TaskUs is a 60,000-person BPO whose primary work is in customer experience, trust and safety, and AI services—helping enterprises adopt and scale AI, and bring new technologies to contact centers.

Lessons Learned from the Front Lines of Enterprise CX

In conversation at Ai4, Regal COO Sahil Mehta and TaskUs CCO Jarrod Johnson traded operator-level notes on where Voice AI is truly working in the enterprise.

The premise was clear from the outset: This isn’t just theory.

“Although the price point for voice is higher, the ROI is like three times better,” says Jarrod.

Sahil added, “We’ve collectively deluded ourselves into believing that chat was a better experience.” Because, in theory, “it was a better experience for our cost basis.”

Over the course of 17 minutes, the two break this down and show why “voice is the purest customer interaction, and the way that everyone would prefer if they could choose.”

They explore this notion by tracing the TaskUs–Regal partnership, spelling out enterprise requirements (across integration, observability, and operational alignment), and tie it all together by highlighting a real-world example that shows materially higher conversion at scale.

Lesson 1: It’s Not Just About Good Tech

TaskUs has built their business on the back of high-growth tech companies, having successfully built foundational AI models with many of their clients. When discussing the company’s confidence in betting on Voice AI Agents, Jarrod said:

“What became clear to us over the past few years was that we built up this business supporting foundation models—we started to see the impact that was going to have on commercial applications…

So, we took some of the learnings we had from helping build the technology, and said, ‘how can we pivot those resources to put them back into the marketplace,’ to help clients adopt technology as part of their strategy for customer experience.”

Working with Regal allowed them to combine their knowledge of building these foundational models, with a partner that understood Voice AI and CCaaS better than anyone in the market:

“Nobody was exclusively focused on voice at the time we were talking to the market… It felt like the Regal team understood the customer use cases, their technology stacks, their telephony plans… Better than any other player.”

He went on to add that 90% of TaskUs’s partners were still using chatbots, which helped Regal stand out as a leader at the forefront of generative AI for voice.

Lesson 2: Most Enterprises Still Run QA the Old Way

Sahil and Jarrod discussed how Voice AI platforms are redefining what “quality” means in enterprise sales and support environments.

Jarrod said that, “80 to 90% of my clients today are sampling 5 to 10% of their calls,” discussing how they’re doing manual reviews of those calls, getting sentiment analysis, and then making policy, process, and product decisions based on that.

Reviewing only 5 to 10% of calls is not nearly enough signal to steer product, policy, or process at an enterprise scale. 

With Voice AI, quality becomes data-complete. According to Jarrod, “the difference that we see with Agentic AI technologies is, natively, they’re sitting on top of all of that data all the time, which provides us droves of information.”

“The really advanced platforms allow us to actually simulate and A/B test things—policy changes, process changes, technology changes—to drive higher quality results than they’re actually able to achieve today.”

What Enterprise Winners Are Doing Differently

Leaders treat Voice AI as its own operating model, not a bolt-on. 

Jarrod discussed three primary things he sees from customers that are the most successful at bringing Voice AI into their enterprise:

1. Aligning Product and Ops more tightly. 

“Where we see projects fail is when the teams who actually run customer support or sales are not aligned with product or tech teams. You just get this massive gap in terms of ROI realization.”

2. Treating knowledge as a living system. 

“Clients [don’t] anticipate up front all the care and feeding that’s going to be necessary to manage these systems over time… It’s not because the technology is flawed or high-maintenance. It’s because policies change, processes change, products change… And people don’t manage their in-house knowledge very well to begin with.”

3. Defining ROI upfront, and iterating relentlessly. 

“Is it revenue growth, is it conversion rate, is it cost reduction at scale? You got to understand [your ROI] up front and then measure it, document it, and go back, otherwise the business gets impatient.”

How a Mental Wellness Platform Used Voice AI to Drive More Conversions

Jarrod discussed how a leading mental wellness platform had a classic revenue detractor faced by many healthcare providers—getting inbound callers to set up and show for their first appointment.

“That’s a lot harder than you think,” said Jarrod.

“We were able to show them that the labor they were throwing against it today was a super high cost… Benchmark us against it. We’ll give you a more efficient experience, a better conversion rate, and we can give you way more outbound appointment setting conversions than you’re able to get today (due to cost and headcount limitations).”

Ultimately, they decided to work with Regal because it offered a clear path to revenue growth (by lifting appointment conversions) and adding new outbound capabilities, all without increasing spend.

Sahil underscored the point: “It’s not just about cost reduction… If you think about any initiative where you’re investing in new technology, you want to perform on three vectors—make more money, spend less money, enable new capabilities. With voice agents, this customer—and several others we’re working with—are able to hit on all three.”

The Time for Voice AI Is Now

When Voice AI is implemented as its own operating model, it outperforms chat and legacy automations on business return. The leaders we see are already realizing those gains in production.

The time is now, but success isn’t automatic. 

As Sahil clarified in Q&A, Regal’s focus is enterprise scale—“we are specifically focused on several use cases within the enterprise. When you get into larger call volumes, more complex governance needs and security needs, that’s where we find ourselves.”

Enterprise Voice AI takes real effort and operational intelligence across all of these areas—while instrumenting every conversation, simulating changes before rollout, and aligning Product, Ops, and CX around measurable outcomes. All at scale.

Voice has always been the customer’s most natural channel. With the help of Regal, it can also be your most operationally mature AI channel.