CoLab has set a launch window of H1 2027 for a generative CAD product, and it is shipping the first piece of that roadmap now: a capability called generative artifacts that can produce and refine engineering documents such as Design Failure Mode and Effects Analyses (DFMEAs) and root cause analyses while preserving program context and design history. The company announced the plan on Sept. 30, 2026, from St. John's, Newfoundland and Labrador.

Co-founder and CEO Adam Keating framed the announcement as a correction to what he sees as an industry fixation on CAD speed. "There is a lot of excitement about generative CAD accelerating design iterations and allowing engineering teams to explore more concepts," Keating said. "It's an important piece of the puzzle, but if all you do is speed up CAD operations, you're not really solving for program speed. The question engineering leaders should be asking is not 'how can I adopt generative CAD?' but 'how can our organization shift toward generative engineering?'"

The distinction matters commercially. Generative artifacts targets the dozens of non-CAD deliverables that engineering teams must mature and approve before release — documentation that is often the bottleneck in program timelines rather than geometry creation itself. CoLab's platform generates these artifacts within the context of the design history and expert feedback already captured in its system.

For the 2027 generative CAD product, CoLab is betting that trust, not raw generation capability, will be the differentiator. With model capability advancing rapidly, the company says it is focused on guardrails that help AI produce outputs engineers can actually use.

"Our customers are designing complex products that must be manufactured at scale," Keating said. "AI is getting better at generating geometry, so we're focused on the last mile: fine-tuning the outputs using an organization's historical designs, standards and guidelines, and expert knowhow."

The generative CAD offering will address two needs. First, concept generation: CoLab will leverage historical program data and integrations with leading CAD and topology optimization tools, including nTop, to generate new design concepts. Second, an "AutoFix" capability that modifies a design based on a specific annotation or requirement. CoLab says its differentiator is using engineering knowledge and documentation to justify each design decision.

Co-founder and CTO Jeremy Andrews argued that this rationale-based approach is what protects a company's competitive position as AI gains access to CAD. "Your competitive advantage is based on the knowledge only you have," Andrews said. "So AI should leverage internal data and standards, there should be a human in the loop, and design changes should be paired with rationale as part of an auditable design history."

CoLab's data position underpins that claim. Hundreds of engineering organizations already use its Design Engagement System to run virtual design reviews, capturing expert feedback that the company says is rarely documented elsewhere. That captured design rationale, according to CoLab, forms the backbone for reliable generative CAD because it gives models context beyond geometry.

The company is also investing in explicit guidelines for AI. Customers can upload internal standards and guidelines to the platform, and CoLab will license trusted third-party standards. Earlier this year, CoLab announced a licensing agreement for ISO standards, enabling AutoReview — CoLab's AI peer checker — to review designs for compliance and provide citation-backed annotations.

The announcement lands in a market that is moving quickly toward AI-accessible CAD. PTC and Autodesk have both recently announced Model Context Protocol (MCP) offerings, giving AI agents and large language models a standard way to access functionality within Onshape and Fusion. CoLab reads this as confirmation that AI for CAD is accelerating — and as a building block for broader AI for engineering rather than an endpoint.

"Soon, AI agents will be able to use any CAD tool," Andrews said. "But what will ultimately accelerate program speed is when you can combine agentic CAD with the engineering knowledge, reviews, and validation data to mature designs. That's where CoLab is going long term."

CoLab, founded in 2017, sells its EngineeringOS platform to mechanical engineering and hardware development teams at leading global manufacturers. The company positions the platform as a collaborative workspace that connects people, data, and AI, capturing expert knowledge as a byproduct of day-to-day work and applying it automatically through built-in AI agents.

Between now and the H1 2027 generative CAD launch, the competitive question will be whether CoLab's repository of captured design rationale and licensed standards can outperform the geometry-generation capabilities that PTC, Autodesk, and other incumbents are already exposing to AI agents through MCP.