How one CEO introduced AI without damaging trust

Nathan Strum did not begin his company’s AI journey with a strategy deck or a warning. He began with a wardrobe. Faced with the thoroughly ordinary problem of building a shelving system in his closet the Abby Connect chief executive turned to ChatGPT, uploaded pictures of another cabinet in his house and asked a string of practical questions: what kind of wood is this, where do I buy it and how do I build one myself? The project got done. And the story soon made its way into conversations at work.

Abby Connect is a 100-person customer service company based in Las Vegas, with a new office in San Francisco and a business model that sits close to one of the areas many expect AI to reshape first. When I spoke to Strum I expected to hear another familiar story about AI adoption: a business moving quickly, a leadership team keen to innovate and the usual talk of productivity. What emerged was a grounded account of how people come to trust a technology that many still find unsettling. 

Before the company introduced AI more formally leaders invited employees to talk about how they were already using tools like ChatGPT in their personal lives, from solving everyday problems to experimenting with new ways of working. The intention was to build familiarity and ease, to make the technology feel accessible, useful and worth exploring. 

This decision runs through everything Abby Connect has done since. The company is approaching AI as an organisational change process shaped by culture, trust and the realities of day-to-day work. 

Culture sits at the centre of Strum’s story because, in his view, it sits at the centre of the business. “Our secret sauce isn't very secret, it’s a great culture,” he tells me. “It's a very simple thing to say but it's a very difficult thing to actually execute.” 

In a customer service company, where the work can be repetitive, pressured and emotionally wearing, this carries real weight. Abby Connect has spent years building an environment in which people felt supported and able to grow. AI has arrived carrying a different sort of pressure: excitement, uncertainty and the familiar fear that a new technology may narrow careers rather than expand them. Strum’s response has been to deal with that atmosphere directly. “Building confidence around AI that it doesn't have to destroy your career was a big part of that,” he says. 

How to build confidence in AI at work

The route in was simple and, precisely for that reason, effective. Employees began by talking about how they were already using AI in their own lives while Strum shared his wardrobe story. This created a point of familiarity before the company asked them to think about work. Courses followed as did internal conversations about growth, capability and the skills people might need next. “We still want to equip you with this new skill set,” Strum says. “And that’s the way we build confidence.” 

There is also a communication lesson here. Plenty of leaders introduce AI through an efficiency lens and then sound surprised when employees hear risk before opportunity. Strum suggests a more durable route. He held staff meetings and round tables, listened to reactions and invited ideas about where AI could make work easier or more effective. One of the more revealing points in our conversation was that many employees were less preoccupied with fear than with usefulness. “More of the conversation was about how is this going to help me?” he reveals. 

In other words Abby Connect was not searching for grand demonstrations of what AI might one day do but instead looking closely at where work already slowed, where information stalled and where effort was being spent to little effect. “We want to start from the people systems first,” Strum says. “And where can AI fit into that and help where the systems break down?” 

Where AI adds value in real workflows

This way of thinking brings the discussion back to something many organisations still neglect when they talk about adoption: the shape of the work itself. One example is product development. Sales teams were hearing valuable feedback from prospects every day, including the reasons they were not buying and the features they wanted, yet very little of that intelligence reached the product manager in a form that could easily be used. “It’s like pulling teeth sometimes to get them to report these things to the product manager,” he says. 

AI, in this case, provides a way of listening across those calls, categorising requests and feeding them into the development workflow more systematically so that priorities could be spotted earlier and decisions made more quickly. In other words, the technology was deployed against a specific workflow blockage. The result, Strum says, is that things are pushed out to clients faster. 

What Strum is really describing here is a familiar organisational weakness: useful information exists, people are hearing it, yet it does not move cleanly enough through the business to shape decisions at the right time. AI is helping Abby Connect deal with that gap in a more systematic way. In this case the gain is speed. In other organisations it may be visibility, prioritisation or simply less wasted effort.

It is also a reminder that some of the most valuable uses of AI sit inside routine operational problems that teams have learned to live with over time. Handoffs get missed, requests are repeated and insight sits in calls, inboxes and conversations without ever fully reaching the people who need it. HR leaders will recognise that pattern immediately because it shows up in every kind of organisation. Strum came back to the same point more than once in our conversation. “Start with the people systems.” It is a useful phrase because it keeps the focus on work design, which is where so much of this story begins.

Watch Nathan Strum explain how Abby Connect built employee trust before introducing AI tools

Watch the In Practice video and transcript here.

Using AI to reduce stress in frontline roles

Another vivid example is the receptionist role itself. Abby Connect has developed an AI receptionist in response to client demand, particularly from smaller businesses under pressure to reduce costs, yet Strum is candid about its limits. “The AI really couldn't handle 100 % of the calls,” he says. “And you really don't want to trust AI to handle those high- pressure calls.” So the company has built around that reality. When friction appears in the conversation, when the caller asks for a person or when the exchange begins to lose fluency the call is transferred to a human receptionist.

What interests me more, though, is the way the same logic is being applied internally. Strum describes how his principal engineer looked at the receptionist role and saw stress as the problem worth solving. Legal intake calls, for example, require staff to listen closely, reassure the caller and capture a large amount of detail at speed. 

“Our principal engineer said I want to make the receptionist job a little bit less stressful. And I said, okay, I'm shocked because usually you engineers want to build a technology. And I thought you were going to get excited about the AI receptionist. And he said, yes, I'm excited about the AI receptionist but I'm also excited about helping your human receptionist do their jobs better and have less stress. And so now we're looking at every point of stress for the receptionist on the human side, and we're going to introduce AI to help them,” explains Strum.

The resulting AI tool is now being used to listen live and fill in the form as the conversation unfolds. “They’ll be able to sit there with their hands up and the AI will fill in the form for them and take all that stress off their plate so that they can focus on the call and they can focus on connecting with the call,” he says. This is a strong illustration of augmentation in practice. The technology is absorbing administrative load so the human can do the more demanding interpersonal work better. It’s about the quality of work as much as the efficiency of it.

This is an important lesson as many AI workplace stories still circle around cost, productivity and replacement. These issues are real but they can flatten the more interesting question: what happens to human capability when routine strain is reduced? Strum’s answer is simple. “We feel less stressed. We feel happier about our job. And we feel more inspired to do our work.” He is quick to acknowledge that “all the talking heads and the business leaders are going to cringe” at that as a measure of success. Yet he also argues that these feelings show up in calls, in customer interactions and in product building, and says that the company is growing faster than before. In service businesses especially employee experience is rarely separate from customer experience. 

Why failed AI rollouts do lasting damage

None of this means that the company is treating AI casually. Strum is realistic about failure. He is open about experiments that did not work and tools that proved more exciting in theory than in practice. He talks about setting up an isolated environment to test an AI agent because it was “just too dangerous” to let it touch the company directly. In theory it could have served as a personal travel assistant. In practice “it broke down and it just was faster for me to look at the price line myself.” There is a broader point here. Leadership teams damage trust when they roll out immature tools too widely and then retreat after the problems become obvious. Strum has watched others make that mistake and wants to avoid it. “We knew we had to be very careful,” he says. 

Governance comes through in other ways too. Abby Connect has medical clients, so data handling and security was a serious concern from the start. Internally, one of the first policies was to stop employees using personal ChatGPT accounts for work tasks. Company accounts were provided instead, with settings designed to reduce exposure and model training risk. Strum also makes a point that many HR leaders will recognise immediately: ambitious employees can create risk as well as momentum. “We have overachievers and sometimes overachievers can be dangerous with AI,” he says. Capability grows fast but judgment often lags. Any serious AI adoption programme needs to account for both. 

Another area Strum warns about is recruitment. He has clearly thought about this carefully and, importantly, from experience. “I think one mistake that I want to point out and I want to help people not make this mistake because I think it's serving a disservice to our society is relying on AI to screen resumes,” he says. “I think it's very dangerous.” 

He accepts that AI can help categorise information and handle volume. What he rejects is the idea that it should form an opinion about candidate quality or rank people in a way that shapes hiring decisions. This comes from personal experience and the company “turned that off right away.” 

What leaders should take from Abby Connect’s AI adoption case

Strum offers three key lessons for organisations looking to adopt AI. 

The first is honesty. “You've really got to believe what you're saying,” he says. Employees, he adds, “can see right through” leaders who privately view AI as a cost-cutting device while publicly presenting it as empowerment. 

The second is to work from people systems and inefficiencies rather than from “this nice shiny object.” 

Finally, the third is a phrase that stays with me because it sidesteps some of the lazy thinking that still dominates this debate. AI, Strum says, is “a rebalance of expenses.” Money, effort and human attention are being redistributed across the organisation. Some tasks get cheaper, some capabilities improve and some teams become more effective. The question is where the gains go and what kind of work they make possible. 

Many chief executives have a view on AI, but Strum offers something more useful: a close-up account of implementation shaped by trust, practical experimentation and a clear respect for the people doing the work. This is where many AI conversations still fall short. They stay at the level of promise rather than close to operations, judgment and human experience. For leaders trying to introduce AI without draining confidence or damaging culture this makes it worth paying attention to.

Related reading:

• Designing AI for better, not fewer, jobs
• Why frontline work is AI’s biggest opportunity
• Human and machine intelligence: what HR must govern
• Job reinvention – a guide to