top of page
ADVISOR. ANALYST. CLIENT. PROVIDER.
THE FOUR SEATS
I have been giving enterprise software advice for 36 years. At the Big 6, to clients. At Gartner, to clients and markets. At the Irvine Company, to my own executives, and then I had to run what I recommended. At MRI Software, to a roadmap, where advice stopped being a recommendation and shipped.
The seats are the point.
Most people advising in this category have held one seat. Having held all four means the advice accounts for what the other side of the table is actually doing, which is usually the part that decides whether something works.
There are four seats in the name. The fifth one is yours.
Recent Enterprise Software and PropTech Essays
What Stays Scarce in a World of Agents
General knowledge is shared, and every agent will have it. If agents share the same general knowledge, what can a person or a company still differentiate on? Six things stay scarce: context, authority, accountability, process, purpose and culture. This piece follows who brings the agent, what travels and who gets paid.
Oct 2716 min read
Will AI Agents Replace Systems of Record? A 2032 Forecast
Picture 2032, with every core headless and employees bringing their own agents. Work backward and the cores keep the system of record and the access, and the contest moves to who holds how the role is performed. This piece tests that scenario against the 2026 evidence, then gives the three outcomes and the six questions to put in a renewal.
Oct 2716 min read
Application Equilibrium
In 2000 I helped write the Gartner note that said ERP would consume its neighbors. It did not happen for CRM. Every domain of enterprise software is built around one core object. It holds until a core forms around an object that sat inside an existing core.
Oct 2711 min read
Nobody Counts Conversations
Every vendor has a pricing metric. Only some have a value metric, one the customer already counts, already controls, and that tracks what the software is worth. Run the test on tokens, conversations, actions, seats and resolutions.
None of them passes. Tokens are like phone minutes, and we know how that went.
Sep 2512 min read
Target AI at the Reviews, Not the Work
Every AI deployment produces wrong answers, and the ones that survive are deployed where wrongness already has somewhere to land.
- A person who was going to read it anyway.
- A domain narrow enough that there is not much to be wrong about.
- A log that catches it later.
- A gate that holds it before it commits.
Find that first, then pick the AI.
Sep 2415 min read
AI Has Changed the Definition of a Connected Building
The call kept dropping because the modem sits in the middle of a basement that is fully below grade, which is the right place for the equipment and the wrong place for a person holding a phone. Scale that up and it decides something larger. Every building holds four populations: the people who work in it, on it, the people it works for, and the first responders who come through it. Enterprise wireless serves the first, the fire code reaches the fourth, and AI is arriving for
Sep 1417 min read
Attention Is Not a Control
Two kinds of rules governed what the machine was allowed to do. Both were written by me, both were in force the whole time, and both were meant to hold it to a standard. One kind was ignored completely. The other was obeyed without exception.
Sep 1410 min read
Your AI Business Case's Blind Spot
I have worked on hundreds of business cases. Every one was written inside a department or a division, and written for the benefit of it. The only one I ever saw that mentioned disaster recovery was a business case about disaster recovery. That was fine for thirty years. It is not fine now.
Sep 1011 min read
AI Capability Doesn't Always Arrive as Features
Taking a new release used to be the feature gate. Now AI capability moves underneath features you already run, and no gate was built to hold it.
Implementation shapes the product to the organization, the organization lives with that shape, and the distance grows with every release.
That gap used to be countable. Part of what accumulates now is not made of items, which is why the deadlines that forced a re-look no longer catch everything.
Sep 311 min read
The Second Disassociation
I spent four hours on a golf course this week and cleaned up a large dataset during them. I did none of that work. I directed it.
In 2020 I sorted which activities would stay separated from their places, and the mechanism held in six of seven.
Here is the same sort, pointed at people instead of rooms.
Aug 1319 min read
Everyone Sells AI the Same Way. You Have to Buy It Differently.
I asked an AI assistant for a foursome and it gave me five players.
Twice that week, two different vendors, the same failure.
The QA profession has been discussing this for two years.
The person who signs the acceptance form has not been in the room.
Nine questions to ask before your name goes on it.
Aug 1315 min read
Your Algorithm Has Decision Rights Nobody Granted It
A leasing agent declines an application in ninety seconds.
Ask who decided and the answer is the system said decline.
Ask everyone above her and every answer is true, and the chain closes into a circle.
Real estate is two years ahead on this. Everyone else bought the same software later.
Aug 1315 min read
The Year I Stopped Writing About AI
In December 2020 I wrote a list of six things that would matter the following year. All six did.
Not one of them was artificial intelligence, and twenty-four months later ChatGPT was released.
We audit what we said. We do not audit what we stopped saying.
Three steps to find out what fell off yours.
Aug 1317 min read
There are four seats in the name. The fifth one is yours.
bottom of page