Research
Writing on AI value realization, privacy-enhancing technology, provable attribution, and enterprise AI cost optimization.
Subscribe via RSS- ·AI Spend · Attribution · Shapley · Enterprise AI
Why Your AI Spend Audit Is Lying to You — The Attribution Problem
Most enterprise AI spend audits add up costs and call it a day. The number you actually need — which dollar produced which dollar of value — requires solving a harder problem. Here is how Shapley-based attribution does it.
Read - ·Privacy-Enhancing Tech · Differential Privacy · Homomorphic Encryption · Enterprise AI
The Privacy-Enhancing Tech Stack for Enterprise AI — When You Need What
Differential privacy, homomorphic encryption, federated learning, secure enclaves, zero-knowledge proofs. Five tools, very different use cases. A practical buyer's guide for enterprise AI workloads.
Read - ·Data Strategy · Enterprise AI · Maturity Model · Data Asset Valuation
Your Data Is Your Identity — A Maturity Model for Enterprise AI
For people and for agents, identity is the aggregate of data and verified capability over time. A five-stage maturity model that tells you where your organization actually is — not where the board deck says you are.
Read - ·Vision Transformers · Attention · SVD · Continual Learning
Co-Activation Spectra as a Map of General vs. Task-Specific Structure in Pretrained Vision Transformers
SVD analysis of attention head co-activation matrices reveals layer-dependent rank structure that characterizes general vs. task-specific representations.
View Abstract - ·Continual Learning · Catastrophic Forgetting · Neural Architecture · Machine Learning
Latent Topology Networks
A continual learning framework that addresses catastrophic forgetting by discovering and permanently committing sparse network structure during training.
View Abstract - ·Agent Marketplaces · Provable Attribution · Shapley Values · Privacy
Contextual Capability Verification in Agent Marketplaces
A verification framework for agent marketplaces where capability is identity: the posterior evidence that an agent can deliver a specific task at the contracted price with required confidence.
View Abstract - ·Shapley Values · Data Valuation · Privacy · Threshold Cryptography
Marginal Value Decomposition of Overlapping Data Sources Under Disclosure Constraints
A cooperative game-theoretic framework for computing each data source's marginal contribution via Shapley values, with threshold cryptographic attestation and privacy-preserving comparison.
View Abstract