Every AI tool at my last gigs got adopted the same way: Someone signed up, submitted the receipt, and got reimbursed. No department had to approve it. No one explained why it mattered or what it was supposed to accomplish. The door was just open.
In addition, the tools we already had — Zendesk, HubSpot — were shipping new AI features in almost every release, whether anyone asked for them or not. Nobody was tracking which showed up, what they were supposed to do, whether anyone was actually using them, or whether one tool’s AI was quietly working against another’s.
My team split three ways in response. Some people picked up every available tool. Some flatly refused: I like doing my job the way I’ve done it. Some sat in the middle, not against it, just unsure why they’d bother.
Nobody kept a list of who was in which camp, or what they’d actually adopted. I found out our team had wildly uneven access to basic things almost by accident. Some people could see HubSpot data, some couldn’t. Some had a whole library of AI skills built for their role, and most had never heard it existed. Zendesk had built-in AI tools sitting right there for case deflection and support automation. We weren’t using them. When I asked why, nobody had an answer, because nobody had ever been asked to look.
Without a real plan for how AI gets adopted, which workflows it touches, who’s using it and on what data, and how the pieces are supposed to work together instead of against each other, nobody could have told me, with confidence, everything that was actually running, let alone which were adding real value. The missing strategy was the actual problem, full stop. Not knowing what was live was just the symptom that finally made it visible.
The List That Doesn’t Exist
The AI Revenue Risk Assessment opens with this question: Can you name every AI tool making decisions about your customers right now? Not the big platform contracts. Every tool, every plugin, every AI feature somebody quietly switched on inside a tool for which you already pay.
Most leaders can’t answer that in full, and it’s not because they’re careless. It’s because nobody ever made it one person’s job to keep the list. Tools get added the way mine did. An individual signs up, uses it, and the record of that decision lives nowhere except in that person’s head, and maybe an expense report nobody’s reading for this purpose. You can’t govern, measure, or hold anyone accountable for a tool you can’t name in the first place.
Why the Gap Hides So Well
The invisibility isn’t an accident. It’s the natural output of the same two channels I saw firsthand: individuals signing up for whatever helps them personally, and vendors quietly shipping AI into tools you already own. Every one of those paths adds a tool. None of them updates a list anywhere.
The gap also doesn’t sit still. It gets wider every quarter, not smaller. Each new platform renewal adds a few more AI features nobody explicitly enabled. Each new hire brings whatever tools worked at their last job. Nobody’s job description says maintain the list, so nobody does, and the list gets less complete the longer a company waits to build one.
This is exactly the blind spot the AI Revenue Risk Assessment is built to catch first, before any of the other four — not whether your AI works, but whether anyone can even see all of it: see how it works →
What You Can’t Name, You Can’t Prove
You can’t fix a gap nobody sees, because as far as anyone’s day-to-day job is concerned, the gap doesn’t exist. It only shows up when someone asks the inventory question directly, and almost nobody does.
Every question the rest of this series asks assumes you can answer this one first. You can’t measure a baseline for a tool nobody knows is running. You can’t check whether sales, marketing, and CS are sending a customer the same signal if nobody can see the full picture each team is using to make decisions. You can’t hold anyone accountable for a tool that only shows up on one person’s expense report.
The ROI conversation doesn’t start with a number. It starts with a list, and most companies don’t have one.
You can’t govern what you can’t name.
This week, try the test that exposes the gap. Ask everyone on your sales, marketing, and CS teams to write down, independently and without comparing notes, every AI tool they personally use to do their job and why. Then put the three lists side by side.
If they don’t match, and they won’t, you’ve just found the starting point for every other problem this series is going to name.
Post 3 in this series: What happens when sales, marketing, and CS are each scoring the same customer off a different dataset, and why that alone makes any ROI number unreliable.
This is post two in an eight-part series on why most B2B SaaS companies can’t prove their AI is working. If you want to see where your own organization stands, the AI Revenue Risk Assessment gives you the diagnostic, starting with this exact question: take the assessment →
Andrea Mulligan is a B2B SaaS executive and advisor with 30 years of experience building Customer Success, Professional Services, and GTM organizations. She works with PE-backed and growth-stage companies on CS transformation, revenue retention strategy, and post-sale model design. Start a conversation →