Stephen Messer is co-founder of Collective[i] and Intelligence.com and has written about the AI economy Artificial common sense at reloadnyc. This column summarizes much of Messer’s recent writings and is related to several others, including “What it means to be AI-First“”The oldest trick in management simply doesn’t work anymore“”The weakest link“”The next computer is alive,” And “The death of privacy. The rise of unbreakable communication.”
Most companies think they have an AI strategy.
They have licenses. They have pilots. You have a chief AI officer, an oversight committee, a vendor roadmap, and a slide deck that says “responsible innovation” in soothing font.
What they don’t have is another company.
Your salespeople are still typing into CRM systems. Your managers still spend half of their weeks collecting information from one team and passing it on to another. Your customers are still waiting while the work goes through the same approval chains. The old workflows remain. The old hierarchy remains. The old software architecture remains intact. AI was simply added.
This is not a transformation. It’s decoration.
I called it “”.AI Blend“: The corporate habit of swapping one technology logo for another while retaining all the underlying assumptions about how work gets done. It feels like progress because it creates activity. There is no benefit.
The companies that avoid this transition will start with a much more difficult question:What work should no longer exist?
Not: How can AI speed up this process by 10%?
Not: Which chatbot should we license?
Not: How many employees use the tool?
What can we delete? Which decisions can bring you closer to the customer? What information no longer needs to be collected, matched, summarized and shared with a number of people before someone takes action?
This is the difference between introducing AI into a company and becoming an AI-first company.
Start with subtraction
The traditional response of a company to a new technology is addition. Add a tool. Add a dashboard. Add a project team. Add a governance layer. Add another system to the stack.
But the first instinct of an AI-first company should be subtraction.
In “The art of subtraction“I argued that companies should question every requirement, remove unnecessary steps, simplify what’s left, and only then automate. This order is important. Automating a bad process doesn’t make it a good process. It makes the bad process faster, harder to detect, and more expensive to run.
Take sales forecasts. For decades, companies have asked individual salespeople to enter forecasts into the CRM, then asked managers to interpret them, then set up meetings where leadership negotiated a number that everyone knew was partly theater. The data is late, incomplete and biased by incentives. The meeting takes place because the system cannot directly observe the purchase process.
The AI-era alternative is not a more elegant forecast meeting. It is a system that analyzes the buyer’s actual behavior, market conditions, timing, relationships and signals throughout the business process. The goal is not to make the old ritual more efficient. It is intended to make the ritual unnecessary.
Therefore The companies that win with AI are playing a different game. You start with a specific business constraint and a measurable outcome. They don’t measure usage. They measure whether the restriction has shifted.
It’s not just software that is at risk
For this reason, the AI discussion isn’t really about software.
Yes, traditional software is vulnerable. Much of the enterprise stack is designed to organize human data entry: applications that store records, route tasks, create reports, and help managers reconstruct what happened after the fact. AI agents will increasingly observe activity, maintain context, initiate work, and recommend or execute the next best action.
But software doesn’t go down alone.
The management structures based on this are also being questioned. In “Software doesn’t go down alone“I argued that AI will pressure the layers created to collect information, translate it across functions, prepare it for meetings, and pass decisions down.”
This does not mean that leadership disappears. This means that the leaders who create value will be different.
The people who matter most are developers: people who understand a real business problem, can solve it using technology, and are close enough to customers and operations to know if the solution works. The people who become less important will be those whose role is to maintain tensions, control access to information, or manage processes that no one today would design from scratch.
In “Find your property developers. Or they leave and start without you“I argued that too many companies have placed their AI future in the hands of people chosen to prevent failure rather than create new capabilities. Governance is important. Security is important. But a company that treats every low-risk experiment like a high-risk autonomous decision will find that its competitors have learned more while it was still approving a pilot.
The The safest move in AI might be the one that makes you irrelevant. For responsible use, it is not necessary to disable every use case. It requires separating the applications that require tight control from those that require learning to begin now.
The actual trench lies above the model
The debate about AI still gets stuck at the model level: who is the best benchmark, who has the most training throughput, and whether a particular frontier company is overvalued?
These questions are important. They are not the most important.
The models will improve. They will also multiply. Open and closed systems will compete, prices will fall, and features that once seemed exclusive will become available to more companies. The lasting advantage will not be having access to a model that everyone else can rent.
It will come from what sits above it.
In “The only fight that counts in AI“I described this battlefield as the orchestration layer: the systems that determine which model takes on a task, maintain context across work, connect intelligence to proprietary data, and learn from the results of real-world decisions.”
Lock-in lives there. Not in a command prompt. Not in an interface. Not in an employee’s passing familiarity with a tool.
The moat is the learning system: a company’s ability to connect proprietary context, trusted relationships, operational data and feedback from the market. For this reason, in “Your buyer has a process“I argued that commercial information must go beyond what a salesperson enters into a CRM. A buyer’s process unfolds through relationships, timing, incentives and signals that no single salesperson can fully discern.
The same applies to human networks. The old warm introduction was valuable because it compressed trust. But it was also opaque and dependent on gatekeepers. In “The warm intro is dead“I explored how verified relationship information can make that trust more visible and usable – when built with the right controls and consent.
This is bigger than the company
AI is often discussed as a labor or technology budget problem. It is neither one nor the other. It is an institutional matter.
The systems that govern housing, infrastructure, energy, capital formation, communications, and privacy were developed in the same pre-AI world as corporate hierarchies: a world where collecting and interpreting information was slow, expensive, and centralized.
That’s why permission is important. In “Time kills all offers“I argued that America’s permitting machinery has become an economic bottleneck. This is not about automating adjudication. This is about eliminating the red tape that makes building a home, opening a business, or investing in infrastructure a drag.
For this reason, the expansion of AI infrastructure also deserves more attention than the usual “bubble versus no bubble” debate. In “The trillion dollar trade that Wall Street doesn’t see“I argued that data centers, power and computing capacity are not simply costs associated with a speculative technology cycle. They are strategic options for the next industrial architecture.”
And that’s why we should resist simplistic narratives. Circular capital flows can lead to surpluses, as I wrote in “The most expensive money in the room.” However, it is possible that the financing structures are unclear and the underlying transition is real. The important question is what survives when financial enthusiasm wanes: infrastructure, capabilities, proprietary intelligence and operational capabilities – or just expensive stories.
The choice before the managers
Every company now faces the same decision.
With the help of AI, the institution of yesterday can be preserved: the same departments, workflows, data silos, approval chains and administrative rituals – just with a more impressive user interface.
Or it can use AI to build the company that should have always existed: one that sees more, learns faster, operates closer to customers and spends less time managing tasks than creating value.
The first path will produce numerous announcements.
Second, the gap between companies that appear to be adopting AI and companies that are actually being reshaped by it will become a larger gap.
The selection window is now open. It will not remain open indefinitely.
The opinions expressed in Fortune.com comments are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Assets.
Suggested Author Disclosure:Stephen Messer is co-founder of Collective[i] and Intelligence.com. The views expressed are his own.
For publication I would also consider adding a linked endnotes module –“Read the relevant columns on Artificial CommonSense”– with the remaining parts, including “What it means to be AI-First“”The oldest trick in management simply doesn’t work anymore“”The weakest link“”The next computer is alive,” And “The death of privacy. The rise of unbreakable communication.”
This story was originally featured on Fortune.com