We surveyed 123 RevOps practitioners — Managers, Directors, VPs, freelancers — at AI-engaged B2B SaaS companies. Boards are mandating it. Vendors are selling it. The plumbing underneath says something different. And this is the leading edge — the broader market is almost certainly worse.
The boardroom narrative — "we're an AI-first GTM company" — is running ahead of operational reality. Three of every four respondents have a top-down mandate. Less than two of every five have the data foundation to make it work.
AI touches CRM data, lead routing, customer comms, pricing and pipeline forecasts. Ask 100 RevOps leaders who owns the governance for it and a fifth say "nobody". RevOps owns it in just 19% of companies — even though it sits on top of the systems RevOps already runs.
The "wait for the vendor" era is over. 61% of teams are building custom AI tools internally — point solutions, custom CPQ, deal desks, dedup agents, deal coaches. Salesforce-native, n8n-wired, Clay-fed, Claude-reasoned.
AI was supposed to free RevOps to hire more — to do more, faster, with the same people. The data says the opposite. Only 9% of teams are hiring more because of AI. 37% are hiring less. The mandate makes it worse, not better.
78% rate AI 4 or 5 out of 5 for day-to-day importance. Then you look at the actual usage metrics. The outward edge — customer calls, outbound emails — is more AI-touched than the inward plumbing. Internal calls, CRM hygiene: still mostly manual.
The biggest single execution gap in the dataset isn't CRM hygiene. It's a question of who teams are willing to record. External customer calls: 87% median transcription. Internal calls — pipeline reviews, deal coaching, forecast meetings, the conversations where actual decisions get made: 38%. A 49-point gap, on the same hardware, with the same vendors available.
Forecast accuracy lives in internal calls, not customer calls. Reps tell customers the optimistic story; they tell their managers a slightly less optimistic story; they tell themselves the real story. If you only transcribe the first conversation, your AI is being trained on the most polished, least predictive version of reality.
The teams that have closed this gap aren't necessarily more sophisticated — they've just made the political decision earlier. If you're a RevOps leader sitting on a customer-call library and no internal one, that's your 2026 plumbing project.
Two clear stories. Claude has taken pole position as the paid LLM of choice — though this sample carries Anthropic bias. And in automation, n8n has closed the gap with Clay, with Zapier slipping into third place. The code-light Zapier era is fading.
Clay vs n8n is the underrated story. Clay sits at 54%, n8n at 46%, Zapier at 41%. n8n is a self-hosted, open-source, low-code automation platform — the fact that it's almost on par with Clay suggests RevOps teams are increasingly comfortable owning their own automation infrastructure. Two years ago this question would have been Zapier dominant. The maturity curve has shifted hard.
Transparency note: every survey carries sample bias. This one was distributed through Revenue Wizards' network and Substack — meaning respondents are practitioners who care enough to spend 10 minutes on a 30-question survey about RevOps and AI. That self-selects toward mid-market B2B SaaS companies that are actively trying to be good at this. If the leading edge looks like the numbers in this report, the median B2B SaaS company almost certainly looks worse.
Five moves the data points to. Opinionated. Borrow what helps, ignore what doesn't.