Autodesk is expanding its AI strategy beyond individual software products with a next-generation Autodesk Assistant designed to connect project data, specialized AI and workflows across Forma, Fusion and Flow. Previewed at Autodesk University 2026, the agentic AI experience is intended to help architecture, engineering, construction, manufacturing and media teams understand project changes, identify downstream impacts and act across applications without having to determine which Autodesk tool to use first.
Autodesk is positioning its next generation of AI around a simple shift: from answering questions inside individual applications to understanding the project those questions belong to.
At Autodesk University 2026, the company previewed new agentic AI capabilities across its three industry clouds—Autodesk Forma for architecture, engineering and construction (AEC), Autodesk Fusion for design and manufacturing, and Autodesk Flow for media and entertainment. The capabilities are being incorporated into the next generation of Autodesk Assistant, which Autodesk plans to begin rolling out in 2027.
Autodesk Assistant is already available across a range of Autodesk products, with capabilities varying by product. Fusion, for example, now offers natural-language assistance for design, manufacturing guidance, project management and other workflows.
The new direction is broader. Instead of requiring professionals to identify the correct application, feature or workflow before asking for help, Autodesk wants Assistant to start with the intended outcome and use project context to determine what information and capabilities are relevant.
That distinction matters in industries where a seemingly small change can affect multiple stages of a project.
A modification to a building’s structure, for example, can have implications for engineering calculations, fabrication, schedules, costs, construction sequencing and ultimately operations. An AI system that only understands the immediate prompt can describe the change. An agentic system connected to project data can potentially identify related dependencies and bring the relevant engineering, design or production capabilities into the workflow.
Autodesk calls this project intelligence: actionable intelligence created from connected project data, industry context and AI.
The company’s three industry clouds provide the foundation. Forma connects data and workflows across AEC; Fusion links product design, engineering and manufacturing; and Flow connects content creation and production workflows in media and entertainment. Autodesk is attempting to use those connected environments as context for specialized AI rather than treating large language models as standalone interfaces.
The strategy reflects a broader trend in enterprise AI. Organizations are moving from general-purpose chatbots toward AI agents that can interact with business systems, retrieve context and execute multistep workflows. For design and manufacturing, however, the underlying data is considerably more specialized than the documents or business records typically handled by enterprise copilots.
Geometry, material properties, engineering constraints, physics, manufacturing processes, project schedules and production histories can all determine whether an AI-generated recommendation is usable.
Autodesk’s argument is therefore less about having a larger general-purpose model and more about grounding AI in the systems professionals already use.
Its next-generation Assistant is organized around three capabilities.
The first is identifying what requires attention. Autodesk’s Personal Pulse is designed to surface prioritized issues and insights across projects, while Assistant Spaces provides a focused environment where users and AI agents can investigate an issue, collaborate and examine downstream effects.
The second is acting across workflows. Autodesk says Assistant will combine its specialized AI with the design, engineering, simulation, manufacturing and creative capabilities embedded in its products. Natural-language and voice interactions are intended to connect user intent with actions that remain reviewable by professionals.
The third is extensibility. Assistant Builder will allow customers to connect their own AI agents, tools and trusted partner services to Autodesk project context. Autodesk also plans support for third-party agents through its Design & Make Marketplace.
That open approach could become important as companies build increasingly specialized AI agents around their own processes. Rather than forcing every workflow into Autodesk’s native AI capabilities, customers can potentially combine Autodesk project information with their own models, agents and enterprise tools.
Behind the experience is Autodesk’s AI Orchestrator, which the company says evaluates different models according to the requirements of individual tasks, including accuracy, speed, security and cost. That architecture moves Autodesk away from a one-model-fits-all approach and toward task-specific model selection.
The company is also emphasizing AI governance. Autodesk maintains AI Transparency Cards that explain AI feature functionality, data sources and privacy and security safeguards. At AU 2026, it expanded those cards and brought trust information directly into Autodesk Assistant, allowing customers to ask questions about an AI feature’s data handling, safeguards, limitations and underlying model.
The emphasis on context and governance comes as AI adoption accelerates across design and manufacturing industries. Autodesk’s 2026 AI Pulse research, based on responses from 2,500 leaders, found that 98% use at least one AI tool, 84% report increased productivity and 59% either already use or plan to use agentic AI within a year.
Autodesk’s 2027 State of Design & Make research also indicates that AI is moving from experimentation toward operational deployment. The company reports that 40% of organizations have integrated AI into design workflows and 34% are implementing AI-driven automation, while three-quarters plan to increase AI investment over the next three years.
The challenge now is whether agentic AI can operate reliably enough in environments where mistakes have physical and financial consequences.
Autodesk is keeping professionals in the loop rather than positioning Assistant as an autonomous replacement for designers, engineers, manufacturers or creative teams. That distinction is important for enterprise AI applications in which an incorrect recommendation can propagate into a physical product, construction project or production pipeline.
For Autodesk, the next phase is therefore less about adding another chatbot to its software portfolio and more about turning its connected industry clouds into an AI execution layer. The expanded Assistant is not yet generally available, with rollout details and subscription terms expected in 2027.
If Autodesk can connect project intelligence with specialized tools while preserving human review and enterprise controls, its strategy could make agentic AI more directly relevant to the real-world workflows that sit between digital design and physical production.
Market Landscape
Enterprise AI is shifting from conversational interfaces toward systems that can understand context, coordinate tools and execute multistep tasks. Autodesk is applying that model to industries where AI must work with specialized project data rather than generic business information.
Its strategy also reflects the increasing importance of connected data. Autodesk’s 2027 research says 91% of AI “Pacesetter” organizations have connected systems that integrate information across the organization, compared with 60% of mainstream organizations.
The competitive landscape includes general enterprise AI platforms from Microsoft, Google and Amazon, alongside specialized engineering, manufacturing, construction and media software providers. Autodesk’s differentiation is its direct access to industry-specific project data and the applications used to act on that data.
Top Insights
- Autodesk previewed agentic AI designed to connect project intelligence across Forma, Fusion and Flow rather than keeping assistance inside individual products.
- The next-generation Autodesk Assistant is planned to begin rolling out in 2027 and is not yet generally available.
- Personal Pulse, Assistant Spaces and Assistant Builder extend the platform from AI assistance toward project monitoring, investigation and workflow execution.
- Autodesk’s AI Orchestrator is designed to select models according to task-specific requirements including accuracy, speed, security and cost.
- Autodesk’s 2026 AI Pulse found that 59% of organizations already use or plan to use agentic AI within one year.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
