01 · RESEARCHERS
Our team has led product security, from silicon to AI, at Google, NVIDIA, NIO and TikTok. We know how hardware really behaves, and which flaws matter to the people using it.
02 · AI AGENTS
We research how AI can make that judgement reliable beyond human scale, for every product that ships, not just the few that get an expert review.
IN STEALTH · FRONTIER PRODUCT SECURITY
NovAscent Labs is a frontier research lab for the security of connected devices, AI-native hardware and robots.
WHAT WE RESEARCH
From how the hardware behaves to the person relying on it.
We research how AI can make product security reliable beyond human scale, by following each behaviour from the chip, through the finished product, into real-world use.
-
01 · HARDWARE
How it behaves
How chips, firmware and sensors really behave, edge cases included.
-
02 · PRODUCT
How it shows up
How that behaviour surfaces in the finished product and its on-device AI.
-
03 · REAL-WORLD USE
How it holds up
How the product performs in the conditions it is actually used in.
-
04 · PEOPLE
Who it affects
What a failure means for the user in front of it.
Research
- Can AI trace a hardware behaviour all the way to its impact on real users?
- Can AI agents find hardware flaws as reliably as senior researchers?
- How far can crafted sensor inputs steer an on-device model?
- When does a robot's security flaw become a safety failure?
- What evidence should AI-assisted testing give a CRA assessor?
Publications coming soon
Advisory
- Architecture review. Security design for products at any step.
- Hands-on assessment. Researchers and AI agents, together on your hardware.
- CRA readiness. Reporting duties apply from 11 Sep 2026; full application 11 Dec 2027.