Insights
From the field
Real observations from building autonomous organizations. No filler. No generic AI predictions. Lessons from production deployments across financial services, retail, healthcare, and more.
Featured
Why Most Enterprise AI Pilots Fail
The pattern is always the same: a large enterprise buys an AI platform, spins up a pilot, and six months later has nothing to show for it. After running dozens of these engagements across banking, government, and healthcare, I can tell you exactly where it goes wrong.
Strategy · 7 min read · 2026-03-28
Strategy
8 min read
The 80/20 of Business Function Automation
Not all business functions are created equal when it comes to automation. After deploying AI across accounting, operations, customer service, and retail, we have a clear picture of what automates well, what partially automates, and what you should not even try.
Engineering
8 min read
Designing Around Legacy Systems
Every enterprise AI project eventually hits the same wall: a legacy ERP or accounting system that cannot be replaced. After years of integrating with ACCPAC, Business Central, Xero, and QuickBooks, we have developed a playbook for working around systems you cannot change.
Engineering
7 min read
Exception-First Design
Most AI systems are designed for the happy path and handle exceptions as an afterthought. We do it backwards: design the exception path first, then automate the rest. This approach has saved us from catastrophic failures more times than I can count.
Operations
7 min read
The Consulting Firm's Dilemma
Accounting and corporate services firms are the best candidates for AI automation, and also the most resistant to it. Here is why they are perfect, and how to get them past the fear.
Engineering
7 min read
Multi-Jurisdiction Compliance Is an AI Problem
Managing compliance across SFRS, HKFRS, AASB, and other standards is a nightmare for human teams. It turns out AI is not just adequate at this — it is better. Here is what we learned deploying multi-jurisdiction AI accounting.
Strategy
6 min read
Why Your AI Strategy Needs a Kill List
Every company has a list of AI projects they want to build. Very few have a list of AI projects they have decided not to build. The kill list is more important than the build list.
Operations
8 min read
Change Management Is the Hard Part
We deployed AI automation in a national retail chain with hundreds of stores. The technology worked. Getting department heads to actually use it was a different battle entirely. Here is what we learned about the human side of AI deployment.
Leadership
8 min read
From 500 People to 50
Everyone talks about autonomous organizations. Few people have thought seriously about what they actually look like: how they are structured, what the remaining humans do, and what breaks when you remove 90% of the headcount.
Operations
7 min read
The Meeting Intelligence Paradox
We built a meeting intelligence platform called Juliette and processed over 1,500 meeting transcripts. What the data revealed about how organizations actually work was far more interesting than the meeting summaries themselves.
Engineering
7 min read
Computer Vision Beyond Defect Detection
When people hear computer vision, they think factory inspection. Our most interesting deployments have been in conservation, education, and retail — places where cameras already exist but nobody is watching.
Leadership
8 min read
Building AI Products While Running a Consulting Firm
The tension between custom consulting work and scalable product development nearly broke us. Here is how we manage it now, and why the consulting work makes our products better — even though it constantly threatens to consume all available resources.