The systems that decide what you see.
AI agents and the workflows they run. Ad auctions that price attention in milliseconds. Answer engines that reply before you click. Analytics pipelines that count it all. Digitarise takes this machinery apart — as engineering, not folklore — with live instruments you can operate and claims traced to public specs and public data.
Latest insights
sorted by date ↓Anatomy of an AI agent tool-calling loop
A neutral systems reference for the loop that lets a language model select, call, and verify external tools.
Measuring Accessibility, Speed, and Tracker Load
A standards-grounded method for auditing usability barriers, runtime performance, and third-party tracking weight.
Search ad auction rank and cost per click
How rank scores, quality signals, floors, and second-price logic determine search ad position and billed clicks.
WCAG 2.2: what the standard requires
A systems-level reference to WCAG 2.2 conformance, thresholds, failure modes, and minimal implementation.
Core Web Vitals: how the metrics work
A systems-level explanation of LCP, INP, and CLS: what they measure, which APIs define them, and how to diagnose failures.
Operate an instrument
live — no reading requiredEvery analysis here ships with a working model of the system it describes. This one is the generalized second-price auction that prices most search advertising: drag any bid or quality slider and watch rank — and what each advertiser actually pays — recompute.
Advertiser B holds position 1 without the highest bid — quality is doing the work.
What we examine
8 fieldsAI agents & workflows
What a tool-calling loop actually does: orchestration, state, failure modes — and the automations built on top of them.
Advertising systems
The auctions and delivery algorithms that price attention, examined as the distributed systems they are.
AI search & answers
How answer engines retrieve, select, and cite sources — the machinery that now decides visibility on the web.
Digital analytics
Event pipelines, identity, and measurement models: what the numbers actually count, and what they miss.
Web performance
Latency, caching, and rendering budgets, measured where the user is.
Accessibility
The contracts between interfaces and assistive technology, tested against WCAG.
Privacy
Data minimisation and consent mechanics, traced through real systems.
Web standards
What the normative text actually says — read from the specification, not the summary.
Every claim traced to a public specification or a public dataset. Nothing for sale.
Read the methodology