Smart Outsourcing: How BPOs Pair Humans + AI to Elevate Customer Service

• Name: Travis Low
• Role: Head of CX and AI Operations Practice Leader, Hugo
• Experience: 15+ years in managing business process outsourcing solutions for brands across every industry
• Company: Hugo builds scalable, multilingual teams for customer support, digital operations, and data/AI. We pair trained specialists with modern tooling for 24/7 coverage and measurable outcomes.

Interview questions

Q1. What does "great" human-AI support look like in practice?

“Great” human-AI support in practice is giving your customers what they want, in the way they want it, and at the speed they prefer. 

When I started in the BPO space, omnichannel support was a big thing. The concept was “wherever the customer goes, we want to be there.” Then chatbots were introduced, and the conversation switched to wanting to serve everyone with a chatbot because it’s cost-effective and gets people answers quicker. 

But in practice, that’s not what happened. It’s a decision that was made on the business side that assumed customer preferences. I give this background because it’s important to note the industry got it wrong. 

Now we’re in the age of human-AI work. Great human-AI support adapts to actual customer needs and preferences. Customer service tech solutions like Text App provide workspaces with shared analytics that allow you to deliver more personalized customer experiences. It enables businesses to use AI to offer truly responsive support that actually delivers what customers have always wanted.

Q2. Where does AI actually work today — and where shouldn’t it?

AI works well for basic transactional tasks, research, summarizing, extracting insights from massive amounts of text and data, knowledge retrieval, etc. When you attach these capabilities to integrations that do things for you, that's where it delivers the most impact. Cooking is a good example. AI can figure out what you need for a recipe based on a photo of the things in your fridge. Then you can build integrations that automatically create a shopping list. 

Where shouldn’t AI work? Things like high-emotion customer support inquiries that uniquely leverage human empathy. In BPO, there is automation, and then there is artificial intelligence. The two are similar and sometimes used synonymously, but in reality, they are quite different. 

Automation follows an “if-then” logic within predefined parameters. AI can make decisions without requiring explicit parameters. The key is using AI for the routine, high-volume data-driven decisions.

Q3. How do you decide when to automate vs. assist a human customer support agent?

For many companies today, automation is built into their customer support operations. Right when you start live chatting with a brand, you receive an automated response (from a bot). It triages or helps you with simple requests before you need to speak with a human agent.  

Each brand might choose to do it differently. Ultimately, the decision to prioritize human interaction or automation depends on factors like cost and the desired customer experience. 

There are still brands that specifically choose to serve the customers very quickly with a human. American Express is a good example, but you pay for that level of service. Larger ecommerce retailers might do everything possible to automate a conversation before you get to a human. 

Text AI features that support human customer service in servicing complex cases.

Q4. What do the first 90 days working with Hugo look like?

First, we introduce new clients to their Hugo team, including executives and leadership they’ll be working with. Then we provide an onboarding agenda. This outlines different workflows around technology, training, and talent profiles. 

Once your Hugo team is ready — which can be in as little as two weeks — we go through a “nesting period” where new agents handle live tickets under supervision. When you’re satisfied with the team’s performance, we move into full production. Support technology like Text App enables our team to connect to our clients’ external tools quickly and seamlessly. 

The work doesn’t end there, though. We go through levels of optimization to make your customer experience better, faster, and more efficient. And we document the entire process from the very beginning. We work with a lot of startups, so these processes will change over time. But it’s important to have a baseline of how we do things as companies scale quickly.   

Q5. How do you integrate so that context follows every interaction?

All customer interactions, regardless of channel (text, call, email), are logged in a CRM. Support agents know historically how you’ve interacted with the brand. Even if you start a live chat and lose connection before resolving your issue, the conversation is still captured inside a CRM. 

Modern tech stacks are built to match even the most complex cases, so this happens automatically. And Hugo operates within the client’s existing CRM systems. They extend licenses to us, and we integrate to connect with their customers.     

Q6. How do you measure quality across both humans and AI?

Historically, you measured quality through a small sample size of an agent’s monthly engagements. This could be four calls out of 200. The QA would review those calls or emails and score the agent based on a rubric. More recently, AI software has enabled us to QA more engagements, sometimes even 100% of customer interactions. This gives you a much better idea of your agent’s performance relative to a smaller sample size. 

On the AI side, measuring quality is more complex. We can't just look at speed or final resolution rates. AI might resolve inquiries faster, but the engagement quality could be poor. You have to actually look at the chat logs to see what the customer experience was like. Did they have a hard time getting the help they needed?

When customers finally reach a human agent after a frustrating AI interaction, they rate the entire experience based on that human conversation — even though their frustration stems from the AI portion. The customer gets their issue resolved, but the path to resolution was unnecessarily difficult. This new interconnected experience makes measuring AI quality in customer experience more challenging.

Snapshot of default Text App reports with automated chats

Q7. What’s your security approach?

Hugo maintains all the compliance and security needed to operate in any regulated industry. We have SOC 2, HIPAA, PCI compliance, and HiTrust certification. And of course, we ensure the customer service software and tools we use and recommend, like Text App, are all well-trusted in the industry. 

Q8. Which outcomes and SLAs do you commit to — and how are incentives aligned?

SLAs are a commitment to specific service metrics. Every client has different SLAs around CSAT, NPS, resolution time, average handle time (AHT), and other key performance indicators. It all depends on what the brand wants. So a client can say, “We want 90% of emails answered within an hour using 30 agents.”

Incentives are performance-based. The client pays a bonus when you hit or exceed the agreed metrics. If you miss the target, the client receives service credits. This creates a true partnership where our success is tied directly to delivering the outcomes that matter most to their business. Text’s executive dashboards have proven instrumental for tracking key metrics and easy-to-access reports.