B&A takes high-agency founders from an irresistible desire to solve problems in the physical world, to a seed-investable concept or working prototype — built in Shenzhen, where your factory floor is a bike ride from your workbench. Next cohort starts July 1.
Modular, software-controlled factory units. Drop one in, configure it, start producing. Like Shopify, but for manufacturing. No 18-month integration projects.
Hardware that makes power grids smarter. Sensors and controllers that unlock capacity in existing infrastructure, so new solar and wind can actually connect.
Replace fossil fuel furnaces in factories with electric heat pumps. High-temperature. Real industrial use. The biggest decarbonization opportunity nobody talks about.
Robots for the messy jobs: unloading mixed boxes, sorting damaged returns, depalletizing chaos. Not the easy stuff on flat floors. The hard stuff. Pay-per-use model.
Sensors and control systems that make steelmaking cleaner. Measure CO2, optimize energy, improve yield. Retrofit existing plants instead of building new ones.
Affordable compute modules that run AI models on the device itself, not in the cloud. The brain that goes inside every robot, sensor, and smart machine.
The next great AI companies will not live only in chat windows. They will be built into devices, workflows, and machines that perceive, decide, and act.
Model capability, edge compute, sensors, and connected hardware have finally caught up with each other. Physical AI is no longer theoretical — it is buildable now.
In this category, every deployment is a data advantage. The companies that win will turn real-world usage into better models, better products, and faster iteration.
When your factory floor is a bike ride from your workbench, you iterate in days, not quarters. B&A founders don't outsource to Shenzhen — they build from inside it.
The next generation of great products won't separate AI, industrial design, and production into phases. They'll co-evolve. B&A puts founders where all three disciplines overlap from day one.
Shenzhen's hardware ecosystem is unmatched — but historically hard to access for international founders. B&A is the bridge: an English-speaking, globally-networked program embedded in the heart of the world's hardware capital.
Shenzhen should not only be where products are made. It should be where the next generation of AI-native product companies begins.
A robotics founder needed a small-batch BLDC motor supplier but couldn't get past MOQ walls on Alibaba. B&A expert Patrick Liu introduced them to a workshop in Baoan that specializes in prototype-grade motors. First samples arrived in 9 days. Without the intro, the founder estimates it would have taken 2–3 months.
An embedded ML team had impressive technical specs but kept hearing "it looks like a breadboard" from investors. Working with Sarah Chen, they developed an industrial design language, a brand, and a physical prototype that communicated the product vision. They closed their seed round within 8 weeks.
A European IoT founder was quoted $45K for a functional prototype from a German engineering firm. In Shenzhen, through the B&A manufacturing network, the same prototype was produced for $12K — with faster turnaround and direct access to the engineering team for revisions.
A smart home device team was planning to get China CCC certification after production — a common mistake that adds 3–6 months to launch. Anna Hoffmann restructured their timeline to run certification testing in parallel with final engineering, saving 4 months off their go-to-market.
We help founders define the wedge, the use case, and the moat. The goal is not a niche device with limited upside, but a product company with the potential to grow into a category.
We help teams test technical risk, manufacturability, performance, cost, and user experience early — before expensive decisions lock in.
For physical products, brand is not an afterthought. Product identity, industrial design, usability, and engineering constraints need to be developed together from the start.
We help founders validate demand, distribution, and competitiveness. Teams can test not only how fast a product can be built, but how well it stands up in one of the world's most demanding markets.
We back companies that can become core to a workflow, environment, or category — not accessories around someone else's platform.
The first use case can be focused, but it must lead somewhere bigger. Sharp entry points with the potential to expand into large, durable businesses.
The strongest AI-native companies improve through real-world deployment. Data, usage, and feedback should make the product better and the company harder to replace.
Too dependent on another platform to become foundational.
Businesses that end where they begin.
Needs a scientific leap before proving value.
Custom engineering, not scalable products.
Products that don't get stronger with use.
Software-defined microfactory cells for rapid deployment, copy-exactly repeatability, and high-mix assembly.
Register interest →Dynamic line rating, feeder measurement, power-flow control devices to unlock distribution capacity.
Register interest →Hardware replacing fossil process heat — high-temp heat pumps and thermal storage for industrial customers.
Register interest →Robotic systems for high-friction warehouse tasks involving unstructured manipulation and messy interfaces.
Register interest →Sensors, control retrofits, and measurement tools for CO₂, energy, or yield improvements in steelmaking.
Register interest →Edge inference hardware lowering TCO through improvements in power, latency, reliability, and deployment simplicity.
Register interest →Cohort 01 starts July 1. 10 founder slots. Applications reviewed on a rolling basis.