AI is rewriting the managed services story

Recently, an insurance company came to us with a challenge: nearly 1,000 tickets a month, recurring pipeline failures, and month-end processing that could take up to 10 days. Much of the run team’s capacity was spent responding to incidents and working through a growing queue of operational issues.

It’s a familiar tale, one that reflects a managed services model built decades ago to handle and scale operational work. Capacity goes toward keeping up with recurring work, with little room or incentive to address what’s causing that work in the first place, improve the underlying environment, or introduce new capabilities that deliver greater value to the business.

AI gives us an opportunity to rewrite this story, starting with the question: What changes when managed services are accountable for outcomes, rather than activity?

Managed services emerged in an era when global connectivity made it possible to distribute repeatable work across teams and geographies. For work that needed to be performed reliably at scale, global delivery models offered companies new ways to access talent, extend capacity, and manage costs. For decades, incremental automation has been layered onto processes that remain largely unchanged from the dot-com era.

“Most of our customers are operating on a 20-plus-year-old business model,” says Anthony Marquardt, senior managing director of Agentic Managed Services at Slalom. “It’s seen iterative improvements around automation, but fundamentally, nothing about how the work gets done has changed.”

The economics of this legacy managed services model reflect the conditions in which it developed. Scaling support generally meant scaling the people needed to perform the work. That put the focus on consistency and efficiency: clearly defined work, established service levels, and the capacity to deliver. This structure has offered companies a practical way to manage complex technology environments with greater cost predictability. But it doesn’t reward reducing the amount of work that needs to be done in the first place.

At the core is an incentive problem: providers are paid to maintain the service, while root-cause resolution and meaningful improvement sit outside the baseline scope. The model can deliver needed capacity and cost predictability, but it’s ill-equipped for an environment where technology, customer expectations, and operating models are changing quickly.

Today, AI agents can take on high-volume, repeatable work without requiring a proportional increase in human capacity. And their role can extend beyond processing the next ticket to identifying patterns across incidents, surfacing potential root causes, and supporting remediation.

But adding agents to the existing managed services model doesn’t address the core issue of a service designed to process work rather than continually improve it.

“A lot of service providers are saying, ‘We’ll add in agents and reduce the cost,’” says Doug White, data and solution lead for Agentic Managed Services at Slalom. “But when you actually look at what they’re agentifying, it’s the same set of processes with agents layered on top. Like automation efforts beforehand, these incremental adjustments don’t lead to the changes needed to impact the type of work and the way these services are delivered.”

Agentic Managed Services takes a different approach, redesigning human-AI workflows so people, agents, and processes work together to deliver the outcomes business leaders care about.

By combining AI agents, human expertise, and modern engineering practices, Agentic Managed Services operates and continuously improves a service. Agents can execute routine work, identify patterns, and support remediation within defined guardrails, with human expertise applied where judgment, context, and oversight matter most.

The value of that shift can extend to every part of the business. Reducing repeatable work can lower operating costs and avoid future costs, while creating room to improve customer experiences, support growth, or introduce new capabilities.

For the insurer, rewriting the decades-old managed services narrative meant looking beyond a faster way to close nearly 1,000 tickets each month to improving the underlying environment. Through this agentic approach, the team was able to align 75% of incidents to a root-cause issue and reduced month-end processing from as many as 10 days to less than 22 hours. The new operating model also freed up $900,000 in annual IT operations costs for reinvestment.

Where that value goes will depend on the service and the business. Some services may be optimized for the lowest sustainable cost. Others may prioritize reliability, customer experience, or the ability to launch and scale new offerings. The point isn’t simply to create more capacity, but to apply it where it can create the greatest value.

That includes where human expertise is applied. As agents absorb more routine work, people have more room to focus on exception handling, root-cause problem solving, service design, governance, and continuous improvement. Greater automation also makes human accountability even more essential, increasing the need for sound judgment, business context, transparency, controls, and clear escalation paths.

In this new model, continuous improvement becomes both part of the service itself and a shared responsibility.

“Historically, the model has been, ‘We hand over the work, and you’re responsible for delivering the service,’” says Marquardt. “We want to redefine that relationship by changing the ‘P’ from provider to partner. That means both sides have a stake in making the service better over time.”

That requires aligning upfront on what success looks like, how it’ll be measured, and what each party is accountable for. Traditional service level agreements (SLAs) have often focused on measures such as resolution time and uptime. Those measures still matter, but an outcome-based model asks a different question: Is the underlying operation actually getting better?

Consider the insurer again: closing nearly 1,000 tickets a month faster would improve a legacy service metric. Identifying why those incidents keep occurring creates the opportunity to eliminate underlying causes and reduce the need for those tickets in the first place. The distinction is key: one measures how efficiently work gets processed; the other measures whether the need for that work is decreasing.

What counts as a meaningful outcome will vary by service and business. That makes a clear baseline, shared measures, and honest agreement about what each party can control essential.

The partnership must be reflected in the commercial model itself. Legacy managed services contracts could lock in scope, service levels, and commercial terms for years, with change orders layered on as needs shifted. The agentic commercial model is designed differently, with greater flexibility built in from the start. Quarterly reviews provide visibility into cost to serve, performance trends, and outcomes, creating a view of what’s working and where there may be opportunities to optimize. That ongoing visibility helps both parties adapt how capacity and investment are directed as needs evolve.