots reused from the last variant are the result, and each one is a reasonable call made under a deadline with the tools that existed.
Improving only one half of the job doesn't change what the operator gets. A faster way to capture screenshots and assemble pages still ends in a static document, so the operator receives the same flat instruction sooner, with the same missing views and the same gap between the picture and the part in their hands. Moving the instruction into a 3D environment doesn't solve it either. Most of what an instruction carries isn't 3D at all. It's the step text, the torque value, the warning, the photograph of the correct orientation and the short video of a technique that is easier to watch than to describe. A 3D environment is built to show geometry, so all of that has to be pushed into the scene as annotations and pop-ups, and the operator has to navigate a model to find out what to do next.
The better approach runs the other way. Put the 3D content into the instruction, alongside the text, images and video the step needs, and let the operator decide on each step whether to turn the model and look closer or simply use it as a visual reference.
What has changed is that authoring and delivery can now be handled together, in one environment that works from the model and publishes the page the operator opens. A team no longer has to choose between a fast instruction and a good one. The change starts with how the SolidWorks model gets into the instruction.
Envision Creator, the authoring module in Canvas Envision, imports the SolidWorks assembly whole. It reads more than thirty CAD formats and brings the assembly structure and part names through with the geometry, so the instruction stays connected to the engineering data it came from and the digital thread from design to the floor stays unbroken. What the author works on is a secure, lightweight representation of the model rather than a copy of your CAD files, so your engineering data stays where it belongs and the instruction is light enough to open in a browser.
The author builds the steps around that model, and labels come straight from the part names and metadata, so the name on a step is the name engineering gave the part and nobody retypes it. Because the work happens in Envision Creator rather than in SolidWorks, the author can change a step, reorder the sequence or add a warning without anyone touching engineering data. Evie, the AI built into Canvas Envision, takes on the mechanical share of that work. It reads the assembly structure and part metadata, drafts and restructures content, and runs batch changes across a whole instruction, and the author checks what it produces in place.
When the design changes, the author points the instruction at the revised geometry, and Envision Creator maps the new model onto the work already authored, so the views, labels and notes carry across. Changes that need judgment still get it, so a connector moved to a new position gets a new angle of entry described and a part split in two gets its step split with it. The author reviews what actually moved, puts their name to those calls and republishes, which is what it takes to keep work instructions current as engineering changes land.
Zipline, which builds autonomous delivery drones, authors its instructions this way. Its first document in Envision Creator took 1 hour, against about 15 hours on the old route. A later configuration of the same product took 30 minutes against that same 15-hour baseline, and that is the saving that repeats every time the product changes.
Envision Operator, the Canvas Envision application for the shop floor, delivers the instruction to the station as an interactive page. The author lays out that page in Envision Creator with the 3D model as one element among several, alongside text, vector illustrations, photographs, QR codes and embedded video. The page the author publishes is the page the operator opens, so what the author decided the operator should see is exactly what arrives at the bench.
The 3D earns its place where a part has to be seen in space, such as a joint buried under a housing or an angle of entry that a still image turns into a guess. The author isolates the parts a step is about, ghosts the ones in the way, cuts a section where the joint is buried and saves the view. The operator arrives at the angle the author chose and decides whether to turn the model and look closer or keep working from that view as a reference. Elsewhere on the same page, a photograph and a sentence do the job.
Each instruction opens in Envision Operator through a persistent link. When the author publishes a new version, that same link shows it, so a link pasted into a travel sheet last quarter opens the current instruction and every operator is on the latest version the next time they open it. Where a team needs more control, each published version can also keep a link of its own. That preserves a history of what was released, and it lets the team decide when a new version reaches the bench instead of having it appear the moment it is published. For a team working to ISO 9001, this is the ground the standard's clause on control of documented information covers, from making the current version available where it is used to retaining a record of what was released before. Either way, the instruction stays digital from the author's screen to the station.
A SolidWorks team has already done the most expensive work behind a work instruction. The product is modeled, the parts are named, and the person who knows the build order is on the payroll. Until now, turning that work into instructions has meant choosing between speed and quality. Choose speed, and the instruction ships on time with thin steps and missing views, leaving the floor to deal with the quality problems that follow. Choose quality, and the release waits while an engineer captures, saves and assembles every rich detail by hand.
Building the instruction from the model in Envision Creator and delivering it through Envision Operator ends that trade-off. For once, you can have your cake and eat it too, with a faster instruction and a better one coming out of the same piece of work.
Garth Coleman is CEO of Canvas Envision, where he is changing how manufacturers capture, share, and execute the knowledge their best people carry. He brings nearly three decades in enterprise software spanning CAD, Product Lifecycle Management (PLM), and 3D communication, including senior leadership at Dassault Systèmes.
Book a demo with your own CAD data. Most teams are authoring in minutes.
