Nearly all go-to-market teams have put AI agents to work on customer and prospect records, but many cannot say how many agents are operating or what those agents are doing, according to a new survey from LeanData. The company's «2026 State of AI Go-to-Market Readiness Report» found that 93% of GTM teams have deployed at least one AI agent, yet close to one-third of respondents could not specify how many agents were taking actions on their records. Thirty percent said they had discovered actions taken without an audit trail.
The findings point to a widening gap between automation and oversight in marketing and revenue operations. LeanData surveyed 157 B2B practitioners across revenue operations, marketing operations, sales, marketing, IT, and related functions in May 2026. The sample skewed toward operations roles, with 36% of respondents in RevOps and 13% in MOps. Most organizations have moved beyond early experimentation: 79% said they were deploying their first agent use cases or already scaling agents across go-to-market. Only 8% described their AI operations as fully optimized.
One reason for the confusion is that AI agents are drawing on the same customer data, workflows, routing rules, and systems that operations teams have struggled with for years. Data quality and AI readiness were the top AI transformation challenges, cited by 55% of respondents. Seventy percent said data hygiene has degraded GTM execution. As a result, 27% have seen multiple tools or agents contact the same prospect, and 17% have seen marketing sequences fire while a sales representative was already working a deal. Bad data was also the top reason cited for stalling AI initiatives, at 45%, followed by undocumented processes at 37% and siloed teams at 32%.
Agents are arriving from many directions. Sixty-nine percent of respondents use AI features embedded in GTM tools such as Gong, Outreach, or HubSpot. Sixty-two percent use custom applications built on LLM APIs, while 46% use agent platforms such as Agentforce, Copilot, or Gemini Enterprise. The most commonly reported number of agents in use was three or four, but nearly one-third of respondents could not provide a number at all. That creates room for collisions: a prospect might be enriched by one system, scored by another, enrolled in an automated sequence by a third, and assigned to a representative by a fourth. Each system can work as designed and still produce a poor customer experience if it operates with different data, timing, or rules.
When asked what they wanted from technology coordinating GTM activity, the top choice, at 31%, was a complete audit trail of every action taken on every record across systems. Another 21% prioritized making agents follow the same rules as human teams. Ownership of those rules remains unsettled. Forty-two percent of respondents said a cross-functional committee owns GTM AI strategy, while 19% said nobody owns it and AI remains ad hoc. RevOps was the designated owner at 18% of organizations. The people most likely to handle the operational work are already stretched: 66% of GTM operations teams said they either had more work than they could handle or could keep up with daily operations but had no capacity for strategic projects. Only 8% said they had enough staff for both.
The report suggests that before adding more agents, teams need to know which ones can already change a customer or prospect record, what data they use, and which actions they are allowed to take. That work falls squarely in familiar marketing operations territory: customer data, business rules, integrations, and process documentation. Without that foundation, the same data problems that have long plagued GTM execution can be repeated automatically, at machine speed, across every system an agent touches.