Practical thinking for
private AI.
Perspectives on enterprise AI, foundation models, secure agents, and the capability organisations build when they retain control of their intelligence.
View illustrative use cases →When an AI agent gets access: the enterprise security lesson
Public discussion around AI-agent security is a reminder that tool access, data access, and autonomy must be designed together—not added after deployment.
18 July 2026Building a private AI platform: the control plane and data plane
A practical workflow for bringing enterprise data, models, and agents together without losing control of cost, security, or intellectual property.
12 June 2026The compounding value of in-house AI capability
A reusable AI foundation creates more strategic value than a growing collection of disconnected tools.
22 April 2026Designing AI agents for accountable work
Enterprise agents must be designed around authority, evidence, and human accountability—not just tool access.
24 March 2026RAG, fine-tuning, and the discipline of model strategy
The right model architecture is determined by the task, the knowledge lifecycle, and the standard of evidence required.
19 February 2026Generative AI requires a trusted knowledge layer
Reliable AI starts with the quality, provenance, and permissions of the information it is allowed to use.
15 January 2026Private AI is an operating model, not a deployment choice
The organisations that create lasting value from AI will govern intelligence as deliberately as they govern data and risk.