Notes from inside the build.
We write about the parts of this work that are hard to find written down honestly: what modernization actually costs, where automation pays and where it does not, and what happens in the weeks after a system goes live.
AI AutomationEnterprise AI agents doubled in four months. Governance did not.
Deployments roughly doubled between December and April. Monitoring coverage moved about five points, and 85% of organizations still cannot name a person accountable for what an agent does. The gap is the whole story.
July 29, 2026 · 7 min read
Systems ModernizationBuild versus buy in 2026: what changed, and what absolutely did not
The cost of building collapsed, so the answer flipped for a whole category of software. It did not flip for all of it, and the projects that fail now fail for exactly the same reason they failed in 2015.
July 27, 2026 · 5 min read
Systems ModernizationThe integration tax is falling, and it changes which projects are worth doing
For twenty years, connecting systems to each other was the most expensive line in the estimate and the reason most good ideas died. In 2026 that line got dramatically cheaper. Very few organizations have re-run the numbers.
July 24, 2026 · 6 min read
GovernanceShadow AI is already in your operation. The question is what it has read.
Roughly two thirds of employees use AI at work and roughly a fifth of employers have a policy about it. Banning the tools does not close that gap. It just moves the usage somewhere you cannot see.
July 23, 2026 · 5 min read
AI AutomationWhy the right AI model for a regulated business is probably a small one
The frontier model is the one in the headlines. The one that survives a compliance review is usually smaller, cheaper, fine-tuned on your own material, and running somewhere you can point at on a map.
July 20, 2026 · 6 min read
HealthNutrition practice has an adherence problem, not a documentation problem
Almost every AI product aimed at clinical practice automates the notes. Notes were never where the outcome was decided. The plan is kept or lost in the week nobody is watching, at the moment someone orders lunch.
July 17, 2026 · 5 min read- Systems Modernization
What replacing a spreadsheet-run operation actually costs
The quote is the smallest number in the project. Here is the rest of the bill: migration, parallel running, and the six weeks after go-live that decide whether any of it worked.
July 15, 2026 · 4 min read 
Data & AnalyticsYour AI project is a data project, and most of the work happens first
Almost every AI request we receive resolves, within an hour of looking, into four ordinary data problems. The good news is that you do not need clean data. You need the specific data this one decision depends on.
July 13, 2026 · 5 min read
AI AutomationWhat actually got automated in claims, and what still has not
The industry is publishing remarkable numbers about AI in claims. We run a claims platform in production across 28 legal entities, and the honest picture is narrower, more useful, and much less exciting than the press releases.
July 10, 2026 · 7 min read
OperationsNearshoring to Mexico: the systems bill nobody puts in the business case
The model compares labour, freight, tariffs and lead time, and it usually works. What it leaves out is that the new plant will run its first three years on spreadsheets, WhatsApp and a corporate ERP that does not fit.
July 6, 2026 · 6 min read
Systems ModernizationThe weeks after go-live decide the project. Almost nobody staffs them.
Every implementation plan is detailed up to launch day and vague afterwards. That gap is not an oversight in the schedule. It is the period in which a system is actually adopted or quietly abandoned.
July 2, 2026 · 5 min read
Data & AnalyticsAdministrators now use AI more than students. The systems underneath have not caught up.
2026 is the first year university staff out-use students on AI. It is also the year it became obvious that the constraint was never the technology. It is that advising, enrolment, aid and billing still cannot see each other.
June 30, 2026 · 6 min read- AI Automation
Where AI automation pays, and where it quietly does not
The question is never whether a model can do the task. It is what a wrong answer costs, how often you will catch it, and who is sitting in the review queue when you do.
June 24, 2026 · 4 min read