ARTICLE ARCHIVE / SMARTER BI
Articles
Research briefs and practical guidance on AI agents, recommender systems, and responsible content discovery.
-
A completed shell is not an operating data centre
A powered-shell handover, occupier fit-out and first-building opening answer different questions. Manor Farm and Wapseys Wood show why completion dates need a delivery boundary.
-
AI infrastructure research: a guide to the series
Start with the UK’s competitive position, then examine project status, planning consent and power claims. A reading guide to four Smarter BI analyses.
-
What does “power secured” actually mean for a data centre?
A grid application, a reported agreement and an energised connection are different things. Three projects show how to read the claim without overstating it.
-
Planning is not delivery: reading data-centre consent properly
Consent, reserved matters, condition approvals and physical works answer different questions. Two Slough applications show why planning progress is not delivered capacity.
-
Why data-centre project status is harder to verify than headline MW suggests
Four projects show why planning decisions, power claims, customer commitments and delivery forecasts need different evidence, not one status label.
-
Can the UK compete for the next wave of AI infrastructure?
Can the UK compete in AI infrastructure? Compare workload needs, location value, electricity costs and delivery evidence across competing markets.
-
7 Reasons Digital Publishers Should Use Recommender Systems
How responsible recommendations can improve content discovery, reactivate valuable archives, support editorial strategy, and create measurable reader journeys.
-
Semantic IDs: Snapchat’s blueprint for smarter recommendations
How Snap uses semantic IDs to improve ranking and generative retrieval—and what its experience with codebook collapse, collisions, and evaluation teaches other recommender teams.
-
Agent Harness Engineering
A practical summary of the ETCLOVG framework—and why execution, tools, context, orchestration, observability, verification, and governance matter as much as the model.