Program report · filed November 11, 2026

AI in SalesMeasured report

ServiceNow's Wilks: AI Agents Need Governance Before Scale

AI agents acting across CRM, billing and fulfillment systems need visibility, boundaries and human intervention before enterprises scale them, says ServiceNow CRM EMEA's Daniel Wilks.

By James Calloway3 min read593 words

Program notes

  1. Daniel Wilks, VP of ServiceNow CRM EMEA, named visibility, clear boundaries and human intervention as the three foundations for scaling AI agents.
  2. Wilks: AI can now access customer data, interact with systems and take action on a customer's behalf — not just summarize.
  3. Agentic workflows may span CRM, billing, inventory, CPQ, service management and fulfillment systems.
  4. Wilks said organizations need much greater visibility into what AI is doing, accessing, and who is accountable.
  5. The source cites no measured failure-rate or adoption data — the governance argument is directional, not benchmarked.
Why AI Agents Need Workflow Governance Before Scale
PlateWhy AI Agents Need Workflow Governance Before Scale — Goumbik / pixabay

AI agents that access customer data, apply discount rules, trigger approvals, or change service processes carry a different level of responsibility than summarization tools — and ServiceNow's Daniel Wilks argues that difference is what most enterprises have not yet governed.

Wilks, Vice President of ServiceNow CRM EMEA, framed the shift in an interview as a move from assistance to operational accountability. "AI is no longer simply summarising information or recommending a next step; it can access customer data, interact with different systems and take action on a customer's behalf," he said.

The distinction matters for deal-stage economics. A summarization tool helps an agent work faster; a workflow agent that initiates work across CRM, billing, inventory, Configure-Price-Quote, service management, or fulfillment systems changes the risk profile of customer service itself.

What changes when AI acts rather than assists?

Enterprise AI adoption has typically started with lower-risk support tasks: case summaries, knowledge retrieval, response drafting, and next-best-action prompts. Those use cases leave a person responsible for the final decision.

The risk profile changes when an agent begins to initiate work. A customer may experience the result as a faster quote, a case update, a scheduled repair, or a changed order. Internal teams then need to answer three questions: how the decision was made, which system supplied the data, and whether the action followed approved rules.

Wilks argued that visibility becomes central once AI starts acting across workflows. "Organizations need much greater visibility into what AI is doing, what it can access and who is accountable for its actions," he said.

That visibility cannot arrive after the event as a reporting exercise. Agentic AI needs an operating model that defines access, records activity, and gives teams a way to intervene before a poor decision reaches the customer. Customers never see the governance layer — but they feel the outcome when a workflow acts on incomplete context.

What controls do AI agents need before they scale?

Asked what must be in place before enterprises scale AI agents, Wilks compressed the control model into one line: "Three foundations are essential: visibility, clear boundaries and human intervention."

The three break down as follows:

  • Visibility — an audit trail of agent activity.
  • Boundaries — defined limits on the data, systems, actions, and decisions available to the agent.
  • Human intervention — defined points where authority returns to a person when risk, judgment, or exception handling requires it.

Those controls gain weight as customer work crosses functions. A single service workflow may begin in CRM, check an entitlement, review inventory, trigger a field-service task, request an approval, and notify the customer. Each step needs permissioned access and clear limits.

The failure modes are concrete. Poorly governed automation can reproduce the very fragmentation CX teams are trying to remove. An AI agent working from incomplete data may act too early. An agent with broad access but weak boundaries may perform actions that should require approval. Wilks warned that autonomy needs controls matched to the level of action the agent can take.

For CX, IT, operations, and security leaders, the enterprise decision is no longer only where AI can help. It is where AI agents can act, what they can access, and when human authority returns.

What Wilks did not supply is a benchmark: no measured data on failure rates, adoption timelines, or audit outcomes from deployed agent fleets. The claim is directional — faster automation reduces delay, but uncontrolled action creates customer, compliance, and operational risk — and enterprises scaling agents will need to test it against their own incident data.

via cxtoday.com (Source)

Filed under

Share this article:

More from James Calloway

James Calloway

Show full bio

Senior reporter covering technology and product news in Sales technique and methodology, sales enablement at Quota Brief.

143 articles

Also rated

  1. IBM: Agentic AI Reshapes Seller Work, Not Headcount
  2. ThoughtSpot Puts AI Agents at the Center of Revenue Operations
  3. Xactly and ServiceNow Connect AI Agents for Revenue Operations
  4. Zeliq Raises €7M to Put AI Agents Inside the Sales Workspace
  5. FrontlineIQ Taps Plaud for AI Sales Coaching Partnership

« Previous articleNext article »