Zhenya BankouskiAI transformation in B2B companies is uneven. It moves fastest in crowded, low-margin markets and slowest where competitive pressure is low, and inside each company it is carried by a few power users while everyone else works the way they did in 2021. The companies that get results treat AI as a system with an owner, a revenue metric and a maintenance plan, not as a set of personal tools.
Here's the uncomfortable part: most companies aren't transforming. A few people inside them are.
At Revenue Wizards we work on the commercial side of B2B companies (sales, marketing, customer success, the systems underneath). Over the last year, AI has crept into almost every client conversation. Sometimes it's the whole agenda. Sometimes it's the thing the CEO brings up in the last five minutes of the call.
And the picture we see is not the one in the vendor decks. It's patchy. Uneven. Expensive in ways nobody is measuring yet. Our own RevOps in 2026 survey of 123 practitioners at mid-market B2B SaaS companies found the same gap: 73% have an AI mandate, and only 38% have the unified data to support it.
So here's what we actually observe, where the research backs it up, and the questions we now ask before we build a single agent.
AI transformation in B2B in six numbers
| Number | What it measures | Source |
|---|---|---|
| About 5% | GenAI pilots that drove rapid revenue acceleration | MIT NANDA, 2025 |
| 78% | AI users who bring their own AI tools to work | Microsoft and LinkedIn, 2024 |
| 52% | AI users reluctant to admit using it for their most important tasks | Microsoft and LinkedIn, 2024 |
| About 6% | Companies that qualify as AI high performers | McKinsey, 2025 |
| 16% to 43% | Logistics companies taking an active approach to AI, in one year | Annual Study of Logistics and Transportation Trends, 2026 |
| About 205% | Year over year growth in GTM engineer job postings, 2024 to 2025 | Onfire |
Each number is linked to its source in the section where we use it.
AI adoption depends on how hard your market squeezes you
AI transformation is not the same across our client base. Not even close.
Companies in sectors where competition is fierce and margins are thin are the better adopters. They have to be. If your competitor ships faster outbound, sharper targeting or a cheaper support motion, you feel it in the pipeline within a quarter. Being on the edge of technology isn't a nice to have for them, it's survival.
Then there are industries like logistics and manufacturing. Again, I'm talking about the commercial side, not the factory floor. Adoption there is slower, and as far as I can tell it comes down to two things.
- Less pressure. Fewer new entrants, longer customer relationships, slower sales cycles. Nobody is getting disrupted by Friday.
- Lower talent mobility. People stay in roles longer. Fewer new hires walk in with fresh habits from an AI native company.
The research says these industries aren't ignoring AI, they're stuck between trying and scaling. The 2026 Annual Study of Logistics and Transportation Trends found the share of logistics companies taking an active approach to AI jumped from 16% to 43% in a year, but only 7% strongly trust what it recommends (Supply Chain 24/7). In manufacturing, a Parsec survey puts adoption at 72%, with only 10% scaled across their networks (Parsec).
Lots of pilots. Very little rollout. That matches what we see.
AI power users and everyone else inside one company
Zoom in on a single client and the same unevenness shows up again.
There are champions. People who use AI every day and have quietly automated a big chunk of their own workload. And there are people doing the job exactly the way they did it in 2021.
This is the era of the highly enabled individual contributor. One curious account executive with good prompts and a couple of automations can now do the research, prep and follow up that used to take a small team.
What separates them isn't seniority or technical skill. It's curiosity. The people who keep asking "how could this part of my week be faster?" are the ones capturing almost all of the benefit, and burning most of the tokens the company pays for.
A lot of that happens outside official programs. Microsoft and LinkedIn's 2024 Work Trend Index found that 78% of AI users bring their own AI tools to work (Fortune).
Why AI champions keep their knowledge to themselves
This is the single biggest challenge we see at clients.
Knowledge doesn't cross from the champions to everyone else. The person who figured out how to cut call prep from an hour to ten minutes often keeps it to themselves. They don't run a lunch session. They don't write it down.
I don't think it's malice. It's job security. In the same Work Trend Index, 52% of AI users said they were reluctant to admit using it for their most important tasks, and 53% worried it would make them look replaceable. Only 39% had received AI training from their employer (Virtualization Review).
So you end up with a company where AI capability lives in a handful of personal accounts. When one of those people leaves, the capability walks out with them.
Why AI stays a set of personal hacks instead of a system
Even with all the enablement, workshops and licenses, people still struggle to connect the dots.
Picture a typical sales org (an illustration, not a specific client). One rep has a great prompt for call summaries. Someone in marketing runs an enrichment flow. A manager built a custom GPT for proposals. All three are useful. None of them talk to each other, and none of them reliably write back to the CRM.
That's not transformation. That's a collection of personal productivity hacks.
Thinking in systems means asking where data comes in, what happens to it, where it lands and who acts on it. Then deciding where AI belongs in that chain. Most teams haven't been taught to look at their work that way, and AI makes the gap more obvious, not less. Our AI for sales and RevOps playbook shows what that looks like workflow by workflow.
Why nobody is measuring the cost of AI yet
Here's something that surprised me. Right now, almost nobody we work with is worried about what AI costs.
Budgets aren't the problem. If anything, everyone is trying to spend as much as possible on AI. We're in build mode, with experiments, trials, pilots and more pilots. The hope is that efficiency and productivity show up later, and that optimization becomes a problem for future us.
The data says the hope is doing a lot of heavy lifting. MIT NANDA's 2025 report found that only about 5% of generative AI pilots drove rapid revenue acceleration, and that more than half of GenAI budgets go to sales and marketing even though the biggest returns showed up in back office automation (Fortune). It was an early study with a contested sample, so I treat the exact number as directional. The direction is hard to argue with.
My bet is that the optimization phase is coming, and it will hit the companies that built the most without measuring anything. That's my inference, not a stat. When it comes, the way your vendors charge for AI matters as much as what you build, which is why we wrote up how AI pricing models and credits work.
Who maintains the AI agents after the project ends?
This is the part we think about most when we build AI into a client's revenue engine.
AI is still a system. Agents, orchestrators, prompts, knowledge bases, integrations. Someone has to inherit all of it when the project ends. And in most clients, nobody on the team can.
The harder issue is that there are no established patterns in the industry yet. Every team relies on different tools, different knowledge bases, different ways of storing data and context. There is no standard way to build, maintain, improve and run these systems. That means every AI setup is a bit of a snowflake, and snowflakes are expensive to maintain.
That maintenance cost has to be priced in from day one. Not after the first thing breaks.
The companies that do get results approach AI seriously, with proper development and deployment muscle behind it. McKinsey's 2025 State of AI puts AI high performers (5% or more of EBIT attributed to AI) at about 6% of respondents, and names workflow redesign as one of the clearest things that set them apart (McKinsey). Redesign, not bolting a chatbot onto the old process.
Why leadership is behind on AI transformation
Sad to say, but leadership in many startups is still behind. Not everywhere, and it varies a lot across the cohorts of companies we work with. But the pattern is real.
Some leaders don't fully understand what AI can do for their company. Others get it in principle but don't have the time to go deep into their own processes and see where it fits. Either way, AI ends up delegated to whoever is most enthusiastic, which brings us right back to the champions problem.
There's still a long way to go before AI is embedded in the workflows of most companies, let alone in their philosophy.
Why RevOps should own AI transformation on the commercial side
One of the biggest blockers to AI adoption is talent. Because there's no standard way to deploy AI yet, a lot of the work still has to be stitched together by hand. That's why demand for GTM engineers has exploded. One analysis of more than 1,000 listings found postings up about 205% year over year from 2024 to 2025 (Onfire). Most of that data comes from vendors in the space, so read it as a direction, not a census.
The catch with the GTM engineer approach is simple. You can connect every tool, build every agent and orchestrate all of them. You still need someone to guarantee that the whole thing adds up to revenue. New revenue streams, or better ones from the streams you already have.
And remember, optimization isn't on anyone's mind yet. So a lot of money goes into building without anyone checking how it shows up in the number.
That's why I think revenue operations is best positioned to orchestrate AI on the commercial side, with GTM engineering as the build function underneath. RevOps already owns the process, the data model and the metrics. It's the natural place to decide which agents should exist at all. We compared the two roles in detail in GTM engineer vs revenue operations.
The real constraint is people who can do both halves. They need to understand the landscape of AI tools, and speak the business language of the organization well enough to translate it into systems and agents. Put simply, most people still don't know what's feasible, or how their day to day could actually change.
Seven questions to ask before building an AI agent
This is the checklist we run at clients now. Steal it.
- Which revenue metric does this move? If nobody can name one, it's a toy. A fun toy, maybe.
- Who inherits it? A named person on the client side who can explain how it works without us in the room.
- Where does the knowledge live? One documented place, not five personal accounts.
- Are we redesigning the workflow or decorating it? AI on top of a broken process gives you a faster broken process.
- Who are the champions, and what would make them share? Recognition, time in the week, a role in the rollout. Fear of being replaceable doesn't go away by itself.
- What does it cost per month at full usage? Even if nobody is asking today. Someone will.
- Can leadership defend it in a budget review? If the CEO can't explain why it exists, it won't survive the first cost cut.
Where to start your AI transformation this week
Don't start with more agents. Start with a map.
Find out who in your commercial team is actually using AI, for what, and where that knowledge sits. You'll probably find two or three champions carrying most of it, and a pile of disconnected experiments around them. Then pick one revenue workflow (lead routing, call prep, renewals) and redesign it end to end, with an owner, a metric and a cost.
The tokens are cheap right now. The system you'll have to maintain isn't.
If you want a second pair of eyes on that map, book a call with us. Mapping where AI fits in a revenue engine, and who owns it afterwards, is the work we do at Revenue Wizards.
Frequently asked questions about AI transformation
What is AI transformation in a B2B company?
AI transformation is the shift from individual people using AI tools to AI being built into how the company runs its workflows, with owners, metrics and maintenance. On the commercial side it means redesigning sales, marketing and customer success processes around AI, not adding a chatbot to the old process.
Why is AI adoption so uneven between companies?
Market pressure drives it. Companies in crowded, low-margin markets have to move fast, so they adopt AI earlier and harder. Industries with less competitive pressure or lower talent mobility, like logistics and manufacturing on the commercial side, move slower. Many of them are piloting AI, but far fewer have scaled it beyond experiments.
Why do most AI pilots fail to show business results?
Usually it is an implementation problem, not a model problem. MIT NANDA's 2025 research found only about 5% of generative AI pilots produced rapid revenue acceleration. Common causes are brittle workflows, no clear revenue metric, knowledge scattered across tools, and nobody inside the company able to maintain what was built.
What is the difference between a GTM engineer and RevOps in AI projects?
A GTM engineer stitches tools together, including enrichment, agents, automations and orchestration. RevOps owns the revenue system those pieces plug into, including the process, the data model and the metrics. GTM engineers build the parts, and RevOps decides which parts should exist and makes sure they add up to pipeline and revenue.
How do you make AI agents maintainable after the consultant leaves?
Plan for the handover before you build. Name an internal owner, keep knowledge and prompts in one documented place, tie every agent to a revenue metric, and estimate the monthly running cost at full usage. If nobody on the team can explain how an agent works, it will break quietly within months.
