30 June 2026
7 min read
Migrating a workload without a quality regression
The cheapest possible AI is one that produces the wrong answer. How to move a production workload and prove nothing got worse.
How high-volume AI actually gets cheaper.
Working notes on what it costs, how it is built and how it is run when you move your AI onto machines you control. Nothing behind a form.
19 May 2026
5 min read
Sizing capacity for spiky demand
Provisioned infrastructure has one genuine failure mode: a spike you did not size for. Here is how to design so it never becomes an outage.
22 April 2026
6 min read
Keeping inference inside the perimeter
Data residency stops being a contractual promise and starts being a deployment fact, and what that changes for the people who have to evidence it.
4 March 2026
7 min read
When dedicated infrastructure beats an API
The crossover is not a matter of taste. It is a calculation, and most teams can do it from data they already have.
11 February 2026
6 min read
The anatomy of a per-token bill
Where the money actually goes when you run AI at volume, and why the invoice grows faster than the value it produces.
Would rather talk it through than read it?
The first call is a technical conversation about what you run, not a pitch. Bring your invoice and the AI job you run most.
Talk to Runvo
Tell us the size of the firm and what your client contracts require.