Latest / The Adults In The AI Room / Earley AI Podcast Episode 79: Scaling from 3 Customers to 300,000 with AI

Transcript

Earley AI Podcast Episode 79: Scaling from 3 Customers to 300,000 with AI · The Adults In The AI Room

• 0:01
  SPEAKER_00: This is the Early AI Podcast hosted by Seth Early.
• 0:04
  SPEAKER_00: Join us as we delve deep into the passions, expertise, and experiences of thought leaders and practitioners to talk about what's possible with artificial intelligence.
• 0:14
  SPEAKER_00: The Early AI Podcast is sponsored by Early Information Science, your digital transformation journey with design and deployment of innovative technology solutions, as well as Vector.
• 0:26
  SPEAKER_00: Now enjoy the show.
• 0:28
  SPEAKER_02: Well, welcome to the Early AI Podcast.
• 0:32
  SPEAKER_02: For some reason, I'm seeing your uh screen here.
• 0:34
  SPEAKER_02: Let me try, let me start again.
• 0:36
  SPEAKER_02: Uh hide.
• 0:37
  SPEAKER_02: I don't know why you're showing up.
• 0:38
  SPEAKER_02: Like the screen is all you.
• 0:41
  SPEAKER_02: Um weird.
• 0:42
  SPEAKER_00: And my video is not even on.
• 0:44
  SPEAKER_02: I know, but it but it shows Carolyn Southwick.
• 0:46
  SPEAKER_02: Is that what you see forced?
• 0:48
  SPEAKER_02: Did I only see that when she talks the rest of the time?
• 0:50
  SPEAKER_02: So so go ahead and unmute and then we'll go from there.
• 0:53
  SPEAKER_02: So great.
• 0:53
  SPEAKER_02: Uh and it should switch back to me.
• 0:55
  SPEAKER_02: I don't I don't know why.
• 0:56
  SPEAKER_02: Yeah, I'm seeing you right now.
• 0:57
  SPEAKER_02: Oh, you're seeing me right now.
• 0:58
  SPEAKER_02: Okay, great.
• 0:59
  SPEAKER_02: I only see her when she talks.
• 1:01
  SPEAKER_02: Oh, so no, no target.
• 1:02
  SPEAKER_02: Okay, great.
• 1:03
  SPEAKER_02: Welcome to today's uh Early AI podcast.
• 1:06
  SPEAKER_02: I'm your host, Seth Early.
• 1:08
  SPEAKER_02: In each episode, we explore how artificial intelligence is transforming organizations and how they manage information, improve customer experience, improve employee experience, and create business value.
• 1:19
  SPEAKER_02: Today we're going to be focusing on how AI is empowering small businesses and home service professionals.
• 1:25
  SPEAKER_02: This is an area that is not well uh covered in the industry.
• 1:28
  SPEAKER_02: A lot of uh AI is about enterprise organizations.
• 1:32
  SPEAKER_02: And these are the teams that are managing, scheduling, invoicing, communicating with clients, doing their daily operations.
• 1:38
  SPEAKER_02: They don't have technical teams, they don't have large teams of people to help them do this stuff.
• 1:43
  SPEAKER_02: So AI is beginning to simplify that work and reduce administrative overhead, to reduce those efforts, and help small companies provide more consistent and more efficient and more reliable results.
• 1:56
  SPEAKER_02: Joining me today is Forrest Zeisler.
• 1:58
  SPEAKER_02: He's co-founder and chief technology officer at Jobber.
• 2:01
  SPEAKER_02: And Forrester has spent has spent years building technology for service professionals, including landscapers, electricians, cleaners, contractors.
• 2:09
  SPEAKER_02: He's uh his experience creating these tools that make complex systems simple, gives him a very unique view on how AI can support small teams and help them grow.
• 2:20
  SPEAKER_02: Forrest, welcome to the show.
• 2:21
  SPEAKER_02: Thanks so much for having me.
• 2:23
  SPEAKER_02: So let's uh let's start off with some misconceptions around AI, especially related to smaller businesses.
• 2:29
  SPEAKER_02: So, what are what do people not get around small and mid-sized businesses when it comes to AI?
• 2:34
  SPEAKER_02: Many people think it's we see that there's personal productivity, right?
• 2:38
  SPEAKER_02: Write emails, we can summarize results, we can, we can handle meetings.
• 2:42
  SPEAKER_02: But beyond that, there's not a lot that uh people are doing uh in terms of small businesses because they just don't have the technical capacity.
• 2:49
  SPEAKER_02: And a lot of people think that they advanced AI is really for large organizations with those advanced teams, with those technical teams.
• 2:57
  SPEAKER_02: How do you see that perception changing?
• 2:59
  SPEAKER_02: And what's the reality of service-based businesses today?
• 3:03
  SPEAKER_01: It's I think that's the misconception that I'm uh here trying to change.
• 3:07
  SPEAKER_01: I mean, first off, no one is going to benefit more from AI than these small service businesses.
• 3:12
  SPEAKER_01: You know, painters, landscapers, window washers have nothing to fear from AI.
• 3:16
  SPEAKER_01: AI is not going to start installing furnaces or cutting down trees, but there's so much administrative work that happens within these businesses that is a tax on their productivity.
• 3:27
  SPEAKER_01: Right?
• 3:27
  SPEAKER_01: When they're sending out invoices, when they're dealing with payroll, when they're doing the hundreds of administrative tasks that they are distracted by every day, that's what they're not getting paid for that work.
• 3:36
  SPEAKER_01: That is directly coming out of them delivering their craft, performing their trade.
• 3:41
  SPEAKER_01: And AI is perfect for automating those types of tasks so they can focus on what's important.
• 3:46
  SPEAKER_01: When it comes to the mindset, I've been blown away by how open these small businesses are, these entrepreneurs are, to adopting technology.
• 3:55
  SPEAKER_01: They're not anti-AI or anti-technology, they're anti-wasting time.
• 4:00
  SPEAKER_01: Right?
• 4:00
  SPEAKER_01: Like they don't have time to burn cycles playing with things.
• 4:04
  SPEAKER_01: Uh, like a lot of technology, especially on the like front edge of the early adopter curve, you and I both know you can spend five hours automating a five-minute task, right?
• 4:13
  SPEAKER_01: And uh, or you can go through all this change management and get a tool and discover it only works 80% of the time.
• 4:19
  SPEAKER_01: And then you have to go through all the trouble of changing back.
• 4:21
  SPEAKER_01: And they just want to focus on like reliable, high confidence steps in their business so they can stay focused.
• 4:28
  SPEAKER_01: They don't have time to waste.
• 4:30
  SPEAKER_01: If you can show them that something is going to actually help them and give them confidence and build that trust, they are more than happy to adopt technology.
• 4:39
  SPEAKER_01: Anything that gives them the edge they'll take on.
• 4:40
  SPEAKER_02: Uh refresh my memory again of how many customers you have.
• 4:44
  SPEAKER_02: Because when we first spoke, I was thinking, well, you're just focusing on small businesses and you know what are the lessons learned.
• 4:50
  SPEAKER_02: But you've really scaled this in a significant way.
• 4:53
  SPEAKER_02: So, how many customers do you have at uh Jobard?
• 4:56
  SPEAKER_01: Oh, over 300,000 customers.
• 4:58
  SPEAKER_01: 300,000 customers.
• 4:59
  SPEAKER_01: So yeah, the the they're all you know definitely smaller compared to the enterprise customers that you or businesses that you usually speak to.
• 5:06
  SPEAKER_01: Uh, but at scale, it's we we've delivered over a hundred billion dollars worth of services through our platform now.
• 5:13
  SPEAKER_01: Wow, unbelievable.
• 5:15
  SPEAKER_02: And when you start thinking about um, you know, how uh uh you're taking these complicated workflows uh and you're you're applying AI in a way that feels natural to users, um, how you've defined the derived these and how you've kind of integrated this.
• 5:33
  SPEAKER_02: How are you applying AI in this way where people don't have technical expertise, but they need reliable outcomes, they need reliable value from automation.
• 5:43
  SPEAKER_02: It's it's putting AI in the background.
• 5:45
  SPEAKER_02: And how have you kind of gotten to that point?
• 5:47
  SPEAKER_02: What's been the process for doing this?
• 5:49
  SPEAKER_02: Because you have very, very different businesses who you're supporting on Java.
• 5:53
  SPEAKER_02: So talk a little bit about that evolution and how you've you've arrived at that.
• 5:57
  SPEAKER_02: Sure.
• 5:57
  SPEAKER_01: It's um, do you want the long uh answer or the short answer?
• 6:00
  SPEAKER_01: Sure, give me the long answer.
• 6:02
  SPEAKER_01: We have some time.
• 6:03
  SPEAKER_01: I think the first step, it's cliche, all your listeners know it, but you gotta start by talking to the customers.
• 6:08
  SPEAKER_01: Uh, make sure they're involved in every step of the process, or you will get it wrong.
• 6:12
  SPEAKER_01: It's very easy to get so excited by this technology that you get out over your skis and build something that no one actually wants.
• 6:18
  SPEAKER_01: The second key piece is always focus on the problem you're solving, not the solution.
• 6:23
  SPEAKER_01: Again, cliche, but in the AI world, it's so easy, right?
• 6:27
  SPEAKER_01: Like the customers don't want quarter-inch drill bits, they want quarter-inch holes, right?
• 6:32
  SPEAKER_01: And we could get really excited about the drill bit in the AI world because it's so new and so exciting.
• 6:36
  SPEAKER_01: For us, it was definitely a crawl, walk, run journey.
• 6:39
  SPEAKER_01: And that's kind of the story.
• 6:41
  SPEAKER_01: If I'm going for the longer one, it we actually started back in 20.
• 6:44
  SPEAKER_01: It was the title of the board deck.
• 6:45
  SPEAKER_01: Jobber will be your most trusted employee.
• 6:47
  SPEAKER_01: This idea that we could be your assistant, your office manager, your business coach, all tied into one.
• 6:52
  SPEAKER_01: And early on, we spent burn through millions of dollars hiring PhDs out of the university and trying to train our own models before the technology was ready.
• 7:01
  SPEAKER_01: But bit by bit, we've started to, especially on the back of this next generation of tools, see our momentum grow.
• 7:07
  SPEAKER_01: We started in the same spot everybody else starts, connect your knowledge base up to a rag search, stick in the product.
• 7:13
  SPEAKER_01: You got a little chatbot that can answer basic questions about the product.
• 7:16
  SPEAKER_01: That's the training wheels crawling version.
• 7:18
  SPEAKER_01: But it wasn't yet sounding authentic.
• 7:21
  SPEAKER_01: It couldn't talk in the way our customers talked.
• 7:23
  SPEAKER_01: It wasn't speaking their language.
• 7:24
  SPEAKER_01: So then we started filling it and training on our Job or Academy articles, the transcripts of our uh Masters of Home Service podcasts, 2700 expert written pieces of content.
• 7:35
  SPEAKER_01: And it began to seem more authentic.
• 7:37
  SPEAKER_01: It began to understand their needs better so it could be more flexible.
• 7:41
  SPEAKER_01: But it was still very generic.
• 7:42
  SPEAKER_01: It wasn't speaking to them.
• 7:43
  SPEAKER_01: So when you say 2,700 pieces of expert content, where did that come from?
• 7:47
  SPEAKER_01: Where did that content come from?
• 7:49
  SPEAKER_01: We've been running that blog for 15 years, interviewing customers, learning their best practices, their motivations, how they think about pricing, how they think about employees, how they think when is the time to go get a loan to get more equipment.
• 8:02
  SPEAKER_01: And having that rich content base was really excellent for building a model that could, you know, connect more legitimately to our customers and understand best practices.
• 8:13
  SPEAKER_01: So that's fascinating.
• 8:14
  SPEAKER_02: So when you think about what you did as a foundation, is you built on that tacit knowledge that you were harvesting over all of those 15 years.
• 8:24
  SPEAKER_02: And you started, you know, you were you were doing this, I imagine, primarily as a marketing vehicle and a and a guided guidance for your customers to say, okay, let's draw these people here to find out how to best run their businesses and what uh approaches and techniques they should use and how they should think about this and learn from peers.
• 8:41
  SPEAKER_02: So you were doing these interviews as a mechanism to inform your market, inform your customers.
• 8:48
  SPEAKER_02: And it really is a marketing function, right?
• 8:51
  SPEAKER_02: But then that became a repository of enormous value because then you could train your AI using that content and be able to actually provide people with real answers and information.
• 9:03
  SPEAKER_02: So that's fascinating.
• 9:04
  SPEAKER_02: So talk more about how you sort of came to that point.
• 9:07
  SPEAKER_02: Like when you were first talk a little bit about the evolution of the platform.
• 9:11
  SPEAKER_02: So you built these applications as kind of standalone standalone for uh you know landscapers and people doing painters and so on.
• 9:20
  SPEAKER_02: And and so talk about that evolution uh into building this platform um and and how you leverage sort of the the uh individual application functionality into this platform.
• 9:32
  SPEAKER_01: Early on, you know, when Sam and I we were both freelance developers, just going around talking to small businesses, trying to convince them that we could do work for them.
• 9:41
  SPEAKER_01: Uh and through that journey, we were both exposed to a lot of small businesses, quite a few of them in these home service industries.
• 9:51
  SPEAKER_01: So the one day when Sam and I happened to meet at a coffee shop shop or coincidence, uh, and he mentioned that there might be an opportunity here to build something for these types of businesses, I got very excited because I had I had clients in the space, he had clients in the space, and we got to work.
• 10:07
  SPEAKER_02: So, what was his initial idea?
• 10:09
  SPEAKER_01: Did you say very much the job of today?
• 10:12
  SPEAKER_01: It was that we both had a chance to look behind the covers of these small businesses and recognize that every piece of their business would be better off if it was systemized, organized, and eventually automated.
• 10:25
  SPEAKER_01: Our vision was always to build a platform that could run their businesses end-to-end.
• 10:30
  SPEAKER_01: We had quoting, scheduling, voicing in our very first version that we launched 15 years ago.
• 10:35
  SPEAKER_01: And obviously, it's evolved a lot.
• 10:36
  SPEAKER_01: We've brought in payments, we've brought in marketing, we've brought in a lot more mobile, we've brought in a lot of sophistication to these steps.
• 10:43
  SPEAKER_01: Uh, but the idea of helping small businesses control the chaos and run better, more successful, thriving businesses has always been our mission.
• 10:50
  SPEAKER_01: It's exciting to have never had to pivot.
• 10:52
  SPEAKER_01: That's interesting.
• 10:53
  SPEAKER_02: And so, so continue on.
• 10:54
  SPEAKER_02: So you you were building these individual standalone applications, then you thought, what's what's the commonality here?
• 11:00
  SPEAKER_02: And you really looked at the bread and butter processes that people have to do day in and day out.
• 11:05
  SPEAKER_02: And and at the time, I imagine there was not a lot of AI, but there was automation, right?
• 11:10
  SPEAKER_02: So it was the first steps of AI because it was business process automation or those types of uh or robotic process automation, RPA.
• 11:18
  SPEAKER_02: So tell tell me more about how that kind of evolved.
• 11:21
  SPEAKER_02: So you started standardizing these processes, you started building the application to cover the common themes across these different types of businesses and and and continue to so you had uh what did you have?
• 11:34
  SPEAKER_02: Did you have didn't start as a platform, it started as specialized applications for these different businesses?
• 11:39
  SPEAKER_01: Or did you it actually start as a platform on day one?
• 11:43
  SPEAKER_01: It started as a platform.
• 11:44
  SPEAKER_01: This has always been the vision.
• 11:45
  SPEAKER_01: Um, we had to build it very configurable because uh Java supports over 55 different industry verticals, and our second biggest one is other, right?
• 11:55
  SPEAKER_01: Like the long tail in this space is um wow really impressive.
• 11:59
  SPEAKER_01: Uh, but that was what makes it exciting.
• 12:01
  SPEAKER_01: And even within a very targeted industry, the variability amongst the businesses is substantial.
• 12:09
  SPEAKER_01: Something that is very configurable with custom forms and custom fields and custom workflows so that we can match how they are running today.
• 12:18
  SPEAKER_01: You can't expect businesses to bend over backwards and adapt to your product, your product has to adapt to the businesses.
• 12:23
  SPEAKER_01: And uh that that was you know there on day one.
• 12:27
  SPEAKER_01: Our vision was to solve 80% of the problems for 80% of the customers and build a platform that could deliver on that.
• 12:33
  SPEAKER_01: And that's what we've been working towards uh this whole time.
• 12:35
  SPEAKER_02: And so when you compare where you are today, so so how did you gradually start implementing more AI capabilities?
• 12:42
  SPEAKER_02: So again, initially it was process automation, so you need to really define those processes and you needed to understand how those processes varied across businesses, even if they're in the same vertical, right?
• 12:54
  SPEAKER_02: Even if they're you know selling to the same customers and doing the same services, they still had variations.
• 13:00
  SPEAKER_02: So, so I would imagine that that configuration was more intensive and it took more time and effort.
• 13:06
  SPEAKER_02: So, how did you handle that?
• 13:08
  SPEAKER_02: So, tell me about the the initial development of those configurable applications and then how that evolved.
• 13:14
  SPEAKER_01: Well, beyond just having a really strong automation engine, effectively we're talking an if-else engine, Xavier, but built into the core underlying product.
• 13:24
  SPEAKER_01: As we brought AI in, it brings us to the next step in the journey.
• 13:26
  SPEAKER_01: Once it had the knowledge of how Jobber works, it had the knowledge of how our customers work, it still needed to speak to each individual business.
• 13:33
  SPEAKER_01: And that's where we started teaching it the next step, which was how to query the individual's account.
• 13:39
  SPEAKER_01: We taught it how to do the SQL queries.
• 13:42
  SPEAKER_01: So you could start asking questions about your revenue, about your customers, which employee is driving the most revenue for you.
• 13:48
  SPEAKER_01: And it could sit there and work with you, pulling up the data and having a conversation.
• 13:52
  SPEAKER_01: At that point, it was starting to become really useful, but it still had no sense of what good or bad was.
• 13:57
  SPEAKER_01: Any CRM will show, give you a widget that'll tell you you have a 20% win rate on your quotes.
• 14:02
  SPEAKER_01: But our customers, what does that mean?
• 14:04
  SPEAKER_01: It varies so much, even geography to geography.
• 14:07
  SPEAKER_01: So we started giving our AI benchmarks.
• 14:10
  SPEAKER_01: Now it'll pop up a little widget and say, hey, did you know other four to 10 employee landscapers in Texas have a 10% higher quote win rate than you?
• 14:18
  SPEAKER_01: Let's talk about some best practices to get this number up for you.
• 14:21
  SPEAKER_01: And when you click that, then it'll discuss what you are doing, what aren't you doing, what things, if you change based on your numbers today, how will much better could your business be?
• 14:31
  SPEAKER_01: And then when you decide that you want to adjust your practices, it can guide you through the product, turn things on for you.
• 14:37
  SPEAKER_01: Interesting.
• 14:38
  SPEAKER_01: So that was like the next key piece in what you're asking about.
• 14:41
  SPEAKER_01: How do you build something, automations for such a wide range of customers?
• 14:45
  SPEAKER_01: The AI has to actually understand the customer.
• 14:48
  SPEAKER_02: And it does that through asking questions about the business, and then based on the answers, providing uh uh recommendations and suggestions based on this very broad and deep knowledge base that was built up over this period of time.
• 15:03
  SPEAKER_02: Early, so I'm trying to get a sense of what the steps were to this process, right?
• 15:07
  SPEAKER_02: That's kind of where you are today.
• 15:08
  SPEAKER_02: But when you first were uh building the platform, how much work had to be uh put into the configuration and uh talking to customers and so on?
• 15:18
  SPEAKER_02: Like what was the what was the manual component of that early on in the process?
• 15:22
  SPEAKER_02: What did it take to honor the customer?
• 15:24
  SPEAKER_01: Well, let's just talk about what it took to get our first customer.
• 15:27
  SPEAKER_01: Okay.
• 15:28
  SPEAKER_01: Sam and I built our first prototype product.
• 15:31
  SPEAKER_01: Uh, it took us two and a half months, and we started reaching out to all the small businesses we knew.
• 15:35
  SPEAKER_01: We very quickly exhausted that supply and ended up just cold calling out the phone book.
• 15:39
  SPEAKER_01: Uh, we had a whole little script and everything.
• 15:42
  SPEAKER_01: But you know, every day or two we would drive out to a job site with our computers, we would walk somebody through what we had built, show them how it could be working for their business, and we'd end every demo with is this something you would use for your product?
• 15:54
  SPEAKER_01: Or is something you'd use for your business?
• 15:56
  SPEAKER_01: And every time without fail, they'd say no.
• 15:58
  SPEAKER_01: But then we'd have to ask why.
• 16:00
  SPEAKER_01: And they'd explain.
• 16:01
  SPEAKER_01: And they were always good explanations, maybe not always, but there were wonderful insights, wonderful feedback, all feedback is a gift.
• 16:07
  SPEAKER_01: And we would take that feedback, yeah, and go and iterate on the product.
• 16:12
  SPEAKER_01: What were they just for more than almost every single day?
• 16:15
  SPEAKER_02: So, what were they telling you early on?
• 16:17
  SPEAKER_02: Like, what was the feedback they were giving you when they saw things and said, No, I would not use this?
• 16:22
  SPEAKER_02: What kinds of things were you hearing?
• 16:23
  SPEAKER_01: So early on, we sort of had uh an idealized how we thought business worked in our heads.
• 16:29
  SPEAKER_01: And then you start discovering all the edge cases, all the unhappy paths, all the you know, like, yeah, I work most of my clients are like that.
• 16:37
  SPEAKER_01: But this one client, I need to build them this other way.
• 16:39
  SPEAKER_01: And this employee, you know, can't go on this truck over here, and I really need to track these things for the business.
• 16:45
  SPEAKER_01: You gotta remember, we were our entire goal was to get people off of pen and paper.
• 16:50
  SPEAKER_01: Yeah.
• 16:50
  SPEAKER_01: Even today, that's our number one competitor.
• 16:52
  SPEAKER_01: But back then was like the metal clipboard with the handwritten carbon copy invoice.
• 16:56
  SPEAKER_01: That was our competition.
• 16:58
  SPEAKER_01: And even though it sounds like it should be easy to beat that, there's almost nothing more configurable and flexible than a blank piece of paper.
• 17:05
  SPEAKER_01: The flexibility you have to build into a product to allow their existing systems that are running on pen and paper to uh come into your software.
• 17:13
  SPEAKER_01: That was the real challenge.
• 17:14
  SPEAKER_01: We also just faced skepticism if I'm being honest.
• 17:17
  SPEAKER_01: Oh, there was a lot of it's worked this way for 20 years.
• 17:20
  SPEAKER_01: Why do I need to change?
• 17:21
  SPEAKER_01: Even though you could sit there and look at the business, and there were so many problems.
• 17:26
  SPEAKER_01: They but they felt it was okay that an invoice would sometimes get lost under the driver's seat, or that they had$400 a month in SMS overages uh trying to coordinate their team.
• 17:39
  SPEAKER_01: Like those things were considered just the harp of the course.
• 17:42
  SPEAKER_01: Yeah.
• 17:43
  SPEAKER_01: This idea that there was a better way, there was definitely skepticism of that in the early days.
• 17:47
  SPEAKER_02: So I I love that quote.
• 17:49
  SPEAKER_02: You know, there's nothing more configurable than a blank piece of paper, than a pen and a blank piece of paper.
• 17:54
  SPEAKER_02: I mean, that's kind of brilliant in a way, because that is what you're dealing with, and that is what the the mindset is look, I I just need to I just need to do this the way it works in each of these situations.
• 18:04
  SPEAKER_02: And so so you took six months of doing that day to day.
• 18:07
  SPEAKER_02: Was there one industry or one type of of business you were focusing on initially?
• 18:12
  SPEAKER_01: No, it's I go through waves depending on where I was in the uh phone book.
• 18:16
  SPEAKER_01: You know, sometimes I'm in the plumber section.
• 18:18
  SPEAKER_01: I get a bunch talk to a bunch of plumbers, and sometimes I was in the painting section talking to a bunch of painters.
• 18:23
  SPEAKER_01: But yeah, it took six months until when we asked that faithful question, would you use this for your business?
• 18:28
  SPEAKER_01: Somebody said yes.
• 18:29
  SPEAKER_02: And so then what uh so that was gathering all those requirements and finding out what those issues were and those challenges and why, you know, why digital was better than pen and paper.
• 18:38
  SPEAKER_02: And then once you started to get that, you probably I would imagine you had a very long list of configurable options for these uh businesses.
• 18:48
  SPEAKER_02: So, what did it take to start to onboard from this manual perspective, right?
• 18:53
  SPEAKER_02: People had to be walked through that process, they had to choose, you know, setting setting up their their business in all the different ways and nuances.
• 19:00
  SPEAKER_02: What did it take initially to do that once you got to that first customer?
• 19:04
  SPEAKER_01: It's early on.
• 19:05
  SPEAKER_01: You always, you know, again, I'm gonna drop a lot of cliches, but do things that don't scale.
• 19:09
  SPEAKER_01: We definitely white gloved every single customer early on, but we were careful as we were building because we knew our vision was impossible without strong self-serve motion.
• 19:20
  SPEAKER_01: You know, 6.2 million small home service businesses in just North America.
• 19:26
  SPEAKER_01: 97% of these are under 25 employees.
• 19:28
  SPEAKER_01: If you don't have a strong self-serve motion, the business just won't work.
• 19:32
  SPEAKER_01: Sure.
• 19:33
  SPEAKER_01: Uh, so really good usability, in-product onboarding, strong defaults.
• 19:38
  SPEAKER_01: When you pick your industry as you're signing up, it would pre-configure the account.
• 19:43
  SPEAKER_01: As we talked, though, even amongst painters or plumbers, you're going to have differences in configuration, but a good starting point, good interview questions built right into the signup flow that adapt the product.
• 19:54
  SPEAKER_01: Those are all things we took uh really seriously.
• 19:56
  SPEAKER_01: And I'll say AI is helping us take that to the next level right now.
• 20:00
  SPEAKER_01: Right now.
• 20:00
  SPEAKER_01: But again, I'm trying to get to the plan here, right?
• 20:03
  SPEAKER_02: Because it was all very manual at first.
• 20:05
  SPEAKER_02: And then when you were onboarding your initial customers, you knew had you had to give that white glove service.
• 20:10
  SPEAKER_02: What would it what would it take?
• 20:12
  SPEAKER_02: And you had to do that at a loss, obviously, because you're not going to charge them consulting fees, right?
• 20:16
  SPEAKER_02: You're still you're just you're simply amortizing that investment in the process in the configuration over this period of time.
• 20:24
  SPEAKER_02: And so, so what was what was the level of effort that it took initially to do this?
• 20:30
  SPEAKER_02: And and then when did it start to become when did you incrementally build in more of the self-service uh and self-configuration?
• 20:37
  SPEAKER_02: I'm taking you way back to the beginning.
• 20:39
  SPEAKER_01: So yeah, uh to understand the journey.
• 20:42
  SPEAKER_01: It's funny.
• 20:42
  SPEAKER_01: I get flashbacks as you ask some of those questions because you talk about operating on loss.
• 20:46
  SPEAKER_01: 12 months in, we had three customers paying us 29 bucks a month.
• 20:49
  SPEAKER_01: Well, everyone had my personal cell phone and would call me all hours of day or night.
• 20:54
  SPEAKER_01: And I had to pull up my computer in bed.
• 20:56
  SPEAKER_01: Uh, my wife was never happy with me, but uh, we'd sit there and uh talk to them on the phone and help answer any question they have.
• 21:03
  SPEAKER_01: From three customers to 300,000 customers.
• 21:06
  SPEAKER_02: I think that's a good, that's a good you know, headline for this uh for this initiative.
• 21:11
  SPEAKER_02: And then the first 12 months, three customers paying$29 a month, and then uh, and so so you were putting a tremendous amount of effort into servicing these customers, and then tell me more about how that kind of uh smoothed out and how those those pieces evolved.
• 21:26
  SPEAKER_01: So, what was your second year like, for example?
• 21:29
  SPEAKER_01: We built in that strong serve motion.
• 21:32
  SPEAKER_01: Anybody could sign up and get started.
• 21:34
  SPEAKER_01: They did not know we were going to offer this white glove service as they were signing up and starting to use the product.
• 21:39
  SPEAKER_01: We would call them and see if they had any questions, any concerns, if there was anything we could help them with.
• 21:46
  SPEAKER_01: And of course they did, and we tried to go that extra mile, but we again built in that motion.
• 21:52
  SPEAKER_01: So if we were answering the same set of questions again and again, again, all right, let's imp change the product to answer that question, then.
• 22:00
  SPEAKER_02: you know that moves to the what's the next most common thing we're having to help people with how can we invest in that piece of the journey so how were you supporting yourself at that time in other words did you have uh any funding was it from personal investment did you did you raise money did you have with a side job or what was the what was the uh mechanism for for doing this a little bit of side job a little bit of friends and family and quite a bit of credit card debt yeah okay and then we we did eventually uh fundraise after a couple years we got our a very small seed round but it was enough to keep us going uh how many already a couple years later yeah and how many customers did you have at that point so three in the first year and then by the time you got to funding how many customers did you have do you recall?
• 22:44
  SPEAKER_01: From starting the journey it's funny because fundraising people act like it's a single point in time you know especially that first raise from starting the conversation some money actually hitting the bank that's you know a six month plus journey.
• 22:55
  SPEAKER_01: But we started having conversations around that one year mark trying to figure out how we could uh raise capital not really understanding it I think it's actually been a strength of the business is that we were founded in Edmonton Alberta Canada about as far away from Silicon Valley as you can get very much a blue collar town.
• 23:11
  SPEAKER_01: It was quite a while before I'd ever talked to somebody who worked in a company that had raised venture capital.
• 23:17
  SPEAKER_01: It was just a foreign world and I think that was actually a benefit in disguise.
• 23:20
  SPEAKER_01: But it would have been around 2012 that we raised our first seed round I want to say we had maybe a dozen customers then it took us three and a half years before we got to our first million in revenue.
• 23:33
  SPEAKER_01: Wow wow okay uh took us another two years after that to get to four million in revenue so five and a half years in we were only four million revenue and now we're doing hundreds of millions.
• 23:44
  SPEAKER_02: Right, right well I mean it was recurring revenue so that's great.
• 23:48
  SPEAKER_02: I mean four million run rate you know a few years in is is is awesome.
• 23:52
  SPEAKER_02: And then so obviously you're and you're gonna have to give me exactly what you you raised but you you were doing a lot of research and development and putting money into this tool and this product what was the so in the first few years you were getting to that point.
• 24:07
  SPEAKER_02: And then at that point was it uh was it it when did it become self-sustaining?
• 24:12
  SPEAKER_02: When did it become profitable?
• 24:13
  SPEAKER_01: When were we able to um to handle your your your overhead and have a profit.
• 24:18
  SPEAKER_01: Once we went down the path of fundraising we never wanted to get out over our skis be one of those companies that's just burning through capital and it has to raise yeah right but we also you don't raise money for it to go from sitting in their bank account to sitting in your bank account.
• 24:32
  SPEAKER_01: The point is to deploy it for growth.
• 24:34
  SPEAKER_01: Sure.
• 24:35
  SPEAKER_01: So we always operate it at a very small controlled loss and we become profitable just at the right time.
• 24:41
  SPEAKER_01: You know we'd always pick a line in the sand where we will not let the bank account drop below this number and we'd burn to that point and then plateau there and we'd stay there.
• 24:49
  SPEAKER_01: And if we raise more money we would then go and repeat that process.
• 24:53
  SPEAKER_01: So we've been profitable many times through the history of the company but the goal on average especially back then when you're that small is to you deploy the money for marketing for sales for more engineering so you can improve your product market fit faster.
• 25:06
  SPEAKER_01: And that's what we did.
• 25:07
  SPEAKER_01: Yeah we raised our first that first seed round in 2012 I want to say was 25000 Canadian.
• 25:14
  SPEAKER_01: So not even series A we've gone went through multiple seed rounds multiple series and uh you know by 2015 we were raising a six million dollar US series A.
• 25:26
  SPEAKER_01: And then recently in 2023 we actually raised 100 million dollars in our series wow wow that's a pretty that's a pretty interesting journey from uh 2500 to 100 million.
• 25:37
  SPEAKER_02: So what's the saying or 15 years to an overnight success.
• 25:40
  SPEAKER_02: Yeah right exactly and uh so how much in uh total did you raise up until that$100 million uh raise we would have been about 80 million give or take so we've done 180 million in total raises now oh okay and so and now you're you're at a significant run rate and um I imagine have a good valuation and uh and it when did when did ai become a more significant part of the product right because initially you were doing this hinn you know white glove service configuration for these businesses taking that feedback building more self-service into it at what point do you would you say you started to significantly leverage AI?
• 26:19
  SPEAKER_02: I mean you did process automation early on but at what point did you did you really begin to leverage AI?
• 26:26
  SPEAKER_01: Great question.
• 26:27
  SPEAKER_01: As I mentioned we started investing heavily in 2017 but in beta as in labs none of those were good enough to go live um so even though we we started investing early I don't want to overstate it that was not actually production worthy AI I learned a lot of ways not to build the light bulb it was probably 2021 um when things start to really take off uh we were able to move fast because of all those learnings in the early days we could take designs right off the shelf we already had you know a lot of thoughts embedded into the company on how this could work so we were able to move fast but it was really on the back of GPT 3.0 that things start to click and we could go from trying to use our own in-house models to using something better than what we had been able to build.
• 27:15
  SPEAKER_02: And so there was really a lot of failures that you built that on and and all of those lessons learned from and when did you say you first started to experiment with large language models?
• 27:25
  SPEAKER_02: Did you say that was uh 2017?
• 27:27
  SPEAKER_01: Yeah.
• 27:28
  SPEAKER_02: Okay.
• 27:29
  SPEAKER_01: And not just large large language models.
• 27:31
  SPEAKER_01: I'm talking all machine learning.
• 27:33
  SPEAKER_01: Yeah we had a lot of ideas we were trying to build dashboards that would automatically configure to show you whatever you might need at that point in time uh based off the data and your preferences and your behaviors.
• 27:45
  SPEAKER_01: So if you do your invoicing every night when you pull Java up at night it would just show you know all your invoicing information if you happen to be more of the type of person who sends out your invoices first thing in the morning that's when those would suddenly show up.
• 27:56
  SPEAKER_01: But we try we were trying all kinds of really fun experiments to understand.
• 28:01
  SPEAKER_01: And as I said they were good but they weren't good enough to overcome the hurdle.
• 28:06
  SPEAKER_01: Right.
• 28:06
  SPEAKER_02: Interesting and and so when you were trying those early experiments what were the things that were most problematic for you and why didn't they work?
• 28:15
  SPEAKER_01: Was it what tell me about where where the failures were and what the root cause was probably a couple different categories.
• 28:23
  SPEAKER_01: One was just reliability.
• 28:25
  SPEAKER_01: As I mentioned our customers are more than happy to adopt technology that will reliably save them time and save them money.
• 28:32
  SPEAKER_01: They want predictability they want confidence so if something's really impressive 80% of the time you know okay 10% of the time and messing up 10% of the time they are not going to adopt it for their business and we just couldn't get the reliability high enough on some of our features.
• 28:44
  SPEAKER_01: Also right now people are becoming a little more open just to the concepts of AI like they're self-driving cars on the roads there's AI all around us.
• 28:53
  SPEAKER_01: So it's more open.
• 28:54
  SPEAKER_01: But back then it was like felt very sci fi ish.
• 28:59
  SPEAKER_01: So there was just the market wasn't ready for it as well.
• 29:02
  SPEAKER_01: Right.
• 29:03
  SPEAKER_02: That's so interesting yeah yeah and um and so uh how how large is the organization now how many uh how many folks are you?
• 29:12
  SPEAKER_02: 1100 employees eleven hundred and um what percentage of those are kind of the sales marketing business development versus um you know research and development product development and and so on we're we're about a third RD.
• 29:24
  SPEAKER_02: A third RD okay and the rest are building the market that's interesting.
• 29:29
  SPEAKER_02: Exactly yeah how many salespeople do you have you know I'd have to check that number okay I'm just curious yeah because I again uh I would imagine that so much of this is is based on digital marketing and there's a lot of automation there but you still have a lot of humans who are reaching out to humans right yes even though not in the outreach from an outbound perspective as much but when people are coming in the door we still want to surprise them.
• 29:55
  SPEAKER_01: They're not signing up for the product expecting to have to talk to humans and they certainly don't have to but we like to offer still like hey if you have any questions and concerns like why don't we set you up for a demo talk about how this can like impact your business we like to go that extra mile and provide that white glove service like again selfishly it does help the conversion rates it does help the retention if they sign up on their own maybe they're using eight of the 10 features they're paying for if we help talk them through it and give them that extra push we can get them using all 10 of the 10 features they're paying for and that's great for them and great for us.
• 30:29
  SPEAKER_01: So we do still have a good sized sales team that we use to sort of go above and beyond on the service front.
• 30:35
  SPEAKER_02: Your customer service customer support customer engagement and then they can upsell to more features in the product?
• 30:42
  SPEAKER_02: They certainly can we try and take a consultative mindset not an upsell mindset but yes it turns out when you're talking to them about their problems there are usually ways we can help that they haven't even considered and of course they do end up uh as I said engaging more broadly with the full platform right very interesting so so what are your big lessons learned so I have a couple of questions that I would like to explore is kind of what are your big lessons learned from this journey and then I I want to talk about how those lessons learned can apply to mid-size and larger organizations because they'll have teams of people that are smaller that are analogous to you know your smaller uh businesses so so let's let's start with that what are your big lessons learned I think as I mentioned you got to crawl before you walk before you run we we only got partway through our AI uh story but there's still layers of functionality and value that we've brought on top it's you we couldn't have jumped to the end state we had to go through each one of those progressions making it systematically better and I think you know there's this real desire to start at the end right but if you're built on a weak foundation it's not going to accomplish your vision.
• 31:51
  SPEAKER_01: The next big lesson I think is becoming well documented in the AI world but it's going to take you a lot longer and be more expensive than you think.
• 31:59
  SPEAKER_01: It's very easy you can build an 80% version in 5% of the time and a 90% version in 150% of the time but real world customers they need a 95 or 100% solution like it has to work.
• 32:12
  SPEAKER_01: That's like the key thing you need to build that trust it takes a long time to build trust very easy to lose trust so expect to be spilt spending a lot of time and a lot of money to get these things from a really impressive two-day hackathon into something that you can proudly go out and sell to your customers.
• 32:31
  SPEAKER_02: Right right right what are the other pieces you said we didn't talk about the full AI journey what are the other high points of that journey that that we should talk about well it's we only got through the phase of the journey that was one way.
• 32:43
  SPEAKER_01: It was still a bot it understood the industry the benchmarks your business best practices and how Jobber works but it was only giving you information.
• 32:52
  SPEAKER_01: That was the point where we decided it was time to start bringing agent of technology in getting the AI to start performing actions in your account on your behalf.
• 33:01
  SPEAKER_01: We also want to start bringing voice technology in so the next layer we wrapped around our Jobber AI was we started in our with our AI receptionist.
• 33:11
  SPEAKER_01: This seemed like a perfect case study because of three things one very valuable for our customers 76% of phone calls come in after hours 33% of phone calls go unanswered and as you know first to contact first to contract.
• 33:24
  SPEAKER_01: If you don't pick up the phone they're going to the very next number on Google and calling that number so it was very valuable to our customers it forced us to develop strong, robust voice technology and we were sort of operating in a safe sandbox because it was only doing things that we already let people do through the web portal.
• 33:43
  SPEAKER_01: Homeowners could already log in, book new work, request services, cancel things so the potential for the AI to go do something unexpected was limited.
• 33:52
  SPEAKER_01: We brought that feature to bear it's been fantastic.
• 33:54
  SPEAKER_01: We've done over 3000 phone calls now.
• 33:57
  SPEAKER_01: When was that introduced?
• 33:58
  SPEAKER_01: That would have been maybe a year ago now even though it only it was in beta for more than six months uh but that's been a fantastic product for our customers from there we then had voice technology figured out we had agentive it could start to take actions so we brought it into the core product we started developing teaching it how to work with our API so that could take any action in your account this required building a lot more robust audit logging and safeguards into our core foundation and we brought the voice in now you can you know hey jobber I need to prepare a new quote and as you're walking around the house it's like all right I see four walls we're gonna need to trim that door got three windowsills and you'll the quote can be game built and put together right in front of you.
• 34:41
  SPEAKER_01: So now it's you can just have a conversation with it.
• 34:43
  SPEAKER_01: It can take actions on your behalf you're still in the room though like you're it's doing it right in front of you.
• 34:49
  SPEAKER_01: Right.
• 34:49
  SPEAKER_01: So that takes us to the stage of the journey we're on now is how do we build more confidence in it running these automations when you're not around right if it has the ability to take actions in front of you sometimes it should be able to take those actions autonomously when you're not around and we're having some really fun challenges in that space right now.
• 35:07
  SPEAKER_01: Tell me about those tell me about some of those challenges key word is trust.
• 35:10
  SPEAKER_01: How do you you know as I said you got to build it so step one might be as you're preparing quotes it's giving helpful suggestions.
• 35:18
  SPEAKER_01: After you've accepted enough of those suggestions go, hey for jobs like this do you just want me to draft the quote automatically for you you're still going to review you're still going to hit send but I can draft this for you then after four or five times in a row if you accepting with no edits I'm a draft a quote for quotes like this you just want me to send it for you?
• 35:35
  SPEAKER_01: Yeah right right these are typical yeah yeah so it's in context building trust very much like you'd have with a human on day one when you hire a human they're not just okay go run my business right you're you're reviewing everything they're doing.
• 35:48
  SPEAKER_02: Exactly and bit by bit you sort of give them the gr the green light to take it to the next level and we've had to build some of those patterns into job so you you know if you hired a new estimator you would look at every estimate that goes out the door right you don't want them doing something you know winning the deal so that you know they get a commission or whatever it might be and then you running the job at a loss right so you need to make sure that you follow on what what they're doing review what they're doing approve what they're doing and then as they get better and better and you build that trust then you can let them go ahead and take care of that.
• 36:20
  SPEAKER_02: You just do give a quick glance and say oh yeah all right we can do that.
• 36:24
  SPEAKER_02: What other things are you trying to do from an agentic perspective and what are the big challenges there?
• 36:29
  SPEAKER_02: Like what what does that landscape look like not have a pun intended but what what does that look like in terms of your your world?
• 36:36
  SPEAKER_01: It's something I'm finding AI is bringing a lot of value into the space is traditional automations are good for the happy path.
• 36:43
  SPEAKER_01: When everything's going according to plan it's very rare in the real world that everything goes according to plan.
• 36:49
  SPEAKER_01: Sure.
• 36:49
  SPEAKER_01: You get raindies and sick employees and clients not answering their door andor things being asked last minute and contractors not showing subcontractors not showing up.
• 37:00
  SPEAKER_02: Yeah exactly part not the wrong part being ordered whatever it might be.
• 37:04
  SPEAKER_01: So I find LLMs are bringing a level of flexibility and resilience to adapt to the real world situations that occur.
• 37:13
  SPEAKER_01: I think that's some of the really big strengths the other category I'd say is common sense right again our goal is not to stick a lot of chat bots in front of our customers it's to make Jobber just magically always seem like it knows what you need when you need it.
• 37:27
  SPEAKER_01: We want to measure our success by how little we're sticking in front of our customers.
• 37:32
  SPEAKER_01: But LLMs unlock some new opportunities to do this.
• 37:35
  SPEAKER_01: Let's take a super contrived simple example of a plumbing company two work requests can come in the front door two calls.
• 37:42
  SPEAKER_01: One's help my basement is flooding and the other is I need you to come by and prepare an estimate for a kitchen reno.
• 37:49
  SPEAKER_01: It's very easy to find when's the next available spot in my calendar it's very easy to find out when am I next in the area.
• 37:54
  SPEAKER_01: The hard part is knowing when to do which decision like find this next opening in my calendar algorithmically that's been around since the beginning of Jobber but making it so when you respond to that problem we can just show Susan can be there in 20 minutes and a go button and nothing else.
• 38:13
  SPEAKER_01: That's magic, right?
• 38:14
  SPEAKER_01: So I think LLMs, we're just at the beginning of this journey there's so long way to go but every single decision that we ask our customers we should be starting by going do we need to ask them or is there a common sense based off how they've made decisions in the past, what we know about them can we short circuit the decision making process and turn them into just a big approved rubber stamp recommendations yeah yeah so I see a lot of the value of LLMs is not that we can put new interfaces in front of them but all the interface we can remove from them and its ability to look at the context and make some pretty reasonable decisions.
• 38:49
  SPEAKER_02: Right.
• 38:49
  SPEAKER_02: Many times when I'm given choices by an LLM I'll say what is your what what's your uh what do you recommend based on on my goals in your judgment?
• 38:59
  SPEAKER_02: Right?
• 38:59
  SPEAKER_02: And it'll say well I recommend you do BA.
• 39:01
  SPEAKER_02: Okay, do Bob so that's interesting.
• 39:03
  SPEAKER_02: So that's fascinating.
• 39:04
  SPEAKER_02: And it starts now are you I imagine you're training your own models or starting with uh open source or or are you using uh public models?
• 39:12
  SPEAKER_01: How are you how are you uh evolving that it's we have a pretty sophisticated architecture.
• 39:17
  SPEAKER_01: We are using public models often with fine tuning often a pretty complicated orchestration layer a typical request to job or AI spends up more than a dozen agents working in parallel to try and address that problem.
• 39:31
  SPEAKER_01: I did a quick audit the other day we had um 11 distinct models with different you know fine tunings and levels because a human brain we got our reasoning center we got our analytics center a human brain we have different ways of thinking different parts of our brain that are optimized to solve different types of problems that's where we're at with Job or AI as well so it's a very robust architecture.
• 39:54
  SPEAKER_02: That's fascinating but we we find that in doing some of the work we do around document processing and componentization and entity extraction and and processing images and tables we use about six different language models and uh and a variety of algorithms in order to process PDFs, right?
• 40:12
  SPEAKER_02: Because no one size will fit all.
• 40:14
  SPEAKER_02: So you're doing the same thing with specialized algorithms to be able to handle these different parts of the problem uh and that's fascinating but then I think that's where you know organizations are evolving realizing that no one size fits all.
• 40:26
  SPEAKER_02: And then you have to do that orchestration of those different uh models in those different regions.
• 40:31
  SPEAKER_01: But again the orchestration is so important for this customer base because they you know they don't want to waste their time trying to learn AI, figure out all the different ways see what works see what doesn't that's our job.
• 40:44
  SPEAKER_01: It's our job to to democratize access to this technology so that they can focus on running their businesses.
• 40:49
  SPEAKER_01: And that's why we put so much time in that orchestration layer.
• 40:51
  SPEAKER_01: If done well, yeah it should be invisible feel like magic.
• 40:56
  SPEAKER_01: Right.
• 40:56
  SPEAKER_02: And I feel like a trusted advisor or a a key employee who has knowledge of the processes and the customers and the industry and the quotes and and uh and the best practices and all of those things right and that's what we're trying to make invisible is is what uh what the behind the scenes is and and and get to the result that these uh agents and these agentic technologies are providing what what is there's so many questions I could continue to ask you because this is such a fascinating journey and we'll probably have to do another session.
• 41:27
  SPEAKER_02: First of all, what what are the what's the largest but what are the larger customers uh how many employees or revenue or whatever that you have as you're saying you're small mostly small to mid-sized business but what's the higher end of of those customers?
• 41:41
  SPEAKER_01: A higher end will be sort of 50 to 100 employees like we but that is the upper end it's uh we also we do support franchises where it's you know 200 locations but in my mind that's 200 individual businesses because that's how they operate like they each have that workflow.
• 41:58
  SPEAKER_01: We of course support a reporting layer so you can see see all the details in aggregate but each location uh I tend to view as its own business.
• 42:07
  SPEAKER_01: Yep yep no that's great.
• 42:08
  SPEAKER_02: And then what would you say is the are some of the lessons learned if uh for uh the enterprise for larger businesses you know who will have smaller teams you know focused on certain areas will they have business units or have departments what what do you think the is the the what are the transferable lessons learned and I know you gave us some of them already you know you talked about listen to your customer understand the process look at the end state uh keep the end state in mind no one size fits all in terms of these uh AI models and LLMs and and and your journey and your path you know it was a long journey and a long path in order to get to this level of sophistication and people need to understand that you have to go through all of these steps in order to get there and that it does take an evolution what are some of the bigger lessons learned that you would say would apply to larger uh enterprises it's probably two internal changes we made that I think have really helped one is we have a Slack channel that anybody can join uh just called Social AI where cool news, new toys, people hacking on things in their free time like anything related to AI, people are dumping in that channel.
• 43:15
  SPEAKER_01: There are a lot of enthusiasts throughout every company right now and giving them a chance for learnings to spread rapidly as you said there are little silos everywhere.
• 43:22
  SPEAKER_01: How do you break those silos open and create the shared learning is creating those fun spots for these enthusiasts like it's not something they should think is like oh it's just me and my own time right like bring that into the open so that everybody can be sharing.
• 43:36
  SPEAKER_01: And are you doing this across your customers so this is public?
• 43:39
  SPEAKER_01: Oh no sorry this is internal to our internal organization.
• 43:42
  SPEAKER_01: Okay the other thing from an enterprise perspective uh is we really fast tracked pieces of the procurement process for AI experimentation.
• 43:52
  SPEAKER_01: There's so many new tools coming out all the time.
• 43:56
  SPEAKER_01: You know 80 or 90% of them are probably not going to live up to their hype.
• 44:00
  SPEAKER_01: Of course but some of them are going to be game changing for the organization.
• 44:04
  SPEAKER_01: You know every every organization has a procurement process for a reason.
• 44:08
  SPEAKER_01: There are a lot of steps from you know security and budget and approvals and everything.
• 44:13
  SPEAKER_01: But people weren't able to just like it wasn't well set up for I want to just play with this one tool for one month.
• 44:19
  SPEAKER_01: Right, right so uh so I actually created my own budget I got a little envelope of virtual cash set aside that I just approve tools directly and we have just a a two-step procurement process.
• 44:32
  SPEAKER_01: Nice any tool that any AI tool that you want to experiment with security just has to sign off and then I'll sign off and it'll come out of my personal uh budget for up to three months.
• 44:43
  SPEAKER_01: So anybody can play with any tool for up to three months uh and they don't have to go through the whole departmental budget approval process.
• 44:51
  SPEAKER_01: Nice and if it's if it's working well for them then they still have to go through the slower process but now we know yeah and that's really unlocked a lot of internal experimentation.
• 45:01
  SPEAKER_02: Tremendous that's tremendous I we're right up at the the top of the R and I do have a hard stop I wish I had a little more time to spend with you but um this has been fantastic.
• 45:10
  SPEAKER_02: What's down the road?
• 45:11
  SPEAKER_02: What's the future what what are your what are your aspirations?
• 45:14
  SPEAKER_02: Jobber will be your most trusted employee.
• 45:16
  SPEAKER_02: Yeah okay that's really interesting I just listened to a podcast called Shell Game which is about uh a journalist who tried to create a company with just AI agents and it's fascinating it's really gets into some strange spaces and and uh but it it's fascinating I understand they're they're anthropomorphizing things too much and they're they're trying to infer intent and and all of this stuff but it's a fascinating listen so you might want to check that out um uh I can send you the link and I also um uh would love to stay in touch with you this has been fantastic thank you so much for your time uh this has really been this has really been uh great so force pleasure is all mine thank you so much you know force thank I really appreciate the conversation and you're sharing all these lessons learned and these perspectives you know on how AI can streamline you know businesses and small businesses and your especially your journey you know and deliver those you know those those improved customer experiences but you know if it's this is so perfect for those industries because they're not going to be replaced by AI, right?
• 46:18
  SPEAKER_02: And AI is going to be so instrumental in how they continue and how they grow.
• 46:23
  SPEAKER_02: And to our listeners thanks very much for listening in and joining the conversation about how AI is reshaping work uh how you can make decisions how you can create value across all industries using AI and using uh these these new and emerging technologies there's so much more we could talk about force but uh thank you for today and thank you for your time and it's been really great uh meeting you and speaking with you.
• 46:48
  SPEAKER_01: Thank you for having me.
• 46:49
  SPEAKER_00: It's been a pleasure thank you for joining us on another deep dive into AI innovation.
• 46:55
  SPEAKER_00: Tune in next time when we introduce another industry expert and discuss how to maximize AI in your world.
• 47:01
  SPEAKER_00: The Early AI Podcast is sponsored by Early Information Science and Vector.
• 47:06
  SPEAKER_00: To learn more visit early dot com that's ear e y dot com.
• 47:12
  SPEAKER_00: Thank you for tuning in