Mid-market is the missing layer in the Big Four AI rollouts.

I keep three numbers on a sticky note next to my desk.

$3 billion. That is what Accenture committed to its AI rollout in 2024. The number was floor-buyer, not ceiling. By the time you read this, Accenture has shipped a fleet of agents inside its own delivery model and pushed the same playbook outward into F500 transformation programs.

$2 billion. That is KPMG. Same playbook, different brand.

150 agents. That is what EY shipped in tax this year. Each agent does the work of two or three senior associates, and the firm gets to keep the senior associates focused on partner-level judgment instead of spreadsheet plumbing.

And then there is Zora. Deloitte announced Zora as a productized client surface, a packaged AI agent platform F500 buyers can integrate into their existing enterprise stack. Six-figure platform, seven-figure consulting. Built right for the problem set Deloitte already serves.

Five firms. Roughly $8 billion in committed AI rollout dollars, all aimed at the same buyer.

That buyer is the F500. Every dollar of that spend is justified by the price band where F500 transformation work lives. The economics make sense for the firms. The work fits the buyer. None of that is the problem.

The problem is the floor below them.

The mid-market is where the work actually has to land.

I run an operating company. Most of our clients sit between $5M and $200M in annual revenue. They have growth functions running on three to ten people. Their AI line item, if it exists at all, is somewhere between a $25k pilot and a $250k annual budget. Their decision cycle is one founder or one CMO, with one CFO and one board member sanity-checking the math.

None of those clients can hire EY for a 150-agent rollout. None of them can buy Zora. The Deloitte minimum scoping document is longer than the mid-market client’s entire growth-function org chart.

This is not a slight against the Big Four. They are doing the work they were built to do, at the scale they were built to do it. The seam is real and structural. Their cost of sales does not flex below a certain engagement size. Their delivery model does not flex below a certain operational complexity. Both of those constraints push them upward, toward the F500 buyer who can absorb the cost and the complexity.

And below them, the buyers who need AI most are sitting with a problem the Big Four cannot serve profitably.

$8 billion of AI investment from five firms. None of it lands on a $40M ARR SaaS company or a 12-location dental group.

What the mid-market actually needs.

I have spent the last year asking mid-market operators what their AI line item is. The answers cluster into three things, and none of them look like a Big Four rollout.

They need someone who has actually shipped an agentic workflow inside a mid-market growth function, not a partner who is sponsoring one. They need a working surface their team can touch this quarter, not a six-month roadmap deck. They need a senior operator on every call, not a tiered delivery team where the first two hours are a discovery workshop and the third hour is a junior consultant taking notes.

Most importantly, they need the price to fit. A $30M ARR SaaS company cannot pay six figures upfront for an agent platform, no matter how good the product is. The cost of capital at that scale is real. The opportunity cost of paying for a platform you cannot integrate inside two quarters is worse.

Operating companies are built for that economics. One P&L means one priority list. One operator on the engagement means no relay race. Cross-capability work is the default behavior, not a separate sale. The price band that results lands inside what the mid-market can absorb, because the structure does not carry the cost of a partner-and-tiered-team delivery model on top of every workstream.

That is the gap. The mid-market needs operating companies built for the AI moment. The Big Four are doing the F500 work. The seam in between is what the operating companies have to claim.

Why I built IG this way.

Innovative Group was rebuilt around this argument. Six specialty teams under one P&L. One operator on every engagement. Productized client surfaces (Beat and Next Best Action) so the client can see the work in real time. Capital alongside services, because some mid-market clients need the funding more than they need the consulting.

None of that is an accident. The structure exists because the mid-market buyer does not have a Deloitte option and does not want a holding-company option either. The holding-company shape pushes the agency burden of coordination back onto the buyer. The Big Four push the buyer up to a price band they cannot afford. The operating-company shape resolves both pressures inside the same engagement.

I think this is going to be the dominant shape in B2B services for the next decade, at least for mid-market work. The economics are right and the AI moment makes the cross-capability coordination question matter more, not less. A buyer who is figuring out how to deploy AI inside their growth function does not have time to manage four agency vendors. They need one team that can run the strategy, build the system, ship the surface, and connect it to the capital they need to scale the experiment that works.

Operating companies are built for that. Holding companies are not. Big Four firms are not, at this price point.

What the next twelve months look like.

I expect the Big Four AI numbers to keep climbing. EY will ship the next 150-agent rollout. Deloitte will expand Zora into adjacent verticals. The press will cover all of it as a sign that AI consulting has arrived at the F500. That will be true.

I also expect a quieter wave of operating companies underneath. Some will look like IG. Some will look different. The shape that matters is the single P&L, the cross-capability team, and the price band that fits the buyer. The mid-market is going to figure out that this shape exists, and the agencies that did not rebuild will start losing engagements to companies that did.

If you run a mid-market growth function and you have been trying to figure out what your AI line item should look like, the answer probably is not a Big Four engagement. It is also not four separate agencies stitched together. It is one operating team that already has the operators, the products, and the capital lined up. Whether that is IG or someone else, the shape is the thing that matters.

The mid-market is where the work has to land. The Big Four have shipped their AI rollouts and missed that floor. The operating companies are showing up to claim it. The next year is going to be loud underneath the F500 noise, and that is where the real shift is happening.