AI in Operations Is Exposing Work That Never Got Done

Key Insights

• AI targets are reopening old process work. Clients promised big back-office gains often haven't documented their processes.
• In plants, basics still beat technology. A $70,000 parallel line doubled the output of a $500,000 automated one.
• Firms forecast autonomous operations. Their engagements are mostly fixing data, processes, and lean fundamentals first.

---

AI companies are telling operations leaders they can get 40% to 80% improvement in back-office operations. One practice leader told us he hears that range at a number of large clients. Then the work starts, and the client doesn't have its back-office processes written down.

That gap between what AI is supposed to deliver and what a client's operations can support came up again and again in the firm submissions, practice leader conversations, and market research behind Management Consulted's 2027 Top Operations Consulting Firms ranking.

The ranking is global and covers operations consulting as the work that changes how a business buys, makes, moves, and delivers what it sells. This article covers AI applied to that work – back offices, plants, and supply chains. Firms that sell AI strategy and implementation as their main offer are ranked separately in our Top AI Consulting Firms list. The pattern below held in both back-office and plant work, though it looks different in each.

In operations, AI is creating consulting work before it replaces any. Most of that work is the process, data, and lean basics that clients skipped.

AI Targets Are Sending Clients Back to Unfinished Process Work

The leader who hears the 40-80% promise runs an operations practice that does a lot of shared services and operating model work. He didn't expect that. "I thought the majority of that work was done in the '80s and '90s," he said. "There's a tremendous amount of work being done in that space now because of AI."

A company can't automate a process it hasn't mapped, and many haven't. Clients "don't have the back office operations documented," he said. "They haven't thought through what does it mean to be AI forward and how do you build a foundation for doing that."

So the engagement becomes the foundation. "It's the operating model, it's the communication, it's all the exceptions," he said. The goal is to take a client "from an ambition that's highly aggressive to a reality that's bold but practical."

He was blunt about where the market stands. "Everybody's early stages on it," he said, and anyone claiming otherwise doesn't have a client doing it. The technology that was supposed to make knowledge move "closer, faster, better" hasn't arrived yet. "It's on the horizon. But it's not there yet, which is why people need help."

In Plants, the Basics Still Beat the Technology

On the manufacturing side, the evidence came with price tags.

A large industrial manufacturer spent about half a million dollars modernizing a production line with screens, digital tools, and robots. A firm came in and built a parallel line for about $70,000. It doubled the output. The firm's leader said they were careful with the messaging, because the client had already made the bigger investment.

That firm has a rule for its smart factory work: it won't take a client there until the client has gotten "the maximum out of your basics." Lean comes first. Digital tools come second, and the firm decides which one a client needs based on where its operations actually are.

A leader at another firm described the same order. "The thing that they need to get from where they are to an incremental better place is not AI," he said. "That can get you from incremental better place to perfect."

His firm recently matched the gains a client expected from €1–2 million of planned capital projects in 11 weeks, with no capital spending. Its digital team now works alongside the consultants during the two-week site assessment, so clients can see an operational fix clearly enough to choose it over new equipment.

When AI work does start, the first step often isn't AI. One leader said his team expects the plant's data to be in place and often finds it isn't. "Then our first use case becomes more before AI, let's set up the digital infrastructure." That matches what Gartner hears from data leaders: 63% of organizations lack the right data practices for AI or aren't sure they have them, and Gartner expects companies to abandon 60% of AI projects that lack AI-ready data.

The same leader estimated that 80% of his firm's operations work still comes from "the very bare basics" – how to design a factory, how to set up lean processes. That's the core of manufacturing consulting, and AI hasn't changed it much. AI has sped up some operations work but not all of it. He said procurement work like building a supplier RFQ now moves much faster, and planning decisions that used to take a long S&OP meeting can now be made "in a minute." Factory design and manufacturing improvement haven't sped up.

Most AI Pressure Starts at the Top

Leaders described the push for AI as coming from boards and executives, not from the people running the operations. "The AI hype is coming more top-down than it is bottoms-up," one said. Plant and functional teams respond by pointing to AI being used "in here, and then there."

Stanford's AI Index reported that 78% of organizations used AI in 2024, up from 55% a year earlier. Use is not the same as results. Gartner predicts that more than 40% of agentic AI projects will be canceled by the end of 2027, citing rising costs, unclear business value, and weak risk controls.

The firms we heard from handle this by starting with the decision, not the tool. "Don't let the tool drive you," one leader tells client audiences. "You figure out what you need and then see what can give you that." He calls it decision-driven AI: "What decision do you need, and what can AI give you to get that decision?"

One submission put it this way: the winners "won't be the organisations with the most pilots. They'll be the ones that redesign their operating models around AI-enabled decision making." Asked what clients should be asking but rarely do, one leader answered: how to prioritize AI and automation versus efficiency.

The New Competitor Sells a Platform, Not a Team

Operations firms used to compete with other operations firms. Now they also compete with software.

One leader described a single large client where his firm works alongside other consulting firms and AI platform companies that bring their own engineers, each in a different workstream. He used to think of those companies as good investments. "Now I think of them as, oh my God, they're coming after consulting."

A leader at a specialist firm described the same shift in sales meetings. The competition used to be other consultancies. Now it can be a company that will "sell you a tool and promise you that you can get results." In some cases, he said, "it's platforms and apps."

His worry is what happens after a client buys the platform. Companies will invest heavily in AI systems, then hand the rollout to internal teams that aren't fluent in them, as a substitute for bringing in outside help for a few months. "Six months, one year, two years will pass and they won't be where they want to be," he said. Going back to an outside partner after that is hard for the executives who made the call. He sees the bigger risk for strategy and advisory firms, since a platform can't stand on a production line and run a study.

Firms are still working out how to talk about their own AI use. About a year ago, a senior client asked one of that firm's partners whether the firm uses AI, and the partner didn't know what the right answer was. Yes sounds modern. It also suggests the client could get the same thing elsewhere. The firm has since set up an AI task force and uses AI mainly for internal efficiency. Some clients have settled the question for it: one current client's confidentiality agreement explicitly bans the use of AI on its data.

Firms Forecast Autonomous Operations While Clients Are Still on the Basics

What firms wrote in their submissions and what they described in conversation pointed in different directions.

Asked where client demand will grow most over the next 12 to 24 months, firms named AI-enabled planning and autonomous operations more often than anything else. Estimates of how much of their new work already includes AI as a meaningful component ranged from under a quarter to three quarters. The firms that spend the most time on the plant floor tended to give the lowest estimates.

In conversation, the same leaders described documenting processes, building data infrastructure, getting lean basics right, and helping teams change how they work. That's the work clients are buying now. Autonomous operations is the next step, and most clients can't take it yet.

That matters because platform companies are making the autonomous operations pitch too, often at a lower price. A firm that leads with the same pitch competes on the platform's terms. The work platforms can't do is the foundation underneath it, and that's where operations firms are winning engagements. One leader named the hardest person to hire right now as an operations expert with AI know-how: someone who can do both.

This doesn't conflict with the ranking's market map, which gives digital manufacturing and Industry 4.0 its own segment. That segment rewards firms that can scale digital tools across many plants, which works only once those plants run well on the basics.

What This Means for Firm Leaders

Sell the foundation as its own engagement. Process documentation, data infrastructure, and lean fundamentals are the first phase of AI work, not a delay before it. Scope and price them that way.

Start every AI conversation with the decision. Ask what decision the client needs to make faster or better, and what data that takes. Then pick the tool.

Decide your answer to "do you use AI?" Know where your firm uses it, where it won't, and how it protects client data. Some clients now write the answer into the contract.

Compete with platforms on what they can't do. Working on site, changing how people work, and handling the exceptions a standard flow misses. Show what happened at clients that installed a tool without that work.

Tie AI claims to the segment you lead. The ranking splits operations into 10 segments, and AI looks different in each. A specific claim about AI in procurement, plant performance, or shared services is easier to believe than a general claim about AI in operations.

The Bottom Line

AI is driving real demand for operations consulting, but not the kind most firms advertise. Clients are paying to document processes, fix data, and get the basics right so AI has something to work on. The firms that win the next few years will be the ones that sell that foundation plainly, instead of the autonomous operations pitch a platform can make for less.

FAQ: AI in Operations

Related reading:

• Cost Reduction Consulting: Why Clients Still Buy Cost First
• Gain Share in Consulting: When It Works and When It Doesn’t