Our thinking on the shift.
Writing on strategy, governed agents, delivery, and how the enterprise is reimagined when agents do part of the work.
AI is the new electricity. Most of it is still lightning.
Spectacular, everywhere, and doing almost no work. What it takes to turn a demonstration into current.
The forty-year gap, and why it doesn't have to repeat.
Electricity was available in the 1880s. Factories were rebuilt around it in the 1920s. The lesson for AI is about the factory, not the power.
The pilot trap.
Why enterprise AI dies between the demo and production, and the three decisions that get it across.
A person holds the switch.
How we design approvals, audit, and autonomy for agents that touch the systems of record.
Why the first agent should be reconciliation.
It is boring, it is daily, it touches real money, and it teaches the organization how to run agents.
Replatforming a legacy platform without a cutover weekend.
What we learned rebuilding a wealth management platform in place, domain by domain, while the business kept running.
What we learned training a public company's engineers on agentic coding.
The tools were everywhere and the standards were nowhere. Training the whole organization, not a pilot team, changed what shipped.
Strategy written for the people who have to deliver it.
Most AI strategies fail the moment they are handed to a team too far from the problem. Here is what we do instead.
Delivery that learns: why Circuit is a graph, not a project manager.
The expensive part of consulting is rediscovering what you already knew. We built the platform to make that a query.
Moving off SAP BPC before the lights go out.
BPC is being retired. The migration to CCH Tagetik that used to take quarters can take half the time when agents already know the platform.
Inside your walls: why we deploy single-tenant, and what that costs us.
Every enterprise AI platform makes a choice about where the data lives. Ours is the expensive one, on purpose.