Thank you! You're subscribed!
Oops! Something went wrong while submitting the form.
AI and Automation
Kustomer interviews Assembled CEO Ryan Wang on the future of AI and support ops
May 14, 2025
Kustomer interviews Assembled CEO Ryan Wang on the future of AI and support ops
Bri Tischner
Product Marketing
At the April 2025 Kustomer CX Summit, Assembled CEO Ryan Wang joined Kustomer CMO Gabe Larsen for a conversation on the future of AI, support operations, and the evolving role of people in CX. Watch the full interview here.
How Stripe’s support playbook inspired Assembled
Kustomer: Could you tell us a little about Assembled and your background?
Assembled: I'm Ryan, co-founder and CEO at Assembled.
I first got into support operations at Stripe. I was there when it was just 80 people, with co-founders Patrick and John Collison doing support themselves and having the whole company over to their apartment for all-company support rotation. We watched Stripe scale to 800 people, and at that point, you can't fit everyone in an apartment anymore. We needed systems like workforce management to help a company scale to millions of customers, and that's how we got into support operations.
At Assembled, I've worn a lot of different hats — I was the second CEO, I was writing code, I was working on marketing, and now I'm kind of pointing people around. We are the AI platform for superhuman support. Our AI agent automates over 70% of interactions over chat, email, and voice. Our workforce management platform automates forecasting, scheduling, real-time management for support operations of all shapes and sizes. Companies like Ashley Furniture, Brooks, and Robinhood all trust our platform.
How AI is helping support teams scale without sacrificing quality
Kustomer: Customer support seems to be at a crossroads with AI and new technologies. What fundamental shift do you see happening?
Assembled: What I'm hearing from many leaders now is "we're trying to do more with the same." One customer's top line is growing 30%, their contact volumes are growing 30%, and the CFO said, "Here's your budget for this year — zero net-new headcount."
AI is part of many companies' strategies, but it's not the whole story. Modern support naturally divides into these three tiers:
- Tier 1: Routine interactions
- Tier 2: Specialized knowledge requirements
- Tier 3: Complex concierge service
There's some amount of automation in tier 1 (maybe 70–80%), AI copilots making people more efficient in tier 2, and then tier 3, where the long-term aspiration is recognizing that support isn't just a cost center.
The best companies understand that support isn't just about preventing churn — it's driving retention, customer happiness, and enabling cross-sell and upsell opportunities. They're asking, "How do we make our people better and upskill them?" rather than just, "How do we cut headcount with AI?"
Kustomer: How are you seeing the relationship between AI and human agents evolve in modern support teams?
Assembled: One of our customers, a president of a large Employer of Record company, put it well. He said in tier 1, about 70% can be automated, but the remaining 30% involves complex questions like, "What are the employment laws in Germany?" or, "How do you set up a business in Australia?" For these questions, he doesn't trust LLMs because they can hallucinate. He wants human specialists who are specifically trained for these complex situations.
For tier 2, he wants AI copilots that make humans more productive. And in tier 3, it's about utilizing specialists most effectively to maximize business impact. It's more nuanced than the headlines from a year ago suggesting "AI will replace everybody."
A lot of what we're seeing with AI is more iterative than people might think. LivePerson went public in the 2000s with conversational chatbots — this isn't the first time we've seen this type of technology. The industry has gone from AI to big data to machine learning and back to AI. In many ways, there's nothing new under the sun.
The voice bot space shows particular promise. These systems are now dynamic and lifelike, offering step-function improvements in capability that should cause companies to rethink their strategies. Some early customers are realizing that maybe the channel doesn't matter anymore — if an AI is equally effective across chat, email, and voice, and you'll eventually need to pass to a human anyway, perhaps we shouldn't be thinking about siloed channel strategies.
When policy misses the point: Redefining support excellence
Kustomer: How do you define operational excellence in customer support?
Assembled: Operational excellence is ultimately about tying numbers to what's actually happening. One director of workforce management at a large delivery company was asked by their CEO, "How do we know we're actually solving our customers' issues at scale?"
They told us about a customer who had an $800 weekend order cancelled. The support team followed policy and issued a refund, but digging deeper, this was for an important family event. The refund alone didn't solve their real problem, and the company likely damaged the relationship with a good customer.
True operational excellence is the ability to connect metrics like SLAs, response times, resolution rates, and deflection rates to the end-to-end customer journey. It's about instrumenting your systems to genuinely know if you're solving customer problems, not just hitting performance metrics.
Kustomer: Where should people start on the journey to operational excellence?
Assembled: I'm coming from the advantage of workforce management, where one of the really simple things we do is help people understand quality and efficiency. These are often seen as trade-offs, but they don't have to be.
The core insight of workforce management is aligning supply and demand. If you identify when calls are coming in (the arrival pattern), translate that into staffing needs based on productivity metrics, and compare that with when people are actually scheduled, you often find misalignments. For example, one of our customers discovered their busy time is Tuesday morning, but they had staffed heavily for Wednesday afternoon.
Shifting schedules to match demand patterns allows you to improve both quality and efficiency simultaneously. Frontline staff typically understand this intuitively — they know when the busy periods are versus when they're twiddling their thumbs. Flattening those peaks and valleys is fundamental operational improvement.
Why the best teams start with the customer, not the tool
Kustomer: With so many new technologies emerging, how should companies think about point solutions versus unified platforms?
Assembled: When we talk to leaders, the fundamental question they're asking is, "How do we make support better?" They have a mental map of their tech stack and customer journey and how these interlock.
You might want a point solution in certain areas like AI if you're focused on innovation and early adoption. Or you might not, if you see it as an iterative space. The most successful approach we've seen is when companies work backward from "Where are the greatest opportunities to improve the customer experience?"
We always present customers with a menu of options and partners with their pros and cons. The market isn't so competitive that any vendor is all bad — there are many different providers with different strengths and weaknesses. It's about understanding what you need where and making informed choices.
The next big question in CX: How do we upskill the team?
Kustomer: Looking ahead, what's the next frontier in support operations that you're most excited about?
Assembled: I'm most excited about answering the difficult question we keep getting: "What is the role for people?" I think we're past the "do more with less, AI replaces humans" talking point, but it has left a chill in terms of people's receptiveness to AI.
It's wonderful technology — everyone uses ChatGPT in some form — but in the business world, there's rightful hesitation. Support leaders are asking how they can make humans and AI work together in tandem. The focus shouldn't be on replacing jobs, but on moving people along an upskilling path.
That's what everyone I talk to wants to achieve. The challenge is finding that path forward — what does this upskilling actually look like? That's the frontier I'm most excited about exploring.
Thank you! You're subscribed!
Oops! Something went wrong while submitting the form.