Azetta AI: Physics Grounded AI Research Lab.

Azetta.ai — Physics Grounded AI Research Lab

We build white-box AI. Explainability is our engine: by understanding every decision a model makes, we make it better, more efficient, and safer — by design, not by patch.

ⵟ — YAT Kernel RKHS click to add neurons
scroll

// OUR RESEARCH

Pioneering the Field of Physics Grounded AI.

The mathematics behind white-box AI — one new kernel, three papers on arXiv.

YAT KERNEL

(x, w) = (x·w + b)² / (‖x − w‖² + ε)

Unlike a dot product, this physics-grounded Mercer kernel measures both alignment and proximity — how strongly a weight acts as an attractor for an input. Each neuron bends representation space into its own gravity well: monosemantic neurons by design, interpretability with no post-hoc approximation.

// THE PROBLEM

AI Should Be Safe, Transparent, and Sustainable.

"People outside the field are often surprised to learn that we do not understand how our own AI creations work. This lack of understanding is unprecedented in the history of technology."

— Dario Amodei, CEO Anthropic · 2025

▶ The Black Box Problem

Today's AI is a black box. Models grow larger every year, yet no one can explain why they hallucinate, discriminate, or fail. Even reverse-engineering them barely works — Anthropic admits that fully understanding a model would take more compute than it took to train it.

We believe this is because frontier models are fundamentally opaque, and that a viable, scalable solution has to go back to the foundations. Our mission is to rebuild AI so that every decision is explainable and every behaviour is correctable — models that improve and scale in smarter, far more efficient ways.

Prompt"the capital of Mexico is"
Output
· · · more layers above · · · · · · more layers below · · · BLACK BOX 96 layers · 8,192 neurons each INPUT HIDDEN LAYERS (96 total) OUTPUT
The core problem: Each neuron fires for dozens of unrelated concepts at once — geography, grammar, culture, sentiment all entangled. This superposition makes models opaque by construction.

▶ The Symptoms — Every Major AI Risk Traces Back to Interpretability

Alignment

click to reveal

Harmful AI incidents hit 233 in 2024 — +56% YoY (Stanford HAI). We cannot verify what a model is optimising for — goals appear aligned but remain uninspectable.

Hallucinations

click to reveal

LLMs hallucinate on 75%+ of legal queries. We can detect errors in outputs — without interpretability, we cannot stop them at the source.

Bias & Fairness

click to reveal

LLMs preferred white-sounding names 85% of the time in hiring simulations. EEOC's first AI bias settlement: $325K. Bias lives inside weights — beyond our reach.

Privacy Leakage

click to reveal

Researchers extracted PII from ChatGPT for ~$200. 5%+ of outputs are verbatim training copies. No mechanism exists to audit what a model retained.

▶ Why It Matters to You

01

The Ceiling on Progress

Model self-improvement requires interpretability. Without it, we cannot guide models reliably or safely.

02

Regulated Industries Are Locked Out

Healthcare, finance, legal, and defence cannot deploy black-box AI where decisions must be audited or legally defended.

03

Full Automation Requires Control

You cannot delegate what you cannot inspect — and today, we cannot inspect these systems.

// OUR APPROACH

Reimagining AI Through a Physics Lens.

⬡ WHITE BOX Tracing Neuron Activation
Prompt"the capital of Mexico is"
Output
· · · interpretable by design · · · INPUT INTERPRETABLE LAYERS (3) OUTPUT

The fix isn't inspecting the black box after the fact — it's building so nothing is opaque to begin with. Information has its own physics; reading it yields models that are interpretable, steerable, and efficient by design.

Interpretable
One neuron, one concept. Full audit trail from input to output.
Steerable
Target specific neurons. Correct behaviour directly — no retraining.
Efficient
10× fewer parameters, 90% faster training.
Performant
Performance matches or exceeds State of the Art.

// WHO WE ARE

The Founding Team.

Mathematical rigour, entrepreneurial experience, and product acumen.

Taha Bouhsine

Taha Bouhsine

Co-Founder & CEO

AI RESEARCHER GOOGLE DEV 2× FOUNDER

AI Google Developer Expert and architect of the Physics Grounded AI framework. Mathematician, computer scientist, and electrical engineer researching interpretable, efficient neural networks, representation learning, and the geometry of how models understand the world. Builds rigorous, explainable AI from first principles.

Jose Miguel L.

Jose Miguel L.

Co-Founder & CPO

EX-APPLE COLUMBIA MBA + MS AI/ML

Ex-Apple Engineering Product Manager for AI/ML products. Founding team at YC-backed startup, leading Product, Data and Tech teams. Schwarzman Scholar and Columbia MBA + MS in AI/ML.

Help us rebuild AI from first principles.

We're building the foundations of transparent, physics-grounded AI — and we're looking for the people who want to build it with us.