AI is changing how people design and make. At Autodesk University 2026, we’re showing what comes next: AI that doesn’t just understand a prompt, but understands the data, relationships, decisions, and workflows behind real projects.
For developers, there’s an equally important part of this story.
We’re building this future to be open and extensible.
That means giving developers new ways to connect their own agents, tools, data, and domain expertise with Autodesk, and build AI-powered workflows that span applications, disciplines, and organizations.
Here’s what you should know.
AI gets more useful when it understands the work
General-purpose AI knows a lot about the world. But design and make workflows require something additional: an understanding of geometry, models, schedules, materials, project relationships, and the decisions people make along the way.
Autodesk Assistant can draw on context from Autodesk design and make data to better interpret those workflows and generate more relevant, explainable responses.
The context graph maps how information, people, decisions, and activities in a project relate to one another, including the why behind decisions and the state of the workflow when they were made.
Why does that matter to developers?
Because context is what turns Autodesk Assistant from a generic chatbot into something capable of interpreting a real workflow. By drawing on relevant Autodesk data, Assistant can better identify relationships and dependencies across a workflow and provide more relevant, explainable outputs.
That creates a much richer foundation for the experiences developers can build.
Your agents + Autodesk agents + partner agents
Real workflows don’t happen inside one application, or even one company.
That’s why we’re introducing Autodesk Assistant Builder, designed to let organizations connect their own agents and trusted partner agents with Autodesk Assistant. The vision is straightforward: your AI, your partners’ AI, and Autodesk AI working together in one experience.
For developers, this opens some particularly interesting possibilities.
Assistant Builder will connect company agents, MCP servers, tools, and data sources to Autodesk Assistant. That means specialized intelligence your team has already built can become part of Autodesk workflows rather than living in a separate AI experience.
That’s a big shift for developers: integrations can increasingly contribute not just data, but intelligence and action.
Agents are moving closer to the work
We’re also bringing agents into specific industry workflows.
In construction, new agents are designed to search project documents, turn proposal PDFs into structured bid data, and evaluate submittals against specifications, with people reviewing the results and making the final call.
In Fusion, an agentic PLM experience brings AI into product lifecycle workflows, using Autodesk Assistant to extract actionable information, identify product impacts, evaluate alternatives, and help coordinate changes.
These examples point toward a broader opportunity: agents grounded in domain data and embedded directly into the workflows where work happens.
And Autodesk Assistant itself is headed in that direction. At Autodesk University, we previewed a next-generation, standalone, agent-first experience designed to reach across products, projects, and workflows. Not as a replacement for existing product experiences, but as a new way of working across them.
The developer opportunity is bigger than a new set of APIs
For years, developers have extended Autodesk products by connecting data, automating tasks, and building specialized workflows.
Agentic AI expands that opportunity.
The emerging architecture connects three things:
- Context gives AI an understanding of the project and workflow.
- Capabilities let Autodesk, customer, and partner intelligence work together.
- Agents can use that intelligence to help people accomplish work across products and systems.
The strategy is deliberately open. As Autodesk CTO Raji Arasu put it in the AU keynote, “Autodesk has always believed in open solutions that help extend your unique workflows. Agentic AI is no different.”
For the APS developer community, that means the opportunity isn’t simply to add AI to an existing integration. It’s to start thinking about what unique context, expertise, tools, and actions your application can contribute to a larger intelligent workflow.
Build what comes next
There’s a consistent idea behind the announcements at AU this year: Autodesk AI is being built for the realities of design and make: project-aware, open and connected, with people retaining control over decisions.
That matters for developers because Autodesk can’t, and shouldn’t, build every specialized workflow our customers need.
You know your customers. You know their processes. And many of you have already built the applications, automations, integrations, and domain expertise they depend on every day.
Now there’s an opportunity to make that expertise part of an agentic ecosystem.
The next generation of Design and Make software won’t be built by Autodesk alone. We’re excited to build it with you.