Physics AI for electronics
Software thatdesigns hardware.
Describe the product you need in plain language. Our Physics AI generates a complete, manufacturable design and gets it right the first time.
Join the waitlistSearch, not guesswork
We generate likely candidate designs, and let our Physics AI choose.
01
Describe
in plain language
One set of requirements: what the product does, where it operates, what it must not exceed. Every requirement becomes an executable constraint.
02
Generate
thousands of designs
Evolutionary search opens the full trade space, rather than betting everything on a single generated answer.
03
Evaluate
in milliseconds each
Each candidate scored against calibrated physics where a real simulator would take hours. Non-compliant designs are killed early. What returns is manufacturable and certifiable, not one lucky guess.
This is our proprietary approach, not a language model wrapper. Design is search over calibrated physics, not next-token prediction.
The cost of one iteration
A typical hardware design takes 2 to 5 build-test cycles to get it right.
$30K+
Prototype and lab, per spin.
$100K+
When respins compound.
6+ mo
Loop latency per cycle.
Months of senior engineering time. Vectis.ai turns it into hours.
Hardware becomes software-like
Why this wins
A good generator plus instant checking beats a brilliant one-shot generator.
Most AI-hardware efforts, and the design tools around them, are chasing a smarter design step: a genius engineer that outputs the perfect board in one shot. It is brittle, and none of them can tell you a design works without building a prototype. We made a different bet. We generate a good engineer and make it possible to check any design against real physics instantly. Cheap, instant verification is what lets software iterate freely, and it is exactly what hardware has always lacked.
Pricing
A seat for every engineer. A fee for every validated design.
Your engineers work in Vectis.ai on a per-seat subscription. Beyond that you pay only when a design comes out validated against calibrated physics and ready to manufacture. The thousands of candidates explored and rejected on the way there cost you nothing, and there is no metering per simulation.
Why now
AI is moving past language to the physical world.
The researchers who built modern AI are betting billions that language models cannot reason about the physical world. The next wave is grounded in it.
AMI Labs
Founded by Yann LeCun
$1.03B seed
World Labs
Founded by Fei-Fei Li
$1.23B raised
PhysicsX
Series C
$300M
Emmi AI
Acquired by Mistral
~$340M
Proven for fluid dynamics and mechanics. Vectis.ai brings it to electronics.
Where we are
First revenue
A paid engagement with a Tier-1 industrial manufacturer.
Pilots in scoping
Active scoping with manufacturers in the EU and the US.
Innovation grant awarded
Non-dilutive funding from VLAIO, the Flemish innovation fund.
NVIDIA Inception
Member of the NVIDIA Inception programme for AI startups.
Backed and supported by
Built by the people who felt the pain
Wael Elrifai
CEO / CTO
Repeat deep-tech founder with a prior exit. Former Global VP at Hitachi Vantara, where he led a 100-person global engineering organisation across $80M in annual revenue. Founded Peak Consulting, a machine learning firm acquired in 2013. Co-founded Proxximos, leading electrical engineering, firmware, and machine learning. Board member and Treasurer of AIOTI, the European Commission's IoT alliance, and a Fellow of the British Computer Society. MSEE, and an MBA from Oxford.
Brooke Treseder
COO / GTM
Twenty-five years scaling enterprise revenue, starting at Microsoft and SAP. SVP Revenue Operations at Checkout.com from pre-Series B to $40B, where she built the GTM engine, then SVP Business Operations at Volt.io and VP Revenue Operations at CoachHub. Named one of the top fifteen revenue operations leaders in the industry. First twenty years in Silicon Valley.
Join the waitlist.
We are working with a small number of manufacturers and investors ahead of general availability.