Field notes
Practical writing on private AI infrastructure.
No hype. No AI trend pieces. Writing that helps businesses make better decisions about compute, models, data, and deployment.
The hidden cost of per-token AI billing at scale
What looks affordable at 50 users becomes a significant operating cost at 200. We break down the unit economics and when dedicated compute makes more financial sense.
Read articleOpen-weight vs frontier models: when each is right for business workflows
Not every task needs the most powerful model. The decision framework we use when choosing models for client deployments.
Read articleWhy private AI infrastructure is becoming practical for smaller businesses
Three years ago, running your own AI required a data centre. Today, a compact node in your office can handle the AI workloads of a 30-person firm.
Read articleWhat to ask before connecting AI to sensitive company data
Most businesses underestimate how much thought goes into safe AI deployment. A checklist of the questions to answer first.
Read articleCloud vs on-premises AI: a practical guide for non-technical buyers
The right deployment model depends on compliance, existing infrastructure, and how sensitive your data is. A plain-English breakdown.
Read articleHow to size AI compute for a small professional services firm
Matching compute to workload is the most important decision in a private AI deployment. We walk through how we approach sizing.
Read articleSubscribe
Get new articles by email.
One article every few weeks. No marketing emails.