Business process outsourcing (BPO) is undergoing another fundamental shift. What began as a model built mainly for scale — measured in headcount and handle time — evolved into delivering a better customer experience (CX) and is now advancing through AI.
In an interview with Outsource Accelerator’s Derek Gallimore, TaskUs SVP of Global Offerings, Phil Tomlinson explains how the company is finding new ways to deliver services by combining what humans and machines each do best.
From traditional agent to agentic AI
GenAI has already unlocked more natural, conversational experiences. Now, we’re in the era of agentic AI — autonomous systems that can reason, take action and orchestrate multi-step workflows without human input.
AI agents handle routine tasks, like: password resets, billing queries, shipping updates, etc. They are built to know their own limitations and when to involve humans.
But despite these automation capabilities, agentic AI is not plug-and-play. “You need guardrails in place to stop it from issuing $500M instead of a $10 refund,” says Phil. “There's a lot of heavy lifting that goes into training, integration, maintaining and constant fine-tuning.”
For example, when building AI agents, you have to protect against prompt injection that could compromise the model, ensure the knowledge base, workflows and training materials are high quality and serve as a reliable source of truth for the models to draw from, according to Phil.
Agentic AI must also be taught to know when to hand tasks over to people — if the interaction calls for empathy, nuance or critical thinking, only a human can provide the right support.
“There are always going to be customer service encounters that require high degrees of emotional intelligence,” notes Phil. “Imagine you’ve rented a car through an app. You’re in the middle of nowhere… your car breaks down and your kids are in the backseat. That’s not when you want to deal with an AI agent.”
The AI evolution in specialized services
Another aspect of BPO evolution, AI is reshaping domains like Trust & Safety by automating parts of processes that were traditionally reactive. For example, on most major platforms, the vast majority of hate speech is already being flagged by AI.
“AI is doing a huge amount of work at the top of the funnel,” says Phil.
Beyond efficiency and scale, the technology also protects employees. AI filters and removes the most egregious content, such as graphic violence or child abuse, before it’s seen by a moderator. This helps reduce the cognitive and emotional toll on the humans behind the scenes.
Content that falls into gray areas, however, like memes or commentary, still demands human judgement. “Detection is not the same as enforcement,” Phil points out. “AI is great at saying, ‘Here's a million pieces of content. I think most of these are bad. Can you double-check them?’ That's where the handoff between AI and humans becomes really important.”
Redefining roles in the age of AI
Given this critical relationship, Phil points out that humans are not being replaced — a fact that’s often lost in the AI conversation. Instead, they’re being redeployed to where they matter most.
In Trust & Safety, moderators are shifting from end-point reviewers to upstream experts who are training models, evaluating system performance and fine-tuning AI outputs. In customer support, frontline agents are becoming quality leads, workflow architects and escalation strategists.
In fact, this shift demands stronger investment in people. At TaskUs, we have a global Wellness & Resiliency team of 200+ clinicians and licensed mental health professionals who support our frontline to reduce mental health risks, enhance employee well-being and build a more resilient workforce equipped for high-impact roles.
Rethinking the BPO model
As the nature of work evolves, so must the business model. Phil says to expect new pricing models that reflect results over hours and headcount, and offerings that include more end-to-end services — to make you more “sticky” in a changing market.
Ultimately, he believes providers who combine people and technology — and deliver measurable outcomes — will win in the market. Despite the hype, the future of outsourcing isn’t an either or proposition.
Listen to the full podcast for more insight: “TaskUs on Agentic AI, Tech-First Solutions, & the Future of Outsourcing – with Phil Tomlinson of TaskUs.”
