Patrick
Hebron
AI Toolmaker + Designer
Software developer, designer, teacher, and author working at the intersection of AI and creative toolmaking. Building tools that deepen thought and extend human capability.
AI + Industrial Reinvention
01 / 07AI + Industrial
Reinvention
An All-Around Better Horse
2025 · illustrated essayThis illustrated essay is the capstone of my work so far on AI-native design and engineering tools. It imagines design software that goes beyond drafting to help frame problems, learn new domains, test alternatives, and connect design decisions to manufacturing and business realities. It argues for purpose-built interfaces with just-in-time functionality.
Foil
2010–ongoing · thesis, software, and research programFoil began as my NYU master’s thesis, where I set out to make CAD a place for developing ideas rather than just executing them. It later grew into a broader hybrid machine-learning IDE and design tool, bringing creative coding, interface and scenegraph tooling, data work, and machine-learning development into one environment. That vision has carried forward through my later work on shared learning environments and AI development studios.
View Project →
A Unified Tool for the Education of Humans and Machines
2018 · essayThis essay brings one of the central ideas in Foil into clearer focus, arguing that AI-native environments should help people and models learn together when the domain, constraints, and even the problem are still unclear. Shared workspaces, interactive demonstrations, and experimental feedback become the basis for collaboration.
Read the Essay →
Intelligence Design Studio
2018 · presentationThis presentation distills aspects of Foil into a broader infrastructure proposal: an extensible AI IDE for coding, simulation, dataset curation, experiment provisioning, and generated interfaces. It argues that AI systems need to be built in whole environments designed around their workflows, not through isolated features bolted onto conventional software.
View Presentation →
Artificial Intelligence and the Decoupling of Creation and Comprehension
2022 · talkThroughout this body of work, I explore new modes of collaboration with AI in design, engineering, and science. This talk focuses on what happens when the technical theories and operating principles behind those systems move beyond the user’s understanding. It argues that tools must keep people informed, able to inspect what is happening, and able to trust what they have made.
Read the Talk →
Against Special Casing
2024 · blog postThis post widens the argument from interface design to software architecture. Industrial work changes from part to part and material to material, so software cannot remain a rigid body of fixed menus, panels, and workflows. Drawing inspiration from organisms like fungi rather than vertebrates, it imagines software that adapts to the problem rather than forcing the problem into a fixed container.
Read the Post →
Morphogenic Game Theory
2026 · forthcoming paperA technical follow-up to An All-Around Better Horse, this paper attempts to implement some of that essay’s key ideas in more concrete form. Drawing on biological morphogenesis and game theory, it explores how parts and assemblies can emerge from local agents, incentives, and negotiation rules. Design becomes less like drawing a finished object and more like specifying the developmental logic by which form differentiates and assembles itself.
Coming SoonAI + Aesthetic Computing
01 / 07AI + Aesthetic
Computing
Rethinking Design Tools in the Age of Machine Learning
2017 · essayThis essay is a foundational introduction to my work on AI and aesthetic computing. It argues that machine learning should not be used to flatten creative work into automation, but to help design tools adapt to the user’s way of thinking. I lay out several directions for that shift, including emergent feature sets, design through exploration, design by description, process organization, and conversational interfaces.
Machine Learning for Designers
2016 · bookThis book distills my learnings from building AI tools into a practical framework for designers. It explains why machine learning changes the design of user-facing software, especially once software begins dealing in fuzzy concepts, probabilistic behavior, and learned representations rather than fixed symbolic logic. It was written to help designers understand not just the technology itself, but the kinds of interfaces and interactions it makes possible.
Creative Partnerships
2016 · webcastThis webcast is an early public statement of an idea that runs through much of my later work on AI and creative software: machine learning should not be treated only as automation, but as the basis for a new kind of partnership between people and tools. It starts moving the discussion away from rigid interfaces and fixed workflows, and toward systems that can respond more fluidly to experimentation, discovery, and changing intent. In that sense, it helps prepare the ground for Rethinking Design Tools.
Neural Filters for Adobe Photoshop
2020 · softwareNeural Filters brought a number of these ideas into a major creative tool used by millions of people. Rather than treating AI as a black-box replacement for creative judgment, this work introduced a new Photoshop workspace for nondestructive, exploratory image editing. Filters like Style Transfer, Colorize, Smart Portraits, and Landscape Mixer point toward a vision of creative software that helps people move through aesthetic possibility spaces more quickly without closing those spaces down.
Against Prediction + Big Data, Big Design
2020–2021 · interviewsThese interviews sharpen the humanistic side of my work on AI design tools. They argue against systems that try to predict the user’s final intention too literally, and instead push toward tools that adapt themselves to the direction the user is trying to go. They also make a broader case that machine intelligence proves most useful not when it imitates human thought, but when it offers a genuinely different kind of intelligence that can expand creative work in partnership with human judgment.
Adobe’s AI-First Approach to Product Vision
2021 · talkThis talk brings the concept of AI as a creative partner into the context of product vision at Adobe. It focuses on what it means to think about AI from the outset, rather than as a feature bolted onto conventional software after the fact. Set alongside Neural Filters and the interviews from this period, it marks a point where my work on creative tools becomes inseparable from questions of product strategy, interface direction, and how new machine-learning capabilities should reshape the software itself.
Authoring Creativity With AI
2024 · podcastThis conversation brings the earlier work forward into the generative AI moment. I talk about the interface problems that come with AI-native software, including imprecision, fallback behavior, discoverability, and the difficulty of helping people understand what a system has understood. The conversation maintains the same concern with creative agency, but now in a world where the tools are far more generative, fluid, and unpredictable.
The Archives
Work
Technical essays, experimental tools, and showcases of industry projects bridging AI and design.
Blog
Short-form articles, project write-ups, and unfinished essays on AI, design, and creative computing.
Talks
Keynotes, guest lectures, and workshops exploring human-AI collaboration and creative systems.
Teaching
Syllabi, reading lists, and pedagogical materials from courses and workshops taught at NYU and beyond.
About
Patrick Hebron is a software developer, designer, teacher, and author working at the intersection of AI and creative toolmaking.
Raised by progressive educators, Patrick spent his childhood prototyping VR headsets with aluminum foil. After undergraduate studies in Philosophy and Art, he built custom filmmaking tools before formalizing CS studies at NYU’s ITP. There, he developed Foil, a hybrid AI IDE and design environment. He later served as a researcher and adjunct professor teaching machine learning and software architecture.
His leadership spans Design, Engineering, Research, and Product, building teams that innovate across architecture, models, and UX. At Adobe, he founded the AI design group and co-initiated Neural Filters and Firefly. As VP of R&D at Stability AI, he focused on creative control in image generation. He currently leads agentic engineering initiatives at NVIDIA.
Patrick wrote Machine Learning for Designers (O’Reilly), and speaks frequently on artificial intelligence and creativity.
His mission is to build tools that deepen thought and extend human capability.
Contact
For collaborations, speaking, or research inquiries.