Interview, Fireside Chat
Text-to-CAD: AI Revolutionizing Hardware Design with Jordan Noone of Zoo
Sourcery with Molly O'SheaJordan Noone, Molly O'Shea, Jessie Frazelle, Ben Horowitz, Vinod Khosla, Alfred Lin, Mike Maples, Roger Ehrenberg
- The company expects the merged modeling app and Text-to-CAD tools to be fully integrated within the coming weeks to enable chat-prompt editing of AI-generated parts.
- Headcount is projected to increase from approximately 20 to about 50 employees within the next 12 months.
- A first local data center to host GPUs for geometric rendering and machine learning training is scheduled for construction over the next year.
- Usage and feedback are anticipated to reach a huge inflection point following the release of the editing feature in the modeling app.
- A public-facing manufacturing API and feature implementation are not expected until the end of the year, with full maturity achieved throughout the following year.
- Capabilities for sending text prompts directly to CNC machines for toolpath generation and verification are planned for deployment within the next 18 months.
- Machine learning efforts will expand to cover geometric designs at a scale comparable to large language models, necessitating significant internal infrastructure development.
- Text-to-CAD functionality will be extended to include design for manufacturability checks, tool pathing, and engineering analysis for stress and pressure loads.
- The addition of manufacturing software capabilities, specifically CNC tool pathing, is positioned as the transition from a design-only entity to a full hardware development lifecycle company.
- An internal team will be bootstrapped over the next year to handle CNC machine operations, tool path automation, and end-to-end verification.
- Primary early adoption is expected from the most valuable companies at the intersection of hardware and software, though universal customer adoption is not guaranteed.
- The manufacturing kickoff aims to consolidate automation across multiple sectors and verticals simultaneously rather than addressing bottlenecks in isolated areas.
- Virtual datasets generated internally via the geometry engine will be utilized to supplement external data gaps during Text-to-CAD model training.
- Embedding design for manufacturability into ML models is intended to eliminate the manual, labor-driven feedback loop between engineers and machinists.
- Production versions of manufacturing features are expected to be shipped six to seven months after initial beta launches occurring at the end of the year.
- Integrated design and manufacturing automation is expected to allow a single individual to consolidate workflows that previously required interaction with multiple engineering specialists.
- Significant labor savings and a reduction in workflow time are projected, enabling users to reach 99% of a design via text and finalize the remaining 1% manually.
- APIs will be opened to third-party developers to enable the creation of bespoke tools featuring manufacturing awareness or auto-generated instructions for CNC machines.
- Computational geometry and machine learning teams will continue to be expanded as the core differentiation backbone for the company.
- The company plans to continue dogfooding its software by internally building tools to ensure API functionality before external customer release.
- Adoption is expected to be driven by the asymmetry between software and hardware development at modern tech companies, favoring customer cohorts with existing modern software talent bases.