Aliddo Labs — Research-grade technology, built.

01 Thesis

Defensible technology comes from original research — and the best research deserves to be built.

We believe the only durable advantage in artificial intelligence and Web3 is original research. Most products are thin wrappers over the same models and the same rails. We work the other direction — from first principles in causal inference and cryptography outward into infrastructure that others cannot easily copy. Aliddo Labs is a research engine that ships.

02 Ventures

The research, built into infrastructure.

Three lines of work, each defensible because the technology underneath is hard to reproduce. Web3 infrastructure and applied causal systems, bound together by verifiability.

Flagship · Web3 infrastructure In development

01

DeParity

// Agentic finance infrastructure

The settlement layer for the agent economy. As autonomous systems begin to transact, they need rails that are programmable, verifiable, and compliant by construction. DeParity provides ZK-secured payment and settlement infrastructure for AI agents operating across borders and chains.

For — agent platforms, fintechs, and settlement networks moving value autonomously.

Zero-knowledge verification Settlement & payment rails Compliance-native Cross-chain

Settlement pathZK-secured

Agent intentautonomous transaction request

Policy & compliance checknative to the protocol

ZK verificationproven correct, data withheld

Cross-chain settlementfinal, programmable

AI · Decision systems Engagements open

Correlation vs. causedo(X)

02

Applied Causal AI

// Causal inference as a commercial moat

Most machine learning finds correlations. The decisions that matter require cause. We build decision systems on causal inference — models that reason about intervention and counterfactuals, not just pattern — for domains where being right about why is the whole game.

For — teams where being wrong about cause is expensive: capital, health, policy, pricing.

Intervention Counterfactuals Decision systems

Venture 03 The bridge · AI ⇄ Web3 Research → product

03

Verifiable Causal Inference

// VCI — where the two sides of the lab meet

A causal claim is only as trustworthy as your ability to check it. VCI uses zero-knowledge proofs to make causal and computational claims cryptographically verifiable — proving that a result was produced correctly without exposing the underlying data or model.

VCI connects the AI side to the Web3 sideprove, don't expose

AI side A causal claim A counterfactual or intervention result, produced by a model on private data.

VCI Zero-knowledge proof Establishes the result was computed correctly — without revealing the data or the model.

Web3 side Verifiable on-chain Anyone can check the claim holds. Trust becomes a matter of proof, not reputation.

For — settings where a result must be trusted by parties who cannot see the data behind it.

Zero-knowledge proofs Verifiable computation Privacy-preserving

03 Research

The engine beneath the ventures.

The ventures stand on original theory. This is the bedrock — slower, deeper work on what intelligence is and how to verify it.

Framework

Compression Under Agency — CUA

A framework for intelligence as compression under the constraints of agency — the theoretical spine connecting world models, active inference, and the lab's applied work.

The spine that connects everything above.

Active

Verifiable causal inference, world models & active inference

Ongoing work across verifiable causal inference, world models, and active inference — the open threads that feed the ventures.

Foundational interests

i

World models

Learned, predictive representations of an environment an agent can plan within.

ii

Active inference & the Free Energy Principle

Perception and action as the minimization of expected surprise.

iii

JEPA-style architectures

Predicting in representation space rather than reconstructing raw signal.

iv

Compression as a theory of intelligence

Treating the search for compact structure as the core of cognition.

04 How we work

We build from first principles.

01

We build from first principles.

02

We prefer methods we can verify over results we have to trust.

03

We work on long horizons — on problems where the technical depth is the point.

04

We would rather be right slowly than wrong fast.

05 Work with us

Build with us.

Aliddo Labs is a venture lab, not a consultancy. We co-build infrastructure where original research is the moat — and take on a small number of applied engagements where causal rigor or verifiability decides the outcome.

Start a conversation info@aliddo.com

01 — Ventures & partnerships

Co-founding and building companies on the lab's research.

DeParity is the first. We partner with founders and teams operating at the same frontier in agentic finance and verifiable systems.

02 — Applied engagements

A limited number of client problems, taken on directly.

For organizations where being right about cause, or being able to prove a result, is the whole game — not a feature.