Joe Gershenson didn’t just want to push technology forward — he wanted to see its impact in the real world.
“I went to grad school because I wanted to push technology forward. And I realized, while I was there, that it was really important to me to see the effect that I was having on the world, and to use technology in a way that made people’s lives better,” he says.
That realization led him to leave academia with a master's degree and jump into the world of startups. From co-founding a Y Combinator-backed company to leading machine learning teams at Stripe and Adept, Joe has spent his career chasing the intersection of cutting-edge AI, real-world applications, and high-impact teams.
Now, as Assembled’s new Head of Engineering, he’s bringing that same mindset to the future of AI-powered customer support.
Joe’s approach to his career has been simple: Find the next big shift in technology and figure out how to make it useful.
“My career has mostly been trying to find problems that I think are in areas that technology is about to substantially change and advance, and finding the best ways to solve those problems and move them forward,” he explains.
After leaving Adept, one thing was clear: LLM technology is a sea change. And while AI’s impact on industries like software development and creative work was widely discussed, Joe saw another area primed for transformation — customer support.
“I think people are still underestimating the number of existing jobs that are going to be substantially changed by even today’s generative AI technology,” he says. “AI for customer success and customer support was a huge space.”
More than just a big market, it was a meaningful one.
“There are a lot of people in this space doing jobs they don’t particularly enjoy,” Joe says. “It’s a grind to respond to Tier 1 support requests that don’t require much skill. If you talk to people who work in call centers or handle these types of tickets, they’ll tell you — it’s not fun to be stuck doing repetitive, low-skill work.”
He saw an opportunity: AI that doesn’t just replace jobs, but makes them better.
“I saw a path for companies to up-level the work humans are doing — making it more meaningful and increasing the dignity of that work — so people feel more fulfilled in their roles,” he says.
Joe didn’t find Assembled by accident. Through a mutual connection, he met Ryan Wang, Assembled’s co-founder and CEO, and quickly realized that the company was already tackling the exact problems he was most interested in.
Then came the real moment of clarity: seeing Assembled’s Assist Copilot in action.
“I was looking at a demo of a very early version of Assist Copilot, where we drafted a response for a support agent and handed it off for them to send,” Joe recalls. “I saw the flow — where we generate a suggested response, the agent tweaks it to match their style, and then sends it — and I thought, Oh, this is exactly how you train the best system to do this.”
For Joe, this wasn’t just a cool AI tool — it was the right approach to AI-human collaboration.
“If you’ve spent any time training AI systems, you know this is the secret sauce,” he explains. “It’s all about getting those precise adjustments — having the data to fine-tune the model for the exactly right behavior. And I thought, Okay, this is why Assembled is going to win.”
Joe has seen firsthand how easy it is for AI startups to get lost in the hype. But Assembled is doing something different.
“People don’t care that you solved their problem using AI, right? People actually care whether you've solved their problem,” he says.
That understanding — deep, customer-first product thinking — is what sets Assembled apart from the flood of AI-powered support tools.
“It has consistently been my experience throughout my career that it's easier to add a flashy technology hype layer on top of a rock-solid business that deeply understands the customer problem than to do the reverse,” Joe explains.
In other words, Assembled isn’t just layering AI onto a broken support experience. The team understands workforce management inside and out, and they’re building with a real vision for how AI can actually help support teams, not just automate them away.
When Joe first saw the Assist AI Copilot, it was a powerful but simple tool — drafting AI-generated responses that support agents could tweak before sending. But Assembled Assist has since evolved into an omnichannel AI agent capable of handling customer inquiries across chat, email, and phone.
Joe puts it simply: “Our big recent launch was about omnichannel support — handling email, phone, and chat — because people really want one platform that serves as a seamless interaction point.”
Beyond just answering customer questions, the future of Assist is about resolving real issues. That means pushing AI beyond surface-level automation and giving it the ability to actively handle tasks — whether that’s retrieving information, making policy-driven decisions, or escalating issues only when necessary.
So, what’s next? Joe sees two big challenges ahead:
“When everyone installed chatbots, they weren’t great at handling complex issues or actually resolving things for users,” he says. “I think we’ve all had the experience of interacting with a chatbot and getting a not-very-useful answer pulled straight from its own help docs.”
Joe sees a better path forward — one where AI doesn’t just regurgitate knowledge, but actively helps customers get what they need. The goal is to make Assist more actionable, so it can do more than answer questions — it can fix problems before they escalate.
Joe sees a fundamental shift happening in how companies manage support.
“If you fast forward two years, every serious company will have internal agents, outsourced vendors, and AI agents — all working together,” he says. “They’ll need to orchestrate handoffs between them seamlessly. And I think that's a space where Assembled is very well positioned to grow.”
In the past, workforce management was about balancing in-house teams and outsourced vendors. Now, AI agents add a third layer — and the real challenge is figuring out how to seamlessly route issues between AI and human agents while ensuring high-quality customer experiences.
Joe believes this is where Assembled has a massive opportunity to lead.
“When working with disruptive technologies like LLMs and generative AI, you have a responsibility to deeply understand the existing systems — the Chesterton’s fence — and ask, Why are things this way? Only then can you determine, Okay, here’s what we should change and how we should do it differently.”
Ultimately, AI support agents shouldn’t just be ‘good enough.’ They should be better than what humans could do alone.
“We live in an era where we should be able to build AI solutions that aren’t just worse but cheaper,” Joe says. “I want the support agent Assembled is building with Assist to actually be a better experience for the end user. That’s really the dream in this space, and anything short of that isn’t good enough.”
So, what kind of engineers should be paying attention to Assembled? According to Joe, it’s simple.
“The people who thrive here are the ones driven by a real motivation to build something that helps people do their jobs better,” Joe says. “They’re excited to work at the cutting edge of LLM technology and figure out how to use these advancements to actually solve real problems for real people.”
And if you like solving big, complex problems at scale, Assembled is the place.
“We have so much to do,” he says. “We really need people who are happy to roll up their sleeves, take on a bigger problem than they've run into before, and start breaking it into parts.”
The way Joe sees it, Assembled is the perfect mix of startup energy, technical ambition, and real-world impact.
“If you want to step in and look at a piece of infrastructure that's been in prototyping mode and say, what would it take to make this scale 100x over the next year? Then we absolutely have a thing that we need you to work on.”
Joe didn’t just join Assembled because it was an interesting problem. He joined because it’s the right problem, at the right time, with the right team — and now, he’s leading the charge in building AI-powered solutions that redefine how support teams work.
Assembled is hiring engineers who want to build AI that actually helps people. If that’s you, we’d love to chat. See our open roles.
