Insights
Notes on shipping AI systems that survive production.
On-prem LLM deployment, agentic AI, AI transformation, and the operations that keep deployed systems alive.
Agentic AI
6 minTokenmaxxing: giving employees frontier AI is not AI transformation
Meta employees reportedly burned 73.7 trillion AI tokens in a month. The lesson is not to use less AI. It is that handing staff frontier-model access is not transformation.
Read
On-prem LLM
6 minWhat sovereign AI actually means for GCC governments
Sovereign AI is not a server in a local data centre. For GCC governments it is a full practice: owned models, controlled infrastructure, and the audit controls a regulator expects.
Read
On-prem LLM
6 minOpen-weight vs open-source AI models: the difference that matters
Most 'open-source' AI models are actually open-weight. The distinction is not pedantry. It decides what you can legally deploy, audit, and own.
Read
Government
9 minA frontier model just became a controlled export. Read what it means for China.
For 19 days, a US frontier model was placed under national-security export controls and switched off for every foreign national on earth. The precedent matters more than the incident.
Read
On-prem LLM
6 minFable is back online. The 19 days it was gone are the part to remember.
Claude Fable 5 returned on 1 July after a 19-day, policy-driven shutdown. The outage was not technical. That is exactly why it should change how you plan.
Read
Government
7 minUAE is moving government onto AI agents. That is the signal to read.
The UAE plans to run half its government on agentic AI by 2028. The number worth watching is not the ambition. It is that agents are being treated as public infrastructure.
ReadAgentic AI
6 minAgentic AI vs workflow automation: when each one wins
Agents and process automation are different tools with different cost curves. Picking the wrong one inflates the build and the operating bill at the same time.
ReadTransformation
5 minAI transformation is a deployment problem, not a strategy problem
The slide decks land. The systems do not. Closing that gap is what AI transformation actually is, and it is engineering work.
ReadOperations
7 minManaged AI operations: why deployed agents need ops
Agents are not ship-and-forget software. Prompts rot, models update, data shifts. Operations is how the system you shipped stays the system you shipped.
ReadOn-prem LLM
8 minOn-prem LLM deployment is a different engineering practice
Sovereign deployment is not a different vendor. It is a different practice with its own constraints, its own tradeoffs, and its own operating model.
ReadOn-prem LLM
9 minRAG on private data: the parts that actually matter
Retrieval-augmented generation on private corpora succeeds or fails on the parts of the system nobody puts in the architecture diagram.
Read
Start with the assessment.
A fixed-scope AI readiness assessment: your workflows, your data, your highest-ROI agent use cases, and a deployment roadmap. Two to four weeks.