Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Some go-to-market teams are deploying AI agents without being able to say how many are acting on customer records or reconstruct what they changed. A 2026 LeanData survey of 157 B2B revenue, marketing, and sales operations leaders found widespread production use alongside gaps in agent visibility and audit trails. That points to an operating-control problem—not proof that every GTM organization has uncontrolled agent sprawl.
What the survey says about agent visibility
LeanData’s 2026 report says 93% of its 157 surveyed B2B revenue, marketing, and sales operations leaders had at least one agent in production, while only 31% believed their infrastructure was ready. A separate article reporting details from the same survey says about one-third of respondents could not say how many agents had touched their records, and 30% found actions taken without an audit trail. The article identifies the survey fieldwork as May 2026. These are findings from LeanData’s survey, not a census of all GTM organizations.
For context, Gartner forecast in 2026 that the average global Fortune 500 enterprise would use more than 150,000 agents by 2028, up from fewer than 15 in 2025. That is a forecast, not a current observed count. Gartner also reported that 13% of organizations thought they had the right AI-agent governance in place. Neither figure measures GTM teams specifically.
Why agents slip past a single inventory
Agents can reach GTM workflows through several routes, not just a centrally approved platform. In the LeanData survey reporting, 69% used AI features built into GTM tools, 62% used custom applications built with LLM APIs, and 46% used agent platforms. Respondents could use more than one route, so these percentages overlap.
#1 Best Overall
An inventory limited to officially named agent platforms can therefore miss agents embedded in sales or marketing software, as well as custom applications connected to APIs. Unapproved tools may be harder still to find. The practical unit to track is not only the product called an “agent,” but any automated capability that can read or change customer records, trigger outreach, or influence a handoff.
How weak data and workflows create GTM coordination failures
An agent does not repair ambiguous data definitions, broken integrations, or undocumented handoffs merely by operating faster. LeanData’s survey found that 55% of respondents named data quality as a top AI challenge; 70% said data hygiene degraded execution. The report also says 45% cited bad data as a reason AI initiatives stall, 37% cited undocumented processes, and 32% cited siloed teams. These are respondent-reported challenges in LeanData’s sample.
The consequences can be visible to prospects: multiple tools or agents may contact the same person, or a marketing sequence may trigger while a sales representative is trying to close a deal. Stale or conflicting records can also send an otherwise capable agent down the wrong path. In a separate 2025 Gartner survey of 413 marketing technology leaders fielded from June through August, 45% of respondents whose agents were in pilots or production said vendor-offered agent capabilities did not meet performance expectations. Gartner also reported that 81% of this survey group were piloting or had implemented agent initiatives. These figures describe a different population and survey period from LeanData’s; they should not be combined.
Recommended Free Tools
Build a useful AI-agent inventory and control process
An inventory is a foundation for control, not a solution by itself. Gartner recommends agent policies, centralized discovery, identity and access controls, lifecycle review, information governance, behavior monitoring, remediation, and employee training. The following sequence translates those recommendations and the GTM survey’s process concerns into an operating checklist; it is a synthesis, not a tested implementation recipe.
- Set creation and action rules. Define who may create or share agents, which connectors are allowed, and what actions an agent may take without human review.
- Discover and register agents across routes. Include sanctioned platform agents, features built into GTM tools, custom API applications, and shadow tools. Record each agent’s purpose and an accountable human owner so that an entry can be acted on, not merely counted.
- Give agents identities and least-necessary access. Scope permissions to the records and actions needed for the task. Review connector access and retire redundant or obsolete agents rather than leaving old credentials active.
- Govern the data agents can use. Establish source-of-truth definitions, freshness expectations, and access boundaries. An agent with broad access to stale or overshared customer data can magnify existing problems.
- Monitor behavior and respond to exceptions. Watch for activity outside an agent’s intended scope, such as unexpected record changes or outreach patterns. Define who can pause an agent, investigate, and remediate a problem.
- Keep useful activity logs. Preserve enough detail to attribute actions and investigate incidents, while limiting who can inspect logs and what personal information they expose.
- Document the GTM processes agents touch. Write down customer-data definitions, routing rules, campaign triggers, and sales-marketing handoffs. These are the operating assumptions an agent needs in order not to create overlapping or mistimed work.
- Train users and share responsible practices. Make it clear how to request an agent, report unexpected behavior, and share useful patterns without bypassing governance.
What visibility should include—and its trade-offs
A 2024 ACM FAccT conference paper offers a conceptual way to think about visibility through three mechanisms: identifiers for agents, real-time monitoring, and activity logs. They answer different questions:
- Identifiers help establish which agent or responsible operator is associated with an action.
- Real-time monitoring can reveal behavior while it is happening, when intervention may still prevent downstream effects.
- Activity logs support later attribution and incident review, but are only useful if they capture meaningful actions and can be interpreted.
More visibility is not automatically better. Detailed monitoring and centralized logs can expose sensitive information or concentrate decision-making power. Set access controls, retention limits, and clear purposes for monitoring; make visibility sufficient for accountability without turning it into unrestricted surveillance. The paper is a conceptual framework, not a test of GTM products or proof that any one mechanism is sufficient.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What agent-governance tools can and cannot do
AI governance platforms and agent inventory or monitoring tools may help discover agents, manage identities, inspect activity, or organize logs. Gartner identifies AI TRiSM tools as one possible way to discover and categorize agents, but its cited release does not endorse a specific vendor. No vendor comparison or product-performance result is established by the cited surveys.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Tooling cannot substitute for named ownership, bounded permissions, reliable customer data, documented processes, and a response plan. When duplicate outreach or conflicting changes appear, check both the agent layer and the underlying routing, campaign, and record rules. Otherwise, an organization can gain a polished inventory while keeping the same coordination failures.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
