85% of CEOs Say They Need Help. The AI Industry Wants to Sell Them a Multi-Quarter Transformation.

They don’t need to replace one consultant with another. In 2026 they need a Jarvis.

By Lazaro Fuentes, Founder & CEO, SQOR.ai

---

I was on a train last week listening to my friend Matt Turck interview Timothée Lacroix, the CTO of Mistral AI, on his podcast. Timothée is brilliant. Mistral has done extraordinary things with a fraction of the compute that American labs burn through. But something he said stopped me mid-commute, and I’ve been chewing on it since.

He was describing how Mistral deploys teams of Forward Deployed Engineers (FDEs), embedded technical specialists, into enterprise clients. They connect systems. They map workflows. They fine-tune models against proprietary data. They automate container release processes at shipping ports. They modernize Excel spreadsheets into Python applications. And when he described this work, he used a word that I think reveals more than he intended.

He called it “control.”

His argument: if you’re going to invest the effort to build an AI advantage, you should own what you build. The software stack should be in your hands. Your expertise should stay yours. Don’t rent. Build.

And I thought: Hmm. Maybe pause a second and think about that.

---

Here is what I cannot stop noticing.

Mistral is not alone in this approach. Palantir has been running FDEs for years, embedding teams inside government agencies and corporations to build bespoke analytical systems on top of their platform. Celonis does something similar in process mining, sending consultants and technical teams to map, instrument, and optimize workflows from the inside. These are impressive companies doing impressive work. But if you step back and squint, what they are really doing is something very old wearing something very new.

They are consulting companies.

Not in the pejorative sense. In the structural sense. Their revenue depends on putting human expertise inside your building, learning your problems, and constructing custom solutions that are deeply entangled with your operations. The platform is the Trojan horse. The services are the business. The lock-in is the moat.

And the pitch is always the same: Build. Own. Control.

Which sounds great until you remember that we are in February 2026, the technology is roughly two and a half years old in its current form, and the ground is still moving under everyone’s feet.

---

But here is what really bothers me about this model, and I think it is the thing almost nobody in the industry is willing to say out loud.

It misses the paradigm shift entirely.

The paradigm shift is not about deploying smarter people into your building. It is about making the people already in your building smarter. Those are two fundamentally different propositions. One scales with headcount and billable hours. The other scales with intelligence. And the fact that the most sophisticated AI companies in the world are defaulting to the first model, embedding teams, building custom workflows, monetizing complexity, tells you that they have not fully internalized what their own technology makes possible.

The whole promise of this era is that intelligence becomes a layer, not a project. That a CEO can ask a question and get an answer grounded in their own data without filing a ticket, waiting three weeks for an analyst to assemble a report, or hiring an FDE to build a pipeline. That a CFO can understand what is actually driving margin compression across six business units without commissioning a workstream. That a division president can see what is going to happen next quarter, not just what happened last quarter, without begging for a dashboard update.

That is not an incremental improvement on consulting. That is the end of the reason consulting exists in its current form.

And yet the industry keeps selling the old model with new vocabulary. “Platform plus services” instead of “strategy plus implementation.” “Embedded engineers” instead of “management consultants.” “Workflow automation” instead of “digital transformation.” The wrapper changes. The dependency does not. And every dollar a company spends reinforcing that dependency is a dollar not spent discovering that the dependency itself is becoming optional.

They are not just repackaging consulting. They are blocking the view of what is actually possible, which is that the intelligence layer they are building bespoke for seven figures can be delivered as a capability that works in weeks, for a fraction of the cost and risk. They are selling the cave when the suit already exists.

Same creature. Different name. And the executives writing the checks deserve to know it.

---

Here is what the ground actually looks like for those executives right now.

Global CEO turnover hit a record high in 2025. Russell Reynolds tracked 234 CEO departures across major stock indices worldwide, up 16 percent from the prior year and 21 percent above the eight-year average. In the S&P 500 alone, there were 59 departures. Publicly traded companies saw their most active year of CEO exits on record: 446, according to Challenger, Gray & Christmas. Average CEO tenure has dropped to 7.1 years, a six-year low. Eleven global CEO appointments in 2025 lasted less than a year.

These are not struggling companies. The Conference Board found that CEO successions at firms in the top three performance quartiles jumped from 7 percent in 2024 to 12 percent in 2025. Boards are not just firing underperformers. They are replacing leaders proactively, seeking new skill sets for a world that looks nothing like it did three years ago.

And it is not just CEOs. CFO turnover hit a seven-year high in 2025, with S&P 500 companies hiring a record 106 new finance chiefs. Fortune’s Power Moves column reads like a weekly casualty report. Disney, PayPal, Kroger, Workday, Walmart. New names at the top, week after week, stretching into 2026 with no sign of slowing.

Now layer this on top: the AlixPartners 2026 Disruption Index, based on surveys of 3,200 senior executives across 11 countries, found that 45 percent of CEOs fear losing their jobs due to disruptive forces. Forty percent say they feel more anxious in their roles than last year. Seventy-two percent say it is increasingly difficult to determine which disruptive forces to even prioritize. And here is the one that really lands: 85 percent of CEOs say they need greater professional and personal support.

Eighty-five percent. Nearly nine out of ten people sitting in the most powerful chair in the building are saying, out loud, I need help.

This is the environment in which someone is asking these executives to commit to a multi-quarter, services-heavy, custom-built AI transformation program. To lay it all on the line. To build, not rent. To own the stack.

I have spent most of my career turning around broken companies for private equity investors. I know what it looks like when someone tells a leader under pressure to commit everything to a single vector of attack without adequate reconnaissance. It does not usually end well.

---

Let me say something about the consulting industry, because what is happening there is a mirror of what is happening in AI deployment, and almost nobody is connecting the two.

Accenture eliminated 22,000 positions in 2025 alone, dropping from over 801,000 employees to 779,000 as part of an $865 million restructuring program. Their CEO publicly warned that employees who could not reskill for AI would be “exited.” McKinsey cut 5,000 people over the past two years, dropping from 45,000 to roughly 40,000, with Bloomberg reporting that thousands more cuts are planned over the next 18 to 24 months. PwC laid off over 3,300 in the US across two rounds. Deloitte began cutting its government consulting division after $371 million in federal contracts were canceled or modified. KPMG trimmed its audit workforce. Fast Company called the McKinsey layoffs a “warning signal for consulting in the AI age.”

And then there is Gartner. The company that has been the default advisory voice for CIOs and technology leaders for decades saw its stock drop 71 percent over 52 weeks. A 31 percent single-day crash on February 3rd, 2026, despite beating earnings estimates, because management issued guidance showing revenue contraction for the first time in nearly two decades. The consulting segment’s revenue fell almost 13 percent. One AI author wrote publicly that “Gartner is dying,” arguing that its model of packaging client-sourced information back to those same clients no longer works when AI tools can do the synthesis faster and cheaper.

So the traditional consulting houses are under siege. Their model, sending smart people into your building to tell you what to do, is being compressed by AI from below and by client skepticism from above.

And yet, simultaneously, a new generation of AI companies is emerging with exactly the same structural model, just dressed in different clothes. FDEs instead of management consultants. “Platform plus services” instead of “strategy plus implementation.” “Workflow automation” instead of “digital transformation.”

The industry that is supposed to be delivering the paradigm shift is instead finding new ways to avoid it.

---

And the data tells you exactly how that is working out.

A research team out of MIT published a report in mid-2025 called “The GenAI Divide,” based on interviews with 52 organizations, surveys of 153 senior leaders, and analysis of over 300 public AI implementations. Their headline finding: despite $30 to $40 billion in enterprise Generative AI (GenAI) investment, 95 percent of organizations are seeing zero measurable return. Only 5 percent of custom enterprise AI tools made it from pilot to production. Sixty percent of organizations evaluated these tools. Twenty percent got to pilot. Five percent made it to deployment. The rest died somewhere in between, killed by brittle workflows, poor integration, and systems that could not learn or adapt.

And here is the part that should matter most to every executive reading this: the MIT researchers found that enterprises, the big companies with the biggest budgets, led in pilot volume but had the lowest rates of pilot-to-scale conversion. Mid-market companies moved from pilot to deployment in 90 days. Enterprises took nine months or longer. More money, more resources, worse outcomes.

Now think about what the services-heavy, build-first model requires of a buyer. It requires you to commit significant budget before value is proven. It requires you to connect systems and move data before governance is mature. It requires you to embed external teams before your own people understand what they are building toward. And it requires you to sustain all of this through executive turnover, budget cycles, priority shifts, and the inevitable moment when someone in the C-suite asks: What exactly are we getting for this?

That is not a technology risk. That is a career risk. And in a year when CEO turnover hit an all-time high, when boards are already looking for reasons to make a change, approving a seven-figure AI initiative that produces no measurable P&L impact is not a theoretical danger. It is the kind of thing that gets you replaced by someone who promises to do it differently.

And the Mistral CTO, to his great credit, essentially admitted as much during the interview. He said enterprise AI is still in a “building phase.” He said the tooling is “in its infancy.” He said their workflow product is not yet generally available. He said observability, the ability to see what the system is actually doing, is “an area where we’re still working.” He said enterprises are building “siloed things because we’re scared of data going through walls.”

He is being honest. The infrastructure is not ready for the bet that the sales pitch is asking executives to make.

---

There is a scene in the first Iron Man film where Tony Stark is trapped in a cave with a box of scraps, and he builds the Mark I suit. Crude, heavy, barely functional. Just enough to survive. It gets him out. But nobody would confuse that suit with the Mark L that eventually gives him the power of a god.

The genius of the Iron Man arc is not the final suit. It is the sequence. And critically, the Mark I, the one built in the cave, had no Jarvis. Tony was alone, guessing, improvising, nearly dying in the process. It was only after he survived and built Jarvis into every suit that followed that the iterations became extraordinary. The intelligence layer is what made the difference between desperation and dominance. It made his decisions faster, his awareness broader, his response time shorter. It did not run the company. It made the person running the company superhuman.

That is the analogy I keep coming back to when I think about what enterprises actually need right now.

They do not need a team of external engineers to come in and rebuild their workflow systems from the ground up using AI that is two and a half years old. That is using an F-35 to kill an ant. It is the most expensive, most complex, most risky way to solve a problem that, for most companies, starts with something much simpler.

Most companies have data. They do not have answers.

They have dashboards nobody looks at. Reports that take weeks to assemble. Metrics that three different people define three different ways. Decision cycles that crawl because nobody trusts the numbers. Key-person dependencies where one analyst holds the whole picture and everyone else is guessing.

And the industry knows it. The era of passive Business Intelligence (BI), where data sits on a screen waiting for a human to notice something and then manually jump into another tool to act on it, is ending. The context gap between where data lives and where decisions happen has been the defining dysfunction of enterprise analytics for two decades. Boards are no longer impressed by “productivity gains” from AI. They are demanding direct financial impact: top-line revenue growth and bottom-line profitability. The argument that your AI tool helped someone work a little faster is collapsing as an ROI metric. What matters now is whether the intelligence layer can sense something, explain it, and connect it to action before the window closes.

That is not a workflow automation problem. That is a decision-intelligence problem. And it can be solved without laying everything on the line.

---

I believe 2026 should be the year of the Ironman suit, not the year of the cave.

What I mean by that: give your leaders a Jarvis. Give them an intelligence layer that connects to the systems they already have, surfaces the metrics that actually matter, and helps them make better decisions faster. Without a six-month implementation. Without a team of embedded engineers. Without ripping out and replacing the things that are already working.

Rent first. Cherry-pick. Start where the pain is most obvious, and for most companies, that is the traditional BI layer that everyone knows is broken but nobody has had a clean alternative for.

Prove value in weeks, not quarters. Expand when the evidence says expand. Keep your teams. Make them stronger. Preserve your optionality so that when the technology matures, and it will, you are in a position of strength, not a position of dependency.

If your company is working, the most dangerous thing you can do right now is bet everything on a full-stack AI transformation before the stack itself has stabilized. The most prudent thing you can do is make every executive, every leader, every decision-maker incrementally more powerful. Give them better information, faster. Give them confidence in the numbers. Give them back the hours they spend chasing data so they can spend those hours thinking.

That is not a small ambition. In an environment where 45 percent of CEOs fear for their jobs and 85 percent say they need more support, making leaders meaningfully better at their jobs might be the highest-ROI investment an enterprise can make.

---

This is why we built SQOR.

Not because we think workflow automation is wrong. It is going to be transformational, eventually. But because we believe the sequence matters. And the sequence that protects executives, preserves optionality, and proves value fast is: augment first, automate later.

SQOR is a Decision Intelligence platform. It connects to the data systems you already have. Read-only, no data replication, no migrations, no six-month onboarding. It extracts and normalizes the Key Performance Indicators (KPIs) that drive your business, over 800 of them, automatically. It does not just visualize them. It tells you what is causing them to move, what is likely to happen next, and what you should consider doing about it. It answers questions in plain language. It monitors execution nightly. It alerts you when something drifts. And because every metric is grounded in a unified semantic layer, governed definitions that mean the same thing to every user across every department, the answers it gives are not hallucinations. They are mathematically derived from your own data.

It is not BI. It is not a system of record. It is the decision layer above your stack. The paradigm shift, delivered. Not as a multi-quarter transformation that might collapse under its own weight, but as the intelligence layer your leaders need to make decisions fully informed, fully augmented, and moving at a speed your competitors cannot match.

We built it because I spent years as a turnaround specialist for private equity (PE) and venture capital (VC) portfolio companies, and in every single one of those companies, the problem was the same: smart people, plenty of data, and no reliable way to turn one into the other fast enough to matter. I watched companies spend millions on BI infrastructure and still struggle to answer basic questions about their own business.

And we built it because I believe the right move in 2026 is not to commit everything to a technology that is still finding its footing. The right move is to put on the suit, power up Jarvis, and start making decisions at a level that changes the trajectory of your business.

The 85 percent of CEOs who said they need help were not asking for another consultant, another platform, or another eighteen-month implementation roadmap. They were asking for the thing that has been missing since the data revolution began: the ability to ask a question and trust the answer.

That is what we built. And it is ready now.

---

Lazaro Fuentes is the Founder and CEO of SQOR Technologies Inc. He is a three-time founder, former turnaround specialist for PE and VC portfolio companies, former investment banker on Jamie Dimon’s team at Salomon Brothers and Smith Barney, and a retired U.S. Army Infantry officer (1LT, 10th Mountain Division). He takes trains, listens to podcasts, and occasionally writes things that are longer than they need to be.

To discuss how SQOR can help your organization, visit sqor.ai or reach out directly. Your first 100 queries are always free.

Originally published on LinkedIn.