From files to shapes: the real shift in CAD automation
CAD shape intelligence is the use of AI geometry search to recognise, compare and act on the actual 3D shape of parts inside engineering models so that agents can automate design reuse, simulation setup and manufacturing planning without relying only on filenames, tags or manual lookup. This is the quiet revolution in CAD: we are moving from document-centric to geometry-centric workflows, and teams that embrace it will turn their sprawling part libraries into searchable design memory instead of dead storage. Tech Soft 3D’s HOOPS AI 1.1 embodies this shift with machine learning tuned for part similarity, design reuse and CAD-aware AI workflows. The takeaway is straightforward: AI that understands shape is not a side feature, it is the new backbone for automated part reuse and agentic CAD automation in serious engineering environments.
Synera shows what CAD shape intelligence looks like in practice
Synera’s Advanced AI module, powered by HOOPS AI 1.1, puts CAD shape intelligence directly inside its agentic engineering workflows. Instead of hunting through folders by name or metadata, teams can bring CAD files into Synera, create searchable collections of parts and run similarity searches based on geometry itself. That matters more than it sounds. When agents can interpret geometry, identify similar parts and link them to costing, manufacturing processes and procurement, design reuse stops being a manual chore and becomes a default behaviour. According to Tech Soft 3D, HOOPS AI 1.1 expands its framework with new shape intelligence capabilities designed specifically for part similarity and CAD-aware AI workflows. In opinion, this is the missing piece that makes agentic CAD automation credible: workflows now act on the shape of the part, not just the text wrapped around it.
ODA’s MCP servers turn AI agents into CAD power users
While Synera makes geometry searchable inside its own environment, the Open Design Alliance (ODA) is doing something arguably more radical: exposing its CAD-format SDKs to AI agents through self-hosted Model Context Protocol (MCP) servers, due in Q3. For more than 25 years, ODA has been the invisible engine behind hundreds of applications that read and manipulate DWG, DGN, IFC, STEP and Revit files, governed by more than 1,200 members and structurally protected from acquisition. Now the same compiled C++ capabilities will sit behind an MCP server that agents query via natural language instead of C++ or .NET APIs. The server mediates access, returning geometry, properties and structure without handing raw binary files to an LLM. In my view, this turns AI from a mere app controller into a first-class CAD tool user: the agent interacts directly with the file-format toolset that developers rely on, but within the firm’s own infrastructure.
Why shape-aware agents matter for ordinary design teams
The practical impact is clear: shape intelligence reduces design duplication and accelerates reuse of existing design context in enterprise environments. With Synera’s AI geometry search, teams can identify potential duplicates, reuse previous designs, compare related components and carry over business data such as costing and manufacturing processes. On the ODA side, internal agents can interrogate BIM models, answer technical questions, generate schedules, identify missing information, compare design revisions, extract quantities and check standards compliance—while project data stays on the firm’s own network. A non-developer will be able to do sophisticated work with CAD and BIM data with the help of AI. The arrival of generative AI changes who benefits from these capabilities: they are no longer reserved for developers wiring up APIs, but for engineers, architects and contractors who can talk to models in plain language while the agent handles the geometry and metadata.
The new normal: AI-assisted CAD as standard workflow, not novelty
General-purpose AI tools have raced ahead on text and images, but engineering teams work with 3D models, simulation data, product data and manufacturing context that were largely out of reach. Built with the HOOPS AI toolkit, Synera’s Advanced AI squarely addresses that gap by giving workflows and agents a way to compare actual CAD geometry, not just names or descriptions. In parallel, ODA’s MCP servers are set to cover DWG, STEP and IFC at launch, with DGN, Revit and Navisworks to follow, and Revit write-back planned as a longer-term project. While none of this is shipping yet, the direction is clear—and, from my perspective, it is the right one. AI-assisted CAD workflows are becoming standard for teams managing complex part libraries and multi-project repositories. The conclusion is blunt: in the next wave of engineering software, if your tools cannot search by shape and automate reuse, they will be viewed as incomplete.
