How did Flux get started?

The founder, Matias Wagner, realized that hardware design tooling hadn't improved like software tooling. After a summer focused on hardware projects, he was inspired to create a smarter, AI-driven design tool, leading to the founding of Flux in 2019.

What role does AI play in the Flux platform?

AI acts as an assistant within Flux, helping with tasks like information gathering, suggesting component replacements, optimizing costs, and even generating designs from a product brief. It aims to collaborate with the user, automating repetitive or time-consuming tasks.

How reliable is AI tool calling in Flux?

Tool calling for AI models has become significantly more reliable in the last year. Flux leverages this to enable its agents to perform actions like searching libraries, checking pricing and lead times, and adding/removing components.

Can Flux help me design a complete electronic product from scratch?

Yes, Flux's vision is to 'prompt hardware into existence.' While currently focused on PCB design, the goal is to allow users to describe a product like a smartphone and have the AI generate the design, significantly reducing the time and complexity.

What's the difference between Flux and traditional PCB design tools?

Traditional tools lack AI integration and direct supply chain connectivity. Flux integrates AI for design assistance and real-time access to component availability and pricing, preventing design delays due to unavailable parts.

How does Flux handle component selection and availability?

Flux integrates with semiconductor distributors to provide real-time information on component stock and lead times. The AI agent can also suggest cheaper or more available alternatives, preventing costly project delays.

What kind of AI infrastructure does Flux use?

Flux uses frameworks like LangChain and LangGraph for building its AI agents. They also leverage LLMs and focus on context engineering and injecting domain-specific knowledge to guide the AI's behavior.

⚡️ Building the AI Hardware Engineer with Matthias Wagner, Co-founder of Flux

Latent Space Podcast

Science & Technology4 min read50 min video

Nov 22, 2025|1,901 views|35|10

Save to Pod

Key Moments

TL;DR

Flux builds AI hardware engineers using LLMs to automate PCB design, slashing creation time from months to minutes.

Key Insights

Hardware design tools have lagged behind software development tools for decades, with Flux aiming to bridge this gap.

Flux utilizes AI agents, powered by LLMs, to transform product briefs into manufacturable PCB designs rapidly.

The integration of LLMs significantly accelerated Flux's development, particularly with advancements in tool-calling capabilities.

Flux combines a collaborative, browser-based CAD tool with AI agents for design, component selection, and supply chain integration.

The platform offers real-time pricing and availability checks for components, mitigating supply chain delays.

Flux is democratizing hardware development, enabling individuals and companies to create custom electronic devices more affordably and efficiently.

THE EVOLUTIONARY GAP IN HARDWARE DESIGN TOOLS

Matthias Wagner, co-founder of Flux, highlights a significant disparity: while software development tooling has undergone immense advancements, hardware design tools have remained largely stagnant. This observation, prompted by his personal experience and a desire to innovate after years in software and a hiatus working on Burning Man projects, fueled the founding of Flux in 2019. The core realization was that the manufacturing supply chain for hardware had modernized, allowing individuals to access custom parts, yet the design tools hadn't kept pace, creating a bottleneck in innovation.

THE BIRTH OF THE AI HARDWARE ENGINEER

Flux's journey initially focused on building a 'smart design tool,' with the vision of integrating machine learning to automate electronics and PCB design. The advent of Large Language Models (LLMs) in 2022 provided a significant technological leap, turbocharging Flux's capabilities. Unlike software AI tools that could leverage existing platforms like VS Code, Flux had to build its browser-based CAD tool from scratch, creating a modern architecture designed for AI integration and user feedback loops, akin to Figma's approach in its domain.

LEVERAGING LLMS FOR DESIGN AUTOMATION

A key inflection point for Flux was the integration of AI chat capabilities before GPT-4's public release, initially helping with information gathering and reducing reliance on external searches for datasheets and specifications. As LLM tool-calling capabilities improved, Flux could empower its agents to perform actions, such as adding components, wiring, and running design reviews. This shift from assisting with information to executing tasks marked a significant advancement, aligning with evolving user expectations for AI tools to deliver complete solutions from prompts.

FLUX'S COMPREHENSIVE DESIGN PLATFORM

Flux offers a real-time collaborative, browser-based CAD tool with versioning and team functionalities. Its integrated AI agents can assist users in various ways, from answering questions to executing complex design tasks based on product briefs. The platform also features integrated supply chain intelligence, providing real-time cost and availability data for components, a critical factor in preventing design-and-build delays that plague traditional tools which often overlook component stock.

INTELLIGENT COMPONENT SELECTION AND PERSONALIZATION

A powerful feature of Flux is its agent's ability to intelligently suggest component replacements, considering factors like pin compatibility, availability, and pricing. This process, which would manually take hours, can now be completed in minutes. The platform also incorporates a user-driven knowledge base, allowing for personalization at individual, project, or organizational levels. This memory mechanism is crucial for AI systems to adapt to specific contexts and user preferences, mirroring the need for real-time personalization seen in other AI applications.

ADDRESSING THE HARDWARE LEARNING CURVE

Flux aims to democratize hardware development by simplifying the often steep learning curve for software engineers venturing into hardware. The AI agents can guide users through complex concepts and debugging, identifying issues like the necessity of decoupling capacitors, which are often overlooked by novices. The platform can generate an execution plan from a product brief, allowing even complex projects like building a custom smart speaker to be initiated with clear steps, thereby empowering more individuals to bring their hardware ideas to fruition.

THE FUTURE OF HARDWARE CREATION: PROMPT-DRIVEN DESIGNING

Flux's grand vision extends beyond PCB design to enabling users to 'prompt hardware into existence,' similar to how LLMs generate text or code. The ultimate goal is to allow users to describe a desired electronic device—like a smartphone—and have the AI generate a manufacturable design, fundamentally changing product creation. This vision challenges the traditional reliance on mass manufacturing for affordability, suggesting that AI-driven design and automation can lead to highly customized, cost-effective electronics for a wide range of applications beyond high-end consumer devices.

BUSINESS GROWTH AND INDUSTRY IMPACT

Flux has experienced significant growth, attracting 7,000 paying customers and achieving a 26-fold increase in revenue year-over-year, largely through organic word-of-mouth and content marketing. Its user base includes individual engineers at large companies and established enterprise clients. The company sees a broad industrial use case, targeting sectors from vending machines and traffic lights to farm automation, wherever electronics and automation intersect. Flux aims to disrupt the traditional OEM model by enabling more customization and capability in electronic products.

ENGINEERING CHALLENGES AND ADVANCEMENTS

The development of Flux involves continuous iteration on its AI agents, balancing automated evaluations with subjective 'vibe checks' to ensure usability and performance. Prompt management is a significant challenge, with the team developing strategies for reusability and domain expert involvement. While evaluation metrics are crucial for deterministic tasks, like component retrieval, iterative refinement is essential for more complex, agentic features. The platform's browser-based canvas also introduces unique rendering and performance challenges, especially for mobile and WebGL-dependent applications.

Mentioned in This Episode

●Products

●Software & Apps

●Companies

●Concepts

●People Referenced

Flux AI Hardware Engineering Workflow

Practical takeaways from this episode

Do This

Consider AI assistance for tasks like information gathering, component selection, and design.

Collaborate with AI agents: iterate and provide feedback on proposed designs and plans.

Specify detailed requirements for AI-driven projects to improve output quality.

Leverage user-generated libraries and community knowledge for component data.

Utilize AI for identifying cost-effective component replacements and checking stock availability.

Focus on prompt engineering and context to guide AI agent behavior.

Explore Flux for democratizing hardware design and lowering the barrier to entry.

Test exported design files in third-party software for verification.

Leverage AI for debugging hardware designs and identifying potential issues.

Avoid This

Don't underestimate the complexity of hardware design; AI assists but doesn't fully replace human expertise yet.

Don't expect AI to perfectly handle complex canvas interactions or dragging gestures without specific training.

Don't rely solely on AI for every aspect; manual review and intervention are still necessary.

Avoid vague project briefs; the more detailed your input, the better the AI's output will be.

Don't assume legacy tools offer the same AI and supply chain integration capabilities as Flux.

Component Pricing Tiers (Example: Volume vs. Price)

Data extracted from this episode

Volume	Price per Unit
1	$264
10	$166
48	$125

Common Questions

Flux is a company building an AI-powered design tool for hardware engineers. It aims to automate and streamline the process of designing Printed Circuit Boards (PCBs) and other electronic hardware, similar to how AI tools assist software engineers.