Cash flow, creativity, and compassion are not mutually exclusive™

Peter Swimm: Stop AI-ing the Wrong Things — Put Your Humans First

We’re all so busy “AI-ing” everything that nobody stops to ask — are we actually solving the right problems for the real humans on the other end? 

Peter Swimm is a conversational AI product lead and solutions expert with nearly three decades of experience building and scaling technology that actually works for humans. Peter got his start in tech in 1996 teaching communities how to use the internet through the Chicago Public Library — and that people-first foundation has shaped everything since. He’s launched and scaled conversational AI products at companies like Pando Networks and led product strategy at Howdy, where he helped build the Slack API ecosystem before Microsoft acquired the team. He has since done enterprise conversational AI work for Walmart, Loews, Rabobank, and UnitedHealth Care, and at Microsoft Copilot Studio helped spearhead enterprise-grade generative AI capabilities. His firm, Toilville, is AI-aware but technology agnostic — because Peter believes the best asset any organization has is the human minds that shape it.

Peter shares why so much technology is being deployed in ways that are shortsighted and frankly unempathetic — cutting costs without actually solving the real problem. Take a listen.

To access the episode transcript, please scroll down below.

Listen in for…

  • How to embed empathy-driven requirements directly into your technology specs
  • What it means to design systems to enable dignified escalation
  • Why the best solutions never cross the line from helper to surveillance
  • Why you need to understand how work actually happens before implementing AI in your organization
  • What operational compassion at scale really looks like in practice

“(Many) clients don’t know how work happens today, and so when they go to automate it, they have no A and B to compare against how it was before. Every company has soft-skilled people that are secretly holding the company together through sheer will.” —  Peter Swimm

From Our Sponsor:
Check out this offer from our sponsor, Better Benefits USA: Are You Overpaying for Employee Benefits Without Knowing It? Take the free 10-minute benefits audit at https://findbetterbenefits.org/

Join Maria at The Rising Executive Women’s Retreat, Nov. 4-7. Use code “Edge500” to get $500.00 off! https://retreat.therisingmastermind.com/ 

Episode References: 

About Peter Swimm, Founder and CEO, Toilville: 

Peter L. Swimm is a conversational AI product lead and solutions expert who started in tech in 1996, teaching the community how to use the internet through the Chicago Public Library.

Since then, Peter has built and scaled conversational products from early adoption to enterprise durability, including work at Pando Networks (Intel Capital) and launching the Slack API ecosystem at Howdy, where Peter led product strategy across Botkit and Botkit Studio before Microsoft acquired XOXCO, the team behind Botkit and the Howdy Slackbot.

Toilville has supported conversational AI work for major enterprises including Walmart, Loews, Rabobank, and UnitedHealth Care, and at Microsoft Copilot Studio helped spearhead enterprise-grade generative and conversational capabilities, spanning connector ecosystems, customization of language understanding, and architectures for LLM-powered actions.

Toilville is AI-Aware but technology-agnostic and believes that the best asset any organization has is the human minds that shape it.

Connect with Peter Swimm:

Toilville: toilville.com or peoplemakeitbetter.com

LinkedIn: linkedin.com/in/peterswimm

Connect with Maria:

Get Maria’s books: Red-Slice.com/books

Hire Maria to speak: Red-Slice.com/Speaker-Maria-Ross

Take the LinkedIn Learning Courses! Leading with Empathy and Balancing Empathy, Accountability, and Results as a Leader 

LinkedIn: Maria Ross

Instagram: @redslicemaria

Facebook: Red Slice

FULL TRANSCRIPT:

Maria Ross  00:04

Welcome to the Empathy Edge podcast, the show that proves why cash flow, creativity, and compassion are not mutually exclusive. I’m your host Maria Ross. I’m a speaker, author, mom, facilitator, and empathy advocate, and here you’ll meet trailblazing leaders and executives, authors and experts who embrace empathy to achieve radical success. We discuss all facets of empathy, from trends and research to the future of work to how to heal societal divisions and collaborate more effectively. Our goal is to redefine success and prove that empathy isn’t just good for society. It’s great for business. We’re all so busy AIing everything that nobody stopped to ask: Are we actually solving the right problems for the real humans on the other end? My guest today is Peter Swim, a conversational AI product lead and solutions expert with nearly three decades of experience building and scaling technology that actually works for humans, Peter got his start in tech in 1996, teaching communities how to use the internet through the Chicago Public Library, and that People First Foundation has shaped everything since. He’s launched and scaled conversational AI products at companies like Pando Networks, backed by Intel Capital, and led product strategy at Howdy, where he helped build the Stack API ecosystem before Microsoft acquired the team. He’s since done enterprise conversational AI work for Walmart, Lowe’s, Rabobank, and United Healthcare, and at Microsoft Copilot Studio, helped spearhead enterprise-grade generative AI capabilities. His firm, Toilville, is AI-aware but technology-agnostic, because Peter believes the best asset any organization has is the human minds that shape it. Today, Peter shares why so much technology is being deployed in ways that are short-sighted and frankly unempathetic, cutting costs without actually solving the real problem. We discuss how to embed empathy-driven requirements directly into your technology specs. Why you must design for what he calls dignified escalation, knowing exactly when a human needs to step in, so customers aren’t left fighting your organizational complexity. He also gets into why the best solutions never cross the line from helper to surveillance. And one of my favorite insights from this conversation that many organizations are simply AIing the wrong use cases entirely because they don’t understand how work actually happens, and we talk about what operational compassion at scale really looks like in practice. If you’re a leader making decisions about technology, automation, or AI, this conversation is an important one to sanity check yourself. Take a listen. Welcome, Peter Swim, to the Empathy Edge podcast. I am excited to have this conversation with you about bringing humans back into technology, because that has been the drum I’ve been beating for a really long time, ever since I was in change management early in my career. So, we’re going to talk a little bit about bringing humans in, empathy-driven requirements for tech projects, all the good things. So, welcome to the show.

Peter Swimm  03:22

Hey, thanks for having me, Maria.

Maria Ross  03:24

So, before we get going, as we do with everyone, tell us a little bit about your story and how you got into this work of being an AI expert and a product leader, someone with a lot of experience in tech. How did you get here, and what makes you most passionate about it.

Peter Swimm  03:39

iYou know, I think a lot of my career trajectory has been falling upwards because my first job in high school, my first job in high school, I was I worked at the library and at the library they just got internet computers and it’s 1996 and so they moved all the car catalogs of the way and they put a couple of Packer Bell computers up with internet connection and I was teaching seniors and people from the community how to search for stuff on the internet.

Maria Ross  04:01

Nice.

Peter Swimm  04:02

And so, like people, the library had its audience, and they wanted to learn, and they understand the concept and pedagogy of learning stuff. And I think a lot of the hype around the internet, which surprise continues its day, tries to paint the internet as a new paradigm when it’s really just a tweak of an existing paradigm, like web pages or magazines or card catalogs, it’s all kind of the same, and that really stuck with me as I started getting into tech and getting more and more jobs, almost in like emperor has no clothes way.

Maria Ross  04:30

Right.

Peter Swimm  04:31

And so over the years, I really became focused on the intersection of like human desire, where it meets the edge of technology to do stuff. And as I became a product owner, it really became clear to me that like technical requirements sometimes get lost in the weeds from the initial human need and desire to the point where the requirement becomes more important than the outcome.

Maria Ross  04:50

Yeah,

Peter Swimm  04:51

and when we get into things like conversational AI now with AI, we’re so close to the wish to the execution of the wish that we have to really. Really, be careful when we execute the wish that the execution is going to result in what we want to do, and that is basically the cornerstone of my consultancy. Is there’s a lot of opportunity for business people to use AI technology to expand and increase the reach of their people, but you need to make sure it happens on your terms and it needs to happen with your outcomes.

Maria Ross  05:21

So, Peter, I love what you’re saying because, as I mentioned to you before we started recording, I started my career in change management, and so much of the technology implementation became about the features and the functions and the widgets. And the coding teams lost sight of what they were actually trying to help humans do with the solution, and they would create these methods and these instructions that were not intuitive at all. And my job as change management was to help translate this so that users could adopt it. And it’s like, no, it’s not just oh, press Control Alt Six exclamation point, and then they can find the menu. It’s like they’re not going to know that, even if they get trained in the moment, they need to be able to use this seamlessly. So, how can we design the technology solution to meet the business requirements of what the person actually needs to accomplish? And it doesn’t matter all the whiz bang features and gadgets and bells and whistles you can add onto it. It’s does this help them accomplish their job and achieve their goal. So, I love what you’re saying here about this idea of keeping the human front and center when you’re working on technology implementations. You talk about empathy-driven requirements for technology projects, AI projects. What do you mean by that? And can you give us some examples?

Peter Swimm  06:38

Sure. I think a lot of a lot of technical challenges in tech are focused so much around user stories to the point where you have valid user stories and invalid user stories, and things that are invalid are considered orthogonal to the result of the valid ones, and is very easy for people to marginalize the out groups of people who fall into the other bucket as an edge case or something that doesn’t affect the most of the people, or not the desired outcome. And what you see now, a lot of companies are AIing their products. Is one they don’t understand why people use their products, and so they’re AIing the wrong cases. So technology often is misused and used by people in different ways than is intended. And if you put that process more on rails and stricter, you make the program less useful for the people off your beaten track.

Maria Ross  07:31

Right.

Peter Swimm  07:32

And because you don’t understand what portion of your audience that is, you could be losing a lot of people with your new features that come out every six months. It drift farther and farther away from this audience that may like your project. I think a lot like during COVID, the United Kingdom tracked all the COVID cases in Excel, and the reason they did that because probably it’s hard to spin up a database and it’s hard to like get access to stuff and everyone can do in Excel and then they hit the limit on Excel tables.

Maria Ross  08:00

Yeah,

Peter Swimm  08:00

because there’s it’s not meant to be a database,

Maria Ross  08:03

right?

Peter Swimm  08:03

And so is that is that the product’s fault or the user’s fault? I don’t think it’s anyone’s fault. Like people use what they had, and you know you need to be acknowledge that they maybe you know as a product owner, I’m sure you’ve seen this where you watch a usability test, and that’s like watching like an alien from outer space putting the mouse in their mouth and not understanding why is he not clicking this program right? Yeah, it’s because you come so focused on design and the pattern of it that you don’t think that like oh this is an open field like a game system and users we have very intuitive pattern matching brains and we won’t pick up the manual we’ll just try to like figure it out, and there’s desire lines all throughout your application that maybe a better design pattern than you thought was the best design pattern.

Maria Ross  08:48

So how do you how do you get to those empathy driven requirements?

Peter Swimm  08:53

One, I think you need to make people requirements as strict as legal compliance gates are, or accessibility requirements are, or other things that you can’t ship without. I think these kind of things are the first thing to go when you’re trying to hit a deadline in a big company. And so, if your end goal is the program must be easy to use and delightful, and you’re like, well, you know, delightful is not really bound, and that’s you know you can’t really measure that, so we’re just going to like wing it. It’s just like, no, you got to find some sort of smoke signal to tie to that and understand that, and realize the cost of putting off human-friendly and human-desirable outcomes. Like for example, I work with a lot of clients that work in contact centers, and the main analytic that people own contact centers are looking for with AI is something called containment, and containment is not talking to a real person, and which makes sense, right? Because a phone call is $5 a phone call; they know how much it costs, and you want to keep those costs down. But when does the $5 keeping the cost down is an. Information disbursement problem or a company greed problem, right?

Maria Ross  10:04

Yeah,

Peter Swimm  10:05

and that is like it’s hard when you make technical KPIs the outcome versus empathetic KPIs.

Maria Ross  10:12

Right, and so what are some of the ways that I know you’ve talked about shadowing, you’ve talked about day in the life research. Can you explain how that actually works for folks that are building these tech solutions.

Peter Swimm  10:22

Yeah, you know, I find a lot of my clients don’t know how work happens today, and so when they go to automate it, they have no A and B to compare against how it is before, which is fine. You know, I like to build that as part of this and create benchmarks. But you know, you need to have, you know, I think every company has soft skilled people that are secretly holding the company together through sheer will.

Maria Ross  10:44

Oh, of course, yeah.

Peter Swimm  10:46

And you know, I don’t think the goal should be to wow. I need 20 of Mike or 20 of Cindy. It should be how do I make Cindy the editor-in-chief of a fleet of tools that don’t waste their time doing the stuff that’s easy for her and optimizes her ability to handle really complex situations and orchestrate them, and that resonates really well with employees and you know heads of contact centers, unless so with CFOs and CTOs,

Maria Ross  11:14

yeah,

Peter Swimm  11:14

because it puts a lot of emphasis on your people and your customers in a way that’s probably uncomfortable to technologists or engineers,

Maria Ross  11:23

right? Well, I mean, it’s the short-term gain of like we’re cutting costs. I can’t remember the name of the company last year that tried to replace all of their customer service agents with AI, and it was a total debacle. And they had to bring people back because it wasn’t about the customer experience; it was just about them trying to cut costs. And so you’re making these decisions that look good on a balance sheet. But what are you doing to the long term revenue and long term sustainability of the of the business? You’re degrading it, and then now now not only do you have to get back on track, you have to regain trust from your customers and from your employees. Right? Imagine that company trying to hire people back. Someone going, well, I don’t know. Are you going to fire this position again? Right? Yeah, yeah. You guys don’t seem to know what you’re doing, right? Yeah. So I think that short term, that short sightedness, even from the most experienced executive, they’re trying to jump on the tech train. They’re trying to jump on the AI train, and they’re not thinking through the people component, as you’re saying. How do your clients do? Because this is something I talked about in my first book, The Empathy Edge, of getting in the trenches and where it’s possible, getting into the experience of your customer, of your client, right? Whether it’s going through your own buying process on your website or calling customer service, or the one that I wish every CEO of an airline would do is go fly incognito and coach and see what that’s like, right? What’s a realistic way that they can do day in the life research?

Peter Swimm  12:47

You know, I think one of it is they have people in the trenches already. They don’t have to put on a fake mustache and go undercover boss. Yeah, like they just need to empower the people on the ground to filter out the relations. Like in my early days in startup, I did a lot of support, and I think support is an incredible place to understand where the rubber hits the road of your stack and the people and what they expect. And I was very blessed to be in a company where the CEO’s desk abutted mine and the CTO’s best desk, and so I could just say, “Hey, Tom, I think there’s something wrong with this, right? Yeah. Look at this email. This is a good example of something that is not-it’s not a bug, you know. Like it’s just like a annoyance.

Maria Ross  13:27

Yeah,

Peter Swimm  13:27

it’s an annoyance, and it’s just like we’re making this unnecessarily hard. Yes. You know, very few companies are segmented with that sort of opacity between the orgs, and so I think this is why like AI is very appealing to sea level people because they think that’s all just like the sausage making, but really it’s the research and it’s also you know the customer signals and the business signals. And if you don’t have a realize, you know, I think companies are optimized now for funding and exits and not so much as customer retention and year-over-year growth. So it’s very hard to like say, yeah. If we pay attention for three quarters, the next six quarters will be that much smarter.

Maria Ross  14:07

Right, and three quarters is like a lifetime to these companies. Like they were like, I might not even be here in three quarters, right? Which is right, right. And that that is my fundamental frustration with the way companies are led these days is that there’s no long-term view. It feels like there’s no long-term view now. There’s a lot of talk out there from different worlds that I’m in, where it’s like, okay, there is no more five-year plan, there is no more 10-year plan because things are changing so rapidly. But you do need contingency plans. You need scenario planning so that we know that we can pivot quickly if scenario one doesn’t pan out, but scenario two is is the reality, right? So I want to talk quickly. I kind of got off on a tangent there. You talk about this idea of respectful human in the loop. So can you talk? Can you explain what that means and how you how you build trust? When you’re trying to automate,

Peter Swimm  15:01

sure. I think a lot of people who, especially those who are part of the system that is human in loop, have issues with the system, is when it crosses the line from being a helper to being surveillance or being like second guessing. Like people really criticize ChatGPT as like autocomplete on steroids, and you know, like it’s telling me stuff I don’t mean to complete and stuff like that. And so I think that comes from these products being made by people who don’t have a human resources background, humanities background. I use ChatGPT experimentally, and the other day it started doing BuzzFeed type breadcrumbs at the end of conversations. Like, would you like five tips about ways to avoid this behavior? But like, no, I asked you the first time to tell me the five tips. Don’t string this out into a conversation. Like, yeah, you know, you’re Google, you’re not BuzzFeed, and I think they kind of misunderstand the usage of the product. So Human a Loop to me is like r2 d2 and Star Wars, where he’s you don’t even notice it’s there doing anything until like it does something that saves your butt, right? And it doesn’t, it’s not talking to you. It’s not trying to ask you about your day or any kind of artificial thing. It’s just giving you information in time. Like my dentist, she has a really great system that pulls up a Facebook card or the caller whenever they call, and so you’ll see my face and my last dental appointment and my next dental appointment and the notes, so she knows everything about me. So she can say, “Hey, Peter, how’s it going? How’s your daughter? Right. And I don’t think it’s how that root canal go. Yeah, exactly. Yeah, yeah. How you feeling? And it’s just like, yeah, it’s helpful. Like that’s the kind of like empathy I think that is useful. Where like here’s five facts about the person that you may never talk to before, but the organization knows. So as representative of the organization, this information is helpful to you in solving problems.

Maria Ross  16:53

I love that. I love that idea of looking at it as a helper versus surveillance. That is key right there. Explain to us a little bit about designing systems to enable dignified escalation. What does that mean?

Peter Swimm  17:06

I think a lot of times we force people to navigate our organization chart in a way that is not like it’s not like escalation of concerns. Like you’re like you okay, I’m talking to front counter person, I’m talking to the manager, and I’m talking to CEO. That’s okay, but I think like if you’re having artificial experience or tech-driven experience, you shouldn’t like make people endure the pain of having to transfer someone to. I’ve like I’ve worked for companies so big that each contact center is the size of a company’s contact center, and they have 18 of them, and so you may get passed around, and it’s the same network and the same system. Like, why can’t a person just like, you know, why can’t a person just say I’m Walmart or I’m so and so and you know like

Maria Ross  17:51

yeah

Peter Swimm  17:51

get the information and

Maria Ross  17:53

yeah

Peter Swimm  17:53

I think this is kind of like where technology overrules the empathy side of things where I called the phone number for Walmart I didn’t call the phone number for sub department, sub department, sub department.

Maria Ross  18:04

Right.

Peter Swimm  18:04

And if you’re going to present yourself as a unified thing, don’t put your complexity on me, you know, as a customer or an employee or whatever. And I think a lot of people they come to me with AI solutions that are very hard because they made it overly hard way back at the beginning of time when they put their org together.

Maria Ross  18:22

Yeah, like I was a management consultant in the ’90s, and this was a problem then. It was the whole like, you know, we’re forcing our customer to navigate our complex org chart versus just give everyone who has access to picking up a phone from a customer the information they need. I mean, and maybe that’s not always realistic, but but the amount of times where it’s you’re getting past and passed and passed, and they’re asking you the same questions. Why is there not a screen following me? Of let me take a second and figure out where you just came from and what your initial question was. And it’s decades we’ve been dealing with that. Will that ever go away?

Peter Swimm  19:00

I mean, we had the opportunity at the beginning of ChatGPT because it has something called a context window, and when I build systems for things, we have ephemeral memory, and I think it works like human memory. We’re just like, okay, you know, it’s today’s the date, and you’re the person, and this is the order, and here’s the facts. Right. So as you get passed around, that little baseball card should bounce around for agent agent, and you should normalize the outcome where a person is like, okay, let me talk, let me send you to the decision maker here, who will ask you some more questions. They’ll know everything we said, and it will be a warm handoff. And the warm handoff problem has been effectively solved for 10 years, but people refuse to build it and use it. And you’ll see, you’ll see it like I’ll go to a website of a technology company that’s announcing new AI technology, and they’ll put a chatbot on it, and it doesn’t even have the same design as the website of the event.

Maria Ross  19:56

Right.

Peter Swimm  19:57

Doesn’t seems to me like you’re not a unified AI company. If all your stacks are completely different, you know

Maria Ross  20:02

exactly.

Peter Swimm  20:03

And I think like a lot of the Occam’s Razor stuff I do with my clients is like, okay, we’re going to cancel these three SaaS services and we’re going to put the data directly in product four. Yeah, you know, because you don’t need those two in middle products anymore because you should just go from email into Salesforce. You don’t need to like pass things around, and it’s safer and it’s more secure, and you’ll cut your bill down. You know, so why not?

Maria Ross  20:29

Yeah, you have talked about a lot of your work is around helping organizations operationalize compassion at scale. What does that mean to you?

Peter Swimm  20:37

I think a lot of it is like every decision you make in the company has a cost, right? And so, like, you can say the customer is always right, and that has a cost. And then along the line, you’d be like, “Wow, the customers are really ripping us off. Let’s make a department. And this is a real, a real department a client had. They had a customer demonology department where they isolated and banned problematic customers.

Maria Ross  20:58

Okay, yeah, no, like you cannot hate your customers.

Peter Swimm  21:03

I mean, like, like it’s one thing to say fraud, but you know, I like I had a job in college where a woman would buy a gallon of milk every week and then return it the day after expired, two thirds drink, and say she got it spoiled.

Maria Ross  21:19

Uh huh.

Peter Swimm  21:19

And so we just gave her new milk, right? And like, it’s milk, you know. Like,

Maria Ross  21:24

right?

Peter Swimm  21:25

She buys hundreds of dollars of groceries, and she like has a little scam she has with us on milk. And I think people should be allowed at the shopping level have to deal with customers, be like, it’s not worth the trouble, right?

Maria Ross  21:36

Have that have that autonomy of like what’s what’s worthwhile and what’s not right? And I’ve always said about customer service policies and how you can make them empathetic or not is don’t design for the exception, design for the rule, and then just the exceptions are going to be there. The fraudsters are going to be there. You’re just going to have to chalk that up as a cost of doing business if you don’t want to piss off all the rest of your customers, right?

Peter Swimm  22:02

Yeah. Well, especially if the goal of your of your contact center is volume, right? Do you really want a person fighting a lost cause for 20 minutes on the phone?

Maria Ross  22:12

Yeah.

Peter Swimm  22:12

You could just like, okay, the minute this phone call costs $5, we’re actually arguing about more than the cost of the phone call.

Maria Ross  22:18

Right.

Peter Swimm  22:18

So let’s cut it loose. And so

Maria Ross  22:21

that’s the thing I notice about Amazon, which it’s so funny. In some cases, they have very empathetic customer service, and other ways the organization is not empathetic. But from a customer service standpoint, all of their customer service agents, no matter where in the world they are, at least in my experience, have been like, “Yeah, just keep the item. We’re going to refund you for the book, or we’re going to refund you for the spoiled food you got from Amazon Fresh, or like there’s no there’s no chain of command they have to go through. They’re empowered to do that and keep me as a happy customer with all the other money that I spend with them, right? And so that process is designed to accommodate, you know, having trust in and seeing the best in your customers, versus there’s going to be some people. There’s going to be point 0000 2% of people that are going to say they didn’t get the item, but they really did. So we need to make a whole policy to punish everybody who says they didn’t get their item. You see what I’m saying? Like it’s no, no, it’s not even worth it.

Peter Swimm  23:20

I mean, you see the same thing with like Medicare fraud or like other things. It’s just like I would rather have 5% fraud than 1% hungry kids. Yeah, you know, as a person, as an organization. And so, like at the end of the day, you’re just like, man, people are stealing all these razors. We better lock them up. It’s like, okay, so every everyone has to wait 20 minutes for the guy the key to be found, and then find the key and unlock it because yeah, you think your profit losses this year are driven by

Maria Ross  23:50

right

Peter Swimm  23:50

slippage.

Maria Ross  23:51

Yeah, that that drives me absolutely bananas at the pharmacies when I go in person. And this is I every time I experience it, I’m like, this is why you guys are losing people to online sales. Like, this is exactly

Peter Swimm  24:04

why. Well, and then you know, like the Target boycott, I think is a really good example. This because if Target was doing everything right and they engendered this boycott, it would probably be less successful. But since people are already half annoyed with them, the boycott is also like, cool. I can have a moral victory as well as an annoyance victory with this company. Yeah. And so, don’t give your customers reasons to hate to leave more than they do. Exactly.

Maria Ross  24:31

Exactly. So I also want to talk about. There’s so many great nuggets that you talk about in your work. You have this idea of a humane incident user experience.

Peter Swimm  24:42

Yeah,

Maria Ross  24:42

is that kind of what we’re talking about right now, or is that something different?

Peter Swimm  24:45

No, I think it’s. I mean, every decision made in a system should be recorded not in a surveillance way, but in an arbitration way, and so you should understand like the acclimation of decisions in the day and the net. Result of that in terms of your company vision and ethics and standard and business and all that stuff, and I think people are thinking like, oh, we need to make sure our employees are productive, and or I need to make sure that I have good ROI on the AOI. Like they always say, how do I measure ROI in AI? I’m like, how do you measure ROI on your day? What’s your ROI on sending an email by hand versus like how would you know to base it off of? Yeah, they don’t, and I think like this is kind of like the f around, find out moment of things where people go to AI and they’re like, not really seeing the effect by automating our existing business process, but 10 times faster.

Maria Ross  25:39

Yeah,

Peter Swimm  25:39

faster. You’re not more. You’re not more elegant. You’re not more sophisticated. Like the AI is garbage in, garbage out.

Maria Ross  25:47

Right.

Peter Swimm  25:48

And you know, it takes a lot of the chief insight for me. I took a day. I was sick, and I took a week off, and I did a vibe week where I just did vibe all week, and I let the vibe do everything, and the chief like thing for me is when you just say your wishes into the wish box, and they come out like based on the median response of that, and it’s very inefficient, and it repeats itself all the time, and I’m not happy with the outcome. So you have to like be as mindful as you would any creative endeavor and process, yeah. And so, if people are like wishing their bad system into the box, it will happily produce the the bad system faster and wider, and

Maria Ross  26:31

exactly, exactly,

Peter Swimm  26:33

and less effective.

Maria Ross  26:34

Well, and also, I years ago I worked for a business process management software company, and the whole thing that we would talk about, and this was like pre AI, right? It was automating your business processes, but that was always our advice: is you’ve got to you’ve got to figure out the optimum process first before you automate it, or you’re just automating inefficiency. You’re just creating inefficiency at scale, faster and wider, and in a more you know, impactful way. So unless you know, unless you got your process down, and to your point, from an empathetic point of view, is it providing the outcome that your employees or your customers need? You got to fix that first before you’re just like, let’s automate the heck out of everything, right? Because then you’re just, you’re just one of the things we used to talk about was business process outsourcing, and it was the same thing of like, don’t outsource a process that you haven’t optimized yourself to get the result that you want because now you’re just doing it at scale.

Peter Swimm  27:30

And humans have this great thing, you know, of if they don’t like something, they’ll they’ll ride it out, you know. And so like, if if you like, okay, now we have a new system, you have to type everything you did in a day into the wish box, and the wish box to tell you what you did wrong, and you’ll do just enough to fulfill the requirement, and you’ll slow roll it, and it’ll fail. And the things that I launch to succeed are things that silently become part of the system and enrich the process, and doesn’t change materially the way they execute the plan of how they work, they are just better informed and better hydrated for the day of doing things, and they’re better aligned. Right, and those are the kind of things that are like, what company has that as KPI, an alignment factor? Yeah. Where like, I when I was at Microsoft and we announced Copilot, there was 109 different Copilots, and they all spoke different programming languages and were in different products, and they used different models. Some used ChatGPT, some didn’t at all, and they couldn’t talk to each other.

Maria Ross  28:29

Yeah

Peter Swimm  28:29

And so this was a mandate of like, okay, you need to make AI, but no one said this is how we as a company are going to make AI, or this is how we’re going to evaluate what makes good AI, right? And these are our guardrails.

Maria Ross  28:42

Yeah, yeah.

Peter Swimm  28:44

I mean, there was there is guardrails in ethical AI, but they were suggestions and they weren’t compliance processes. And there was you know gates you would have to go through in a dog and pony show kind of way. Yeah, it wasn’t like this won’t ship without this kind of moment.

Maria Ross  28:58

Yeah,

Peter Swimm  28:59

and I think that was such a huge wasted opportunity because you know they had a head start. They had the technology in house. We were already working on ChatGPT stuff when ChatGPT came out, and you know they were so sure that the way that they’d be building stuff was always the right way to building stuff. They didn’t miss the opportunity, and the company and the tech industry has missed opportunity of like, what, how does this redefine how we make software? Like, what you know, and like I’m talking to a client right now, and they want to synergize. They have a bunch of information streams from all out their organization, and they want to make AI summaries of things.

Maria Ross  29:34

Okay,

Peter Swimm  29:34

and I’m arguing that people don’t want AI summaries. They want an AI whiteboard. They want to have a Pinterest, and so when your big pattern imagined brain looks at all the information aggregate, you connect the things together and like, oh, here’s the Venn diagram of where all this is being ignored, and this is the opportunity. Right. Sometimes

Maria Ross  29:53

you got to get your hands dirty, and that’s I think that’s the human blending. I’ve talked about this on the show before. There was a. They came out from MIT last year about the five skills that complement AI to get the most ROI out of it, and they’re the human skills that currently AI can’t replicate. And guess what’s you know on there? Empathy, right? Yeah, judgment, ethics, perception, connection, all of those things. And they’re saying that the best ROI you can get out of AI is not necessarily ripping and replacing, but blending, and that’s exactly what you’re describing there. It’s like letting AI do the grunt work of pulling everything together, but then you still want the human brain to look at it and make the connections and get into the data. I was speaking to someone on a past podcast interview about customer insights and leveraging AI to pull everything together, but you still need someone to look in and find the patterns and understand the nuances of the conversations. Right? If you just have AI do everything for you, you miss the nonverbal cues. You miss the the you know pauses or the hesitancies when they respond to something, you miss those connections that you make when you’re actually involved in the process. And this is the thing of I think everyone, I think a lot of companies, and tell me if I’m wrong, they are struggling to find that balance of connecting the human with what AI can do for them, and how those how they how that can work together. That’s where they’re doing so much experimentation right now.

Peter Swimm  31:26

And I think like the very two best conversational AI experiences I’ve ever worked on was one was a bot that read the contract and compared it to 13 months of email to make sure all the agreements from the email made it into the contract.

Maria Ross  31:40

That’s amazing.

Peter Swimm  31:41

Like, absolutely, that is a tedious data-driven job that AI is so good at. And like, oh, you know, you see says here that you’re paying for all goods and services. I thought we were going to split that.

Maria Ross  31:54

Right.

Peter Swimm  31:54

It’s us. Or online, and you can redline all the stuff. The other one was an airline. They in the text message they sent you, your flight was canceled. It offered you three alternate flights, and you can click one button and instantly book it. And I think like, yeah, bad news, good news. Click here, and those 30 people are going to run the counter right now, and you’re going to be in line for an hour. Or you can click this button and go have a beer, you know, and you’re booked, you know, like, and I think those are like the opportunities that we’re leaving on the table with the AI automation thing. We’re just like, okay, if 60% of the people are going to choose this, let’s give them a button to choose it, and then that’s how we’ll contain customer service floods. That only people very strange or unusual or detailed requirements are the ones who have to talk to the people take a lot of time,

Maria Ross  32:43

right? And you’re saying that that’s not a good approach.

Peter Swimm  32:46

No, I love that approach because you know, like, I should never have to call a phone number to find an address, right? Right. I should never have to call a phone number to get office hours or you know a mailing address or fax or you know, or find maybe even what are my policies and my insurance plan, or you know, and everything you have in a company that the context of their answers, you should have a list of those and say, okay, this is something that 100% could be automated, and this is something that could be automated for some people, and this is one that should never be automated.

Maria Ross  33:16

Right.

Peter Swimm  33:16

Like if a lawyer, if if someone calls and says, I’m a lawyer, or I ate something and I got sick. Like that should go to a person. Like don’t even mess around. You don’t mess around with you know don’t mess around with people or anything and force them to go through hoops. Right. And if someone calls and says like my food came cold and it’s a $10 order and it’s going to cost $6 to find out what was wrong with the food, you give them a coupon for the food.

Maria Ross  33:38

Yeah, yeah.

Peter Swimm  33:40

And I think that, like, I worked for a client, and they launched an SMS ordering bot for a taco chain.

Maria Ross  33:45

Okay.

Peter Swimm  33:46

And it could order food, but it couldn’t refund bad orders. And so the bot had very poor analytics. And like, why does everyone hate our bot? It’s like, well, look at all the things that the people who voted you down are complaining about. They’re complaining about the food quality. It’s late. I was double charged, and there’s no one to talk to. Right. So there, there’s absolutely nothing you could do to improve that score until you approve that. And it’s just like, ah, well, you know, it’s like, you know, like you, you can’t, you can’t like make people bend to your will in customer service.

Maria Ross  34:17

Right. Exactly.

Peter Swimm  34:18

You know, you have to meet them where they are, and that’s what we mean. When the customer is always right, it means the customer has a point.

Maria Ross  34:25

Right.

Peter Swimm  34:25

Doesn’t mean it doesn’t mean that it’s a valid point or whatever. But and as a company, you have to decide what is the cost worth of dying on this hill.

Maria Ross  34:32

Right. Right.

Peter Swimm  34:33

And well, and I think that’s also empathy too.

Maria Ross  34:34

Yeah. And there’s been studies about you know 97% of customers in one survey, I believe it was 97% said empathy is the most important part of their customer experience when they’re calling into a call center, even if their problem is not solved. And I think that’s an important point, right? So again, the customer isn’t always right, but how are you handling it? Are you handling it with empathy and compassion? You might have to say no, or we can’t do that, or we can do this other thing that is not necessarily the thing you wanted. But that requires discernment. That’s something that you might need to get in front of a person of just like, wow, that’s terrible. I’m really sorry that you experienced that.

Peter Swimm  35:17

Yeah,

Maria Ross  35:18

we can’t, you know, whatever it is. Like we can’t refund the money on that, but here’s what we can do. Would this be acceptable to you? This is, and you at least feel like the person tried or the organization tried, even if you’re doing it through a bot, right? That’s the most important thing to people is that they feel heard and not dismissed, right? One time I got from a from an online site, they botched an order that I sent to a client, and when I contacted customer service, I immediately-I don’t know if it was a bot or if it was someone because it was a few years ago-and they just immediately said, “Oh, sorry that happened. Here’s a coupon for your next purchase, and I was so pissed off because I was like, “You didn’t even acknowledge that what you did was like a hit on my reputation, I don’t want a coupon because I don’t want to spend money with you again, right? So throwing the coupon at people isn’t always the right answer either. It’s the empathy, it’s the listening, it’s the you know how can we make this right? And for some people, it’s not the coupon. It’s not the you know whatever. It’s something else. It’s even just an acknowledgement that something went wrong.

Peter Swimm  36:22

Yeah, and I think in all systems, AI or otherwise, there should always be arbitration moment, right? And if the arbitration is like this is what it’ll take to keep my continued business, and the company says no, that’s okay, that’s your decision,

Maria Ross  36:34

right?

Peter Swimm  36:34

And I have no actionable legality around this, then that’s what it is. And I think I think companies, and especially people who make those decisions, they mistake empathy for agreement.

Maria Ross  36:46

Yes,

Peter Swimm  36:47

like you don’t have, you don’t have. It doesn’t mean I agree

Maria Ross  36:48

with you. I talk about this all the time. Empathy does not mean I have to agree with you. It just means I understand what you’re going through and I acknowledge it.

Peter Swimm  36:56

Yeah, I mean, is it right for this woman to get half a gallon of free milk every two weeks. I don’t care. Like, like I see at the end of the day, we throw out gallons of milk in the trash that we can’t sell. So how is that any different as a moral dilemma? So how can you judge her? I worked at grocery stores that would purposely damage their food to discourage people from scavenging it out of their dumpsters, put glass and bleach in it

Maria Ross  37:23

Oh my gosh!

Peter Swimm  37:25

We yeah, for the you know I did some numbers and the cost we spend to throw away food in this country, we could ship that all to food centers and save a billion dollars.

Maria Ross  37:34

Yeah, yeah.

Peter Swimm  37:35

So and it’s a it’s a crazy like sunk cost fallacy thing. And sometimes we believe in things we want to be true more than they are true, and that even applies to AI solutions.

Maria Ross  37:46

Absolutely, absolutely. Well, we could probably talk longer, Peter, but we’ve got to we’ve got to wrap up. So we are going to have all your links in the show notes, so people can find out more about you and your work with your company Toilville. But for anyone who’s on the go right now or exercising while they’re listening to us, where’s the best place they can find out more about you and your work?

Peter Swimm  38:04

Sure, you can check us out at peoplemaker.com, or you can find me on LinkedIn under Peter Swimm. I’m the only one.

Maria Ross  38:11

I love it. Thank you so much for your insights and for this great conversation today.

Peter Swimm  38:15

No problem. Thanks for having me, Maria.

Maria Ross  38:16

And thank you everyone for listening to another episode of the Empathy Edge podcast. If you like what you heard, you know what to do. Please rate, review, and share it with a friend or a colleague. And until next time, please remember that cash flow, creativity, and compassion are not mutually exclusive. Take care and be kind. For more on how to achieve radical success through empathy, visit theempathyedge.com. There you can listen to past episodes, access show notes and free resources, book me for a keynote or workshop, and sign up for our email list to get new episodes, insights, news, and events. Please follow me on Instagram at Red Slice Maria. Never forget, empathy is your superpower. Use it to make your work and the world a better place.

Cash flow, creativity, and compassion are not mutually exclusive™

Learn More With Maria

Ready to join the revolution?

Find out how empathetic your brand is RIGHT NOW, and join our newsletter to start shifting your perspective and transforming your impact.

This field is for validation purposes and should be left unchanged.
Privacy