Hestus

Hestus is an AI copilot inside Autodesk Fusion that reads your sketch as you draw and suggests constraints, dimensions, symmetry and cleanup, accepted with one keypress. It is available for Mac and Windows, with SolidWorks in alpha.

Hestus is an AI copilot that runs inside Autodesk Fusion and acts like autocomplete for CAD sketching. Instead of asking users to describe a design in a chat box, it watches the sketch as it is drawn, infers the intent behind it, and offers the next constraints, dimensions, symmetry operations and cleanup steps. The site pitches it with a claim of 2.5x faster work and 4x fewer clicks. In practice, the interaction is built around suggestions. Draw a rectangle and Hestus proposes both centerlines as construction geometry, which one key places. When several ideas are plausible, the user taps through the alternatives and presses Enter on the right one, such as mirroring a circle into a set of four with symmetry constraints included. Continuing to draw simply dismisses the suggestion. The homepage contrasts this with text-prompt tools, arguing that typing a long instruction for a pattern that takes two clicks is a step backwards, so Hestus chooses a keypress, a click or no input at all depending on the job. The target audience is people who already spend their day in parametric sketching: mechanical and product design engineers, manufacturing staff, founders building hardware, hobbyists, educators and students. Testimonials on the page mention repetitive constraints, symmetry, mirror and dangling point handling, and getting designs ready for manufacturing. The company says about 5,000 engineers use it and lists Y Combinator among its backers. Hestus is available for Autodesk Fusion on Mac and Windows, with a free download. A SolidWorks version is offered only through an alpha sign-up. Because it lives inside the host CAD, there are no files to export or transfer and no new application to learn. Among alternatives, it sits between general-purpose AI chat assistants and native CAD features such as built-in constraint inference, with a narrower focus on sketch completion rather than generating whole models from text.