Erik: Welcome to the Industrial IoT Spotlight, your number one spot for insight from industrial IoT thought leaders who are transforming businesses today with your host, Erik Walenza.
Welcome back to the Industrial IoT Spotlight podcast. I'm your host, Erik Walenza, CEO of IoT ONE, the consultancy that helps companies create value from data to accelerate growth. And our guest today is John Mickelssn, CEO of Krista. Krista is a nothing-like code Intelligent automation platform that builds and integrates machine learning and apps so you can optimize business outcomes. In this talk, we discussed how conversational AI can transform business processes to enable knowledge workers to spend less time doing routine processes and more time solving problems. We also explored the problem of proliferation of apps and user interfaces, which require constant learning from your teams and are often either used improperly or often not at all.
If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend a speaker, please email us at team@IoTone.com. Finally, if you have an IoT research strategy or training initiative that you'd like to discuss, you can email me directly at erik.walenza@IoTone.com. Thank you. John, thanks for joining us today.
John: Glad to be here.
Erik: So yeah, I want to understand a little bit more about the founding of Krista. Before we get into the business, I think it's an interesting topic for us because often we're looking at the front end of the ops, we're looking at the factory floor and so forth. And you're really about changing how back office or enterprise operations are done, not just from a technical standpoint, but also rethinking how processes should be done, when the technology enables processes to be done in a different way, if I understand So, maybe if we can start with you just walking us through the founding of Krista and how you identified this as a problem that needs to be solved. Because I think, many of the companies that we talked to are very much focused on automating tools to automate processes as they exist today. My understanding is that it's quite different from your approach.
John: That's true, Erik. And in fact, our focus is everybody in their organization, even all the way to the shop floor. So I'm minding my own business, enjoying a practically retired lifestyle. I've had a number of successful startups and exits. I've got 20 patents. I've started so many companies and I’ve lose count. I've had a really wonderful career, and I'm mostly helping others getting their thing going along. And then this thing keeps gnawing at me that we haven't solved this fundamental problem of how do we really create proper human computer interaction?
We've all seen Star Trek or the equivalent SciFi movie. We all know that's where this all goes. Why isn't anybody taking us there? And when you look at what we are building in terms of automation platforms, and enterprise apps, and where's it going, a lot of that just looks like it's going the wrong way, Erik. So that starts me to thinking, okay, well, what would you do to get going in the right way? And there, you there goes, now I'm stuck. Because now the creative juices start flowing, you start trying to figure out what would that be like. When you dissect that problem and you try to really work out, how am I going to make it possible for people who evolve over Eons going to embrace an even faster and ever faster changing technology landscape that evolves over months, weeks, days, minutes? How are we going to ever make that happen? And that created the attention that made me need to build something, prove out the feasibility of it.
That thing we call Krista, we launched to the market a year ago. We're having a tremendous success, because, frankly, it's a fundamental problem every business has. And so that was the motivation. We get it that the way that we should be interacting with computers is to make them do the job. In fact, we wrote a mission statement for the company to make technology that understands people. I'd submit to you that you have never worked with a company whose mission statement was that. It was always some form of making technology you can eventually figure out if you take enough training or if you learn our jargon, which is entirely not the goal. The goal is for the machines to do the work and us to do less, for them to figure us out, not for us to figure them out.
And so with that as a basis, again, that that just became the motivation to figure out okay, well, if that's true, if it's what we're going to do, how would you do that? What would that really need? And how would you deliver that? And so that's what caused the product. And that's what got me involved and going.
Erik: This morning, I'm sitting here in Shanghai, I was on the call with the bank because my bank account just stopped working, which happens every couple months. And the process is as painful as it was 20 years ago, really, it feels like it has almost not changed. The interface maybe changed a little bit, the voice is a bit smoother, but fundamentally, the process is unchanged. And then another question, why? This is a Bank of America. This is an enormous company that's throwing a lot of resources at this problem. So why haven't we solved this problem or made more steps forward than we have in these areas?
John: Because if you look at automotive and I use the anchor of January of 1990 because that's when I actually entered the workforce as a developer, and I started my first company. At the times, like wow, and I had this beat up 76 Volkswagen, but I was behind it more than I was in it; it was just this rundown clunker. And you look at a Tesla today or you look at some of the beautiful automotive technology that's out there and then you look at what are we doing? I did a screenshot of windows back in January 1990, you would never look at it, compared to today.
But you know what? The tools that it uses to build enterprise apps and to develop automation are indistinguishable from the stuff we were using in January 1990. And I proved this in webinars all the time. I show screenshots of products from knowledge where IP owed in the mid-80s, I put them right beside current automation tools, even practitioners can tell the difference. They even claimed the same features: point and click, data modeling, visual screen designer, flowchart to code. You've just drag all the commands from a pallet onto a screen, you wire them all up, you write your scripts, and then you hit a button that’s generate the app.
I was doing that stuff back in early 90s. I had to throw all that stuff out because that generated code never worked. We've been recycling this idea. We call it Case Tools back then. We call it RAD for a while now. We call it automation platforms. It's essentially the same thing. We are not innovating in the platforms that are what we automate processes with. That's the fundamental statement, Erik. And because of that, we're not getting that innovation that we are in almost every other field. That's got to change.
Erik: So what do you think is the bottleneck here? There's clearly resources being thrown at this problem? So it's not a lack of resources. Is it just the stakeholder dynamics of how you sell something to a procurement department and they have a checklist and that prevents? Why is it that we don't see the innovation here? Because we clearly have the money, we have the brains that involved in this market.
John: I actually asked myself, I thought, surely there's something here that's just impenetrable because there's so much innovation everywhere else. John, you're claiming that we're not getting anywhere. And you know where we're trying to go is the whole Star Trek enterprise, you're on the deck of the enterprise and here you are talking with computers and Androids and people the same way, we all kind of get that as a model.
So surely, we're either making progress toward that. Or if we're not, it's for a really good reason. I have a feeling it has to do with the fact that we don't yet think about users appropriately. And I'll give you a really good example. When I first entered the world as a developer, I'm out of uni and I'm building stuff and I'm delivering it to customers. And they're saying, this is really complicated. I really don't want to do this. I'd rather do it the old way. We were taught to call them stupid or lazy. Don't worry about what the users think. Just they have to do it. It's their job. We fire them and we replace them if they don't do it. It's not a matter of inside IT. This isn't about if they don't like it, they don't have to use it and they'll go elsewhere. Well, they’re captive audience. So innovation is not necessary inside an IT shop. And I don't mean to be rude because these people are trying to do their best work.
The problem, though, is there's no alternative, generally speaking. So they end up with a slower innovation. They end up with the whole notion of consumerization of IT. We talked about that a lot about 10 years ago. You know IT pretty much snuff that out. An enterprise IT still doesn't look anything like consumerization of IT, does it? We've just completely like, no, no, no, no, no, no, guys, that takes way too much thinking on our part. We're just going to keep going this way.
And the fact that we currently think and this is a second reason altogether, there is a very consistent motion that software companies have. And of course, IT departments when they build their own apps, they think you need a capability, I’ll build an app with a UI, I'll give it to you. Do you need another capability? I'll build that app. I'll build a UI for it. I'll give that to you. Now, we're overwhelming the desktop with 50 apps. What's the goal? Get to 70, 100, 120? What are we doing? We have not thought about the final assembly step. How do I compose all of IT’s capabilities in such a way that users and AI can participate in that? And that's a natural evolutionary problem.
We started with just a handful of apps. We've added a few more and a few more and a few more. And now we're the frog in the boiling pot. We haven't thought wait, I need to stop, I need to think this is not the motion that's going to end up working, we're all going to pass from this. So I'm going to stop right now. I'm going to realize I need to do this differently. And instead, we just keep, I hear this all the time. I can't hire people into my organization. I've got a seven page long job requisition. They've got to know 16 different products. No one on the planet does. What am I supposed to do?
So we have to think differently because it was just two or three 20 years ago. It wasn't that bad. But the evolution has not caused anyone to stop and say, rethink, this is the trend that doesn't work. What are we going to do differently? And that's fundamental to why we thought the way we did. We took the consumer perspective, make technology that understands people, people don't want yet another app, even though IT is goal is to change faster. Isn't that funny? Isn't the whole low code application development market? There's a whole market full of products that are designed to make it faster for you to build new user interface based apps and get them to your business users faster. Business users are not asking for more user interfaces that are IT ugly, and yet another thing they got to figure out what to do.
So long-story short, Erik, those are the two challenges. First, it's a lack of innovation because there’s no competition, honestly. And then second, we evolved from a place where it wasn't like there were 100 apps on the desktop in 1990, but clearly, we're heading that way now if we don't do something different.
Erik: That's a great point. I'm involved in a lot of projects through our work at IoTONE. And I'd say very consistently, the project team is focused on understanding feature requirements, technical requirements, not at all thinking about the human because you assume this is a professional, this is a trained engineer, this is an IT professional, this is whatever. But these professionals are humans that are being overwhelmed with the complexity that's being thrown at them. I think you have a complex chain of users here. Because whoever is going to be using your software, and then they're going to be building potentially for somebody else who provides requirements or maybe it's the person that has the requirements and builds for themselves and then there's going to be potentially some end user like me on the phone with my bank trying to use this system to get a job done. What does it typically look like?
John: Yeah, three primary roles. We'll start from the top, if you will, and that is the end user. And we try not even to call them users. We try to just call them people. Because user doesn't sound human anymore. If you ask people for things, you inform people of things, people are involved. So you don't get to say once you're thoroughly trained or once you learn all our jargon or because you must otherwise your job won't keep you, people are the target of our solution. And they interact with a conversational style interface.
What we've done is we've realized that humans actually do embrace change. There's the crazy thing, as long as it feels like more of the same. So Erik, you and I are having a conversation on a topic, we did no training to do. You'll have conversations with dozens of other very interesting people on our dozens of other topics. No training required. Conversation is the human means by which we embrace change.
So when you deliver structured rigid, full of jargon you don't understand, abstractions that you don't understand, you’re wise, you create an impediment to people embracing them. But when you tell someone, hey, by the way, you now do all this new stuff in Krista, they say, oh, yeah, how does it work? Oh, don't worry about it. It's like WhatsApp. Oh, I've never thought about needing to go to training for WhatsApp. I just have conversations on WhatsApp. I actually read what people are saying, I say things that I'm interested in saying and it just works. That's the model for dealing with people.
The second of the three important roles in Krista is the author. This is a super important role. This is the business person that understands the outcomes that need to be automated, need to be better digitized, to the phrase we use all the time, need to orchestrate the activities of people, applications and AI. They're the important people. They're the ones we love. Because the target for them is non-developer. We don't want these people to have to write code to make changes. That is the single biggest challenge.
When a change to an automation requires a coding change, it means it sits on a backlog, it has to get prioritized. It can't keep up with the business. There's a phrase we say often that IT cannot keep up with the mouth of the business. It's just impossible. The business can say it faster than you can code it, never will they keep up. So the business is always in a deficit of the velocity they are looking for always behind. Because IT is the friction, we have to separate the automation, the rules, the steps, who needs to know what, the what systems are involved, all of that, the actual business process or outcome. Separate that from the coding that is required, that's what authors do.
Authors describe, literally, in Krista, like a conversation among people systems and AI. They literally use constructs like ask a system to get all the orders that must be fulfilled this week. For each of those orders, ask our supply chain system do we have that inventory? For if we don't have inventory for this particular item, ask of a purchasing manager, can you source that additional quantity in time? If they can't, do this. If they can, do this.
We literally give authors the ability to describe a business outcome as a conversation among people, the apps that they have access to and of course, AI to make decisions. Like we need to move more product, what would be the optimal discount that maximizes our profit and yet gets us to move the product? Well, that's usually like a 25% if it's a human. But if it's Krista, Krista has been studying every promotion you've ever done and is calculating a value based on all of that historical understanding, far more insightful than a human can be. So, the orchestration of people, system capabilities, apps, and AI.
Now the third of these of these roles is the developer, because there's no way around it. If you're going to talk to systems, you got to write code. That's how they communicate. What we've done to make this so much more valuable is we've isolated it to just the connectivity layer. Those authors are making NLP or human style requests, we personify the back end systems. You heard me say, ask the supply chain system, ask the inventory management system. You literally describe it in human language. And that maps to what the capabilities of that particular system are.
So someone has to write the code that understands those requests that humans will make and makes the actual transactions if you will, in the back end. It's necessary. But the good news is you build it once, you can build hundreds of automations with that system. You bring in a new system, you teach Krista how to talk to that system once, hundreds of automations occur. So that's the goal. And those are the three primary roles.
Erik: So I want to get into more what the tech stack behind Krista looks like how it interfaces with other solutions. But before we go there, let's quickly cover who you're working with. So this is a very horizontal problem. So I imagine you could basically be working with any organization in the world, but everybody needs to focus somehow. So who do you typically work with? Is it particular industries, large, medium, small organizations? Are there particular functions within the organizations that you tend to focus on? Where was the 80% here?
John: Sure. So the current state of our business is because we've been just over a year, in fact, yesterday was our one year anniversary in the market. We were building software for years. This is not a small platform. But we actually launched on the 21st of February last year and we've had a great run in put dozens of customers on the product and all kinds of good stuff is happening. Those customers are in a wide variety of industries. We do focus more horizontally than we do vertically, whether it's the finance department, the sales or field organization, cybersecurity organization, and even special HR, all of the ops areas.
In fact, there's a phrase that we use. And we're actually taught this by customer. Give me any large body of low tech people dealing with IT’s high tech stuff. That's a perfect example of where Krista can do, a ton of good and then of course, owning an outcome end-to-end. I've got to figure out a way to get this particular process that maybe takes hours, weeks, months, who knows, I've got to find a way to optimize that and to more effectively deliver that outcome. That's the other primary use case. So in those areas and in others, customers have taken us all kinds of directions. And so it's more horizontally focused, for sure.
What we have done to verticalize our capability is we work with partners. And every one of our customers is obviously in a particular industry, and the partners that they use to do implementation work now, like in manufacturing, it might be one of the big global SIs that has a really strong practice in that particular endeavor, they have that customer intimacy that we don't. They understand that much better than we do.
A really good example recently is one of the larger telcos in the world has adopted Krista to do a lot of work in their executive team, to help them with insights into their 5G activations, the mix of their products, which stores are doing well, which promotions are doing, all that stuff. And we've now gone into their call center to help them with fraud, to help all of these more complex transactions. There's tons of things that call center deals with that are simple question, answer, anybody can figure it out, the agents don't have in trouble.
But then there are some of these transactions you do with the telco, they're a pain in the neck. Maybe your example with unlocking your account at the bank is a good example of that, where there's so much tech complexity involved in that and those agents are so ill-equipped to handle that complexity that Krista needs to own that outcome and maybe do a better job. So that's the kind of stuff that we end up doing. And partners help us with those very verticalized domain specific examples. So we work a lot with and through partners.
Erik: Those are already two very different scenarios. One year, you're serving an executive team, which probably has a lot of rapidly changing requirements and requests, spanning much of the business. And then on the other hand, you have a process, which might be quite complicated but is a fairly well defined process with particular inputs and outputs. Fundamentally, it's the same architecture servicing both of these, is that right?
John: That's right. Just go back to those three different actors, if you will. They're all just people using a conversational interface. There is an author that is easily describing the outcomes that are desired as a conversation among the people, the apps and the AI that's available. And they're connecting the systems through the extensions that have been developed already, for that customer, either by us or by the customer, or of course, by a partner. So it actually fits that architecture quite well.
Erik: Well, let's get into the architecture of it here. So you have a collaboration platform. You're integrating with existing systems. You also have AI. So I'm curious whether the AI is embedded in crystal or whether you're leveraging also external platforms may be selected by the customer there. So can you just break down? What is the architecture and then what within that is Krista code and where's Krista integrating with the existing systems?
John: So at that top layer for people, we provide our own conversational interface. But even more often, our customers want us to use Slack or Teams or email or text messaging or Facebook Messenger, or you name it. And those are absolutely equally useful because we want to make this as approachable as possible for people. And while our user interface might be richer in certain ways, what you're currently using is much more important than a little bit better.
So our architecture is based on the idea that people communicate through channels. We provide channels. We also provide native mobile apps and web apps and all that stuff that you can use natively. But also again, we work through third party channels as well, because the goal is minimize the amount of change, minimize the complexity, integrate Krista into the way people already communicate.
The way that Krista integrates back end systems through APIs, if they're available, through record and replay of UIs, if that's necessary, whatever is required given the particular system that's involved, our AI stack is actually quite interesting. And we call ourselves an intelligent automation platform. Now that's a bit of a misnomer for most people who call themselves that because they really automation platforms that are bolting on artificial intelligence.
We are an incredibly modern platform. I mean, we were built three years ago. Most of the stuff we're compared to, as I said a minute ago, are 30 plus years old. And even the major RPA players are literally 15 year old companies, they are not a modern platform. AI was completely incapable of doing any of the stuff that it's doing today back then. So AI was an afterthought bolted on. Even users, actually people were a bolted on afterthought. So we're very, very different architecturally.
The way we integrate AI is three different ways. But it's actually pretty simple to think about. There's a spectrum of machine learning models that can be built by any company. On the simpler side, maybe a fully a third of them or more are relatively simple datasets, relatively simple algorithms. Can I expedite this order? What is my likely quantity of this particular product? Do I have enough work hours in my current team to cover this particular machine? Those kinds of things that are relatively stable data, relatively simple, Krista builds those things internally with no data science jargon required, no coding required, just literally builds it all quite simply and elegantly.
On other end of the spectrum, there are very complex models that need to be built. And these are generally for very complex datasets where feature engineering is required and hyper parameters and all of the data science jargon that you would not want me to bore our audience with. But those are very important models as well obviously.
Krista can't build those in an automatic way. But Krista has a very elegant way to integrate those, just like we integrate the other systems, your ERP, your shopfloor manufacturing, your HR system, etc, we can integrate those third party machine learning models very easily.
The third way, because we are so horizontally focused, we've been identifying for many months now, common, very important ML solutions, Machine Learning solutions that we can provide to practically every company. Two of those I'll highlight. One is we call document understanding. Everyone's got thousands of contracts. Which contract is coming up for renewal right now? Which of our contracts have an automatic renewal? Which of our contracts have pricing that goes up when they automatically renew? Or are we allowed to sue so and so for our contract? Whatever that is, you want Krista to have all of that knowledge. You don't want your people to have to be trying to track all that.
And then the other is what we call issue cognition and resolution. And unbelievable, most every business has inbound communication from employees and customers, essentially issues. Can you move my delivery to Tuesday? Do you have tomatoes in stock? What's my current leave balance? Or how do I go about adding a dependent to my insurance? Or anything like this.
Those consume enormous amounts of human time right now. Most of the time, the people who receive that email aren't even the ones who know the answer to that question or can't resolve the case, whatever it is. Krista reads those emails, understands the goal, or the intent of that email, and even resolves them in many cases.
So these are what we call ML-based solutions. They do require some work with the customer. We have to see your contracts. We have to see what your inbound customer enquiries look like. We have to see what your employees call certain things. Because general AI is not going to get you there. We have to tune our general AI into your specific domain. But when we do that, we get a great result. So those are the three layers: building ML, super easy; integrating third party ML, either yours or from other companies; and then third, solutions we're building ourselves that we've been able to use across customers and continue to build out.
Erik: So, working with the customer to define what language means. I think this has been one of the big sticking points for companies. Or even do we have tomatoes in stock? But you still need to, you need to program that in, so it understands what to look for, what the request is. And then I think a lot of these companies have basically booklets of jargon that could be here's the here's the 1,000 words. And this jargon evolves over time as well. You make an acquisition. You integrate somebody, all of a sudden you have new terms you have to deal with.
So, what does this look like from an implementation standpoint? Maybe you can also give us a little bit of detail of how long would it take to go through this process? And then how do you make sure that does it evolve naturally, is it a learning system? Or is it every three months, we're going to do a review and we're going to see if we need to modify or add? Or what's the process to onboard and then to maintain the system?
John: So we are in these solutions area and in the models that Krista builds automatically and maintains for you automatically, we do what is technically called Incremental Training. And what you mentioned, Erik, is what mostly AI geeks will call an ontology. These models don't necessarily know the ontology of a particular business domain, so they have to be trained in that. And that's why there's a little upfront work. And by the way, that upfront work is usually a couple of weeks, it's not that much time.
Fetching the data is usually the challenge. Supplying us with enough data that we can properly train to get started typically requires us doing a little automation work, ironically. We've got to extract some data from systems, sometimes extract some data from people's heads. And when doing so, we get a good start.
Now, the most important challenge that has caused AI, by the way I'm citing a study, AI worldwide return on investment so far 1.3%. Ridiculous. And yet every company that's listening to this podcast has got a ton of open requisitions for data scientists paying them much more money than the average person in the company. Now why is that? We all know there's something valuable there. We don’t have no idea how to get it out. Krista is a great vehicle for getting it out.
And one of those reasons why it's a great vehicle is because, as you described a minute ago, you cannot build this once, run it for three months, and then realize it's become crap. And then you have to completely retrain it again, and then delivered another where it's actually a little bit better. And then it goes to crap again. You can't do that. It has to constantly be incrementally continue to evolve, just like the business does.
Examples I use recently, because of the world we're in right now, if you're in the foodservice business and our new COVID regulation comes out, you're going to get 1,000 requests the very next day about that COVID regulation in your business. If your models are not capable of incrementally it's 20, or 50, or 100, or whatever it is, within a couple of 100 of those requests being observed by the software and realized, I don't understand that question. A human had to help me, now that I've seen 100 of these, I can figure this out and take care of the rest of those other 1,800 and continue to do so into the future. You are not going to get there.
The decay, right of the model will be such that you won't trust it anymore, you'll just throw the thing away. And that's the problem we have in AI right now, is trust as low. As soon as people find a reason not to trust it, they then characterize it as bad and don't use it. It's just like humans. The first time I had to deal with a new COVID regulation of wearing a mask, I was like wait, learn it, figure it out, then deal with it. The AI has to do the same thing. It has to incrementally train just like we humans do. We discover new things, the world changes on us, which takes a few times for us to figure it out and then it's normal. The AI has to do the same thing. And Krista has a really nice mechanism for doing so.
Erik: Let's choose something that's very concrete and just walk through from your first conversations with the customer, understanding their requirements through designing the solution, deploying it and then finally using it and maintaining it. What do you think would be a good case?
John: We could do any number. One of my favorites is a manufacturer of buildings supplies, these are tiles, brick, all of the materials that are basically an industrial buildings. And these guys are actually in Asia. This company has hundreds of sales reps in the field. They have hundreds of people who are doing distribution for them and all that good stuff. And of course, there are classic example, people who know the business domain very, very well.
And in fact, when I interviewed the CEO, this is now months and months ago, he said it very well. He said, look, 30% of the people currently working for me, next year, are going to think this whole selling thing is not for them. They're going to go back to lay tile. They became sales reps for me because they really liked the tile they were laying and they thought, man, really I could do this. But 30% of them are going to go back to laying tile and I have a 30% turnover in this organization.
I put CRM, customer relationship management, sales order management, HR systems expense management systems collection system in front of these people, they have no idea what I'm dealing with. They have no idea what they're dealing with. And they don't use it. They revert to WhatsApp. They find my assistants. And they just type messages, they take pictures on their phone of sales orders, and they send them in and say, look, that's all I can do boss.
I am struggling as a business to scale. I either spend weeks trying to train these folks, only for them to leave, 30% of them, every year. So I'm constantly retraining. I'm constantly threatening them that won't pay their commissions if they don't do it the right way. But of course, I have to pay their commissions because if they get the deal. So I'm constantly in this frustration, and I'm trying to force people to do something they don't want to do.
When I discovered Krista, I was able to tell people that came in, you don't train it, you just use it. You don’t think about that. There’s a cognitive thing. Think about the human embracing of change with you don't learn it, you just use it. That's such a fundamental thing. And I know I'm a geek about this stuff because I've been studying the psychology of human change for a while.
When the CEO of this company can say I can have that 30% turnover, because within a day or two of a new person coming in, they're equally as productive as the guy that left. That is a fundamentally empowering thing for that business. And because of the fact that they're able to do that, they're able to say everyone has to because it's actually easier than to do with the old way where you're trying to figure out who to WhatsApp. You just say, hey, Krista, I have a new sales order. Krista will actually say great, do you have a picture of it? That's great. No, I don't.
So here are the product numbers, and you can help me look them up. And here's the quantities. By the way, if this is really a sales order, I need a signature from the customer. Hand them my Android and let them sign, hit submit. Now Krista is handling all of that order capture and the rep has learned nothing. He's just using a conversation. He's just participating in the conversation.
But now, the benefit of that is tremendous. They get actual compliance to all of these things that they were trying to beat people up to get them to do. Now they just get it. They just use it. It just happens. And in doing so he's able to optimize his inventory levels, improve his cash flow, and he's able to reduce a ton of expense fraud, ton of expense fraud because it's just too easy for people to make certain mistakes or do things that are improper.
But when Krista says, you want to capture an expense, I'm going to grab the length and long of where you absolutely are. And I need you to take a picture of the receipt in that location. All of a sudden, inappropriate receipts are not being submitted for expenses. So you get to the point where the business is bottom line impacted. At the same time, people are dealing with technology in a much easier way.
And the last part about this that is I think very, very important from a technologist perspective, now that Krista is the way they interface with all those backend systems, you could change that CRM from A to B and your people wouldn't even know what happened. But today, the CRM, again, customer relationship management system, any XYZ, any DLA, you change from one major version to another and you're talking about upheaval, migration change management system. You're talking to the union.
And it's understandable. I got a stack of bills I need to deal with sitting on my desk. We're all humans. We just put off the things that are complicated, or we don't want to deal with. Just make it technology that understands people and they'll naturally do it. And once they'll naturally do it, you can very heighten the capabilities you can deliver. So anyway, there is a good example.
The reason this one is my favorite is actually not because that's the biggest company or the biggest deal we did. It's because it's the most clear example of how people will do what you need them to do if you embrace them as they are, not as you're trying to make them. And there's very important valuable lesson in that, that if we build technology that understands people, we will get them to do what we need them to do.
Erik: So let's say you're using Salesforce, and you have 30% of the organization that's using it properly and 30% is kind of putting in some data occasionally and 30% is like I've got my Excel and my black book and that does the trick. And honestly, that's how we work here. We have Salesforce and it's almost like a phone book, like we put data in and occasionally we'll look it up. But we use Excel a lot because it's just a bit of a painful process to go in and do it the right way. So how does this work? Is it go to a sales meeting, come out and I say Krista, I've got new contact, so it says it's okay, do you have a business card? Yeah, I've got a business card. Can you scan it? I scan it. For example, that scanning is that a third party app that is designed to read text and translated or is that Krista? Can you just walk us through this this one use case, what exactly does it look like?
John: So by the way, you're no different than every other business and most every even department of every business. The number of people who use Excel worldwide, 750 billion people use Excel at least once a month because we're all ignoring their actual system of record. We're all doing our day to day work in Excel, and then updating the system of record when necessary. And that's because we don't like those systems, they don't work the way we think.
So back to your specific example, yes, Krista can upload, hey, I have a new contact. Do you want to fill out a form? Or do you have a picture you can give me? Give me the picture. Krista has already embedded OCR capabilities. Or if you're using a third party and you want us to use that one, Krista will OCR through the third party of your choice. That is all completely hidden to you as a user. The person just says I want you to take a picture and figure this out and crystal will obviously OCR that and then prevent present you with that information to make sure that we got it right, you'll then be done. And that is easier than filling out a row on Excel.
And then when you actually want to see it as a table, you'll actually see it as a table in Krista. You can even then say I wanted an Excel, Krista will generate an Excel file for you. So that's the goal. If you want to work in Excel, keep Krista up to date and hit a button and make it an Excel whenever you want it to. You can even go and edit it in Excel and then say hey, Krista, I've made a bunch of changes, go update all of these contacts or opportunities or what have you. And those are all very simple automations that authors create, you're asking yourself to read each row. And for each of the rows, each of those is a contact, update, the contact is for that row, update the contact for that row, it's easier than you would imagine, to be able to do all of that kind of stuff.
But here's the very important thing. No one keeps their deals up to date and Salesforce. It's universal. We walk into companies we don't even know and we say your people never keep Salesforce up to date. You threaten their jobs, you refuse to pay them unless, and yet they still don't do it. Pick any one of the people that's on your mind right now, send them a text message, literally like right now send them a text message, and you tell me how long it'll take them to get back to you. 30-45 seconds. So what if we just had Krista, say, hey, Erik, where are we with the Acne deal as a text message instead? Very likely, Erik's going to reply in 30-45 seconds just like he does every other time I text him.
But when you say hey, Erik, anytime a deal changes, I want you to log into Salesforce, navigate what your menus, go figure out where that is, fill in the form of the way that CRM want you to, that's the problem. We're saying no, I'm not going to embrace how humans think, I'm not going to embrace how humans work. I'm not going to make this something that is natural to people because I think they should be able to do it. Well, so far, we got about 50 years of computer use and we have yet to actually see that happen. So how about we do something different?
So Erik, if I was sending you a message, where are we with acne? And you said, I think we'll get it by October 30th and it'll be about $250,000. That updated Salesforce. And if that triggered a supply chain notification, production management needs to have that in the forecast now. And let's say we made that in April 1st because October, it's too far out.
So here we are. Wait a minute, Erik's deals likely to close April 1st, I got to get that on the production plan. Of course, not close. But we're going to get the probability out of Salesforce and along with that, and we're going to say, hey, wait a minute, we're going to need to start working on the supply chain aspect of this. We got to get in front of the deal desk because here we are, this is our third project with this.
All of that automation just happens. All you did respond to a text message. Krista was responsible for the outcomes. All of that the authors know what they need to do. You're the person who's just responding to a message. The author says, if that deals 80%, and Erik just said, it's likely to close at 150 on April 1st, I've now got a lot of other things I need to do. Krista can do all of those things.
So it's so important for us to recognize this technology that understands people and not the other way around. Because CRM is one good example, it's rampant in every business.
Erik: So, Salesforce I'm sure was one of the first apps that you integrated with?
John: It was.
Erik: I have a business partner who built their own CRM because he likes to manage processes and so they have a custom built internal CRM. What does it look like? Because you're then constantly bumping against these other systems, that could be a legacy system, it could be a niche system. How long does it take? Because you have to then integrate, you have to understand how they store data, how they process it, etc. What does that process look like?
John: And of course, we talked about all these automations, the author makes automations. Without regard to the actual technology, he or she will say things like create a contact, assign a contact to an opportunity, change the status of an opportunity, get all the opportunity is likely to close, all of these kinds of things. These are phrases that actually don't imply which CRM you're using.
The automations are actually independent of the actual underlying CRM. So what Krista does is Krista makes a body of programming automatically generates it and says, hey, programmer that knows that particular system, here's how I need you to do these things for me to integrate you into the rest of this environment. So when I say create an opportunity, here's all the information that the author says you would need. Please do the thing I asked you to do, create that opportunity. That may be one API call. That maybe 100 API calls. That maybe three different systems that need to have that opportunity information. That's up to the developer to do that one time.
Now, once that developer has done that work one time, I can create opportunities in 50 different ways. I can do them by saying, hey, here's a new opportunity. Let me give you this and this and this. I can do it over a voice call. But I've not needed in one additional line of code, nor have I needed to even understand the technical complexity of that particular CRM. All its author has to do is say, I need to create a new opportunity. Oh, well, you're connected to that CRM, and I know how to make that request on that CRM.
You give me the information I need, like the contact, likely close date, the likely amount, whatever, all that normal stuff. So, Krista is essentially a hub and spoke integration model for us techies. Everything integrates to Krista, not each other. That's one of our huge mistakes as we just made an absolute spiderweb out of it, which is not a good idea. Hub and spoke, everything integrates to Krista. Krista brokers, everything else.
We've got a customer with 48, CRMs. So when they say, what, what business are we doing with Company X? Someone's got to go to 48 CRMs, they give up. But no, they asked that to Krista once. Krista goes to 48 CRMs. And by the way, they're not all the same CRM. Krista can go to 48 CRMs faster than you can even log into the third one. By the time you get to three, Krista has already pulled all that back and here it is an organized list of all the act of business you have with that particular customer. And then when you say drill down into that one, Krista knows where that one came from. You don't.
So you teach Krista how to talk to that particular system, then it becomes a participant in our conversations. And it literally is a participant in our conversations. So if there were four or five people in this podcast and someone asked me a question, I would look right back to them naturally. Krista has the same ability.
Again, a technical phrase, it's an integration platform as a service. But that's a part of the technical architecture of Krista so that we can expose the capabilities of IT in a human like language and remove the programming aspect from everything above that integration, which is the process flow, the people involved, the AI involved, who needs to know what to ask them. None of that is code. That's content. It's like what's inside your Excel document, it's just content.
Erik: So, you just hit your first year anniversary of launching the product to market, I imagine you've been probably getting a lot of customer requests over the past year, also, internally identify areas where you want to strengthen the platform. Maybe if we look forward over the next 12-24 months, what are your engineering?
John: Everything. So we've been building out these solutions in our machine learning capability, we'll continue to do that. There are so many that customers are telling us now. Oh, wow, you guys do that and that. Here's another problem we have all the time. One of them I just discussed today with a customer. We're going to sign up for these invoices. It’s like there are thousands of formats of invoices. We cannot automate anything related to inbound invoices, obviously, their own invoices they can automate nicely with Krista already, inbound invoices from thousands of suppliers globally. We're just swimming in this. And it's a prerequisite to being able to do so many other things. That is a hard problem. So we've already embarked on that. We've already been doing it with some customers, we will continue down that path.
The core platform has a few really interesting additions that need to be added to. One of them is that we will strive to continue to get closer and closer to a genuine Star Trek like experience. And I hope that that is a meaningful phrase to enough of your folks. Just think about your Star Wars or you're in a SciFi humans talking to computers and computers sounding like humans. That is an absolute genuine imperative for us. So we will constantly make progress there. We've already gotten there for the people involved. They don't need anything more than the conversational interface and voice and capabilities to be out there.
The challenge for us is creating these automations simpler and simpler and more of a conversation itself. So there's a lot of R&D effort in that. We want Krista to basically interview experts, not have to be told what to do. But to ask experts what should I do when, if I don't understand, I think, Krista. I know who to ask. I know to ask Erik. Erik, I'm in the middle of an automation, I don't know what to do. Can you tell me what to do? Describe to me what I need to do so that we can be much more fluid.
Because one of the things conversations are that Krista is not right now is you're in my conversation was not pre-scripted, you decided what to ask me totally on the fly from your own intuition. Today, every automation has to be pre-thought of, pre-coded, pre-tested and deployed, and then only that is what the system is capable of doing.
We'll deliver this this year that notion of being able to build on the fly. Essentially, when someone gets into the CRM world, and says, Krista, I'm trying to split an order and that automation has not been built, Krista is going to figure out with the use of that person and potentially needing access to an expert how to split an order. And that'll be a brand new day for us because automation has never even tried to sign up for that. It's always write the requirement, put it on the backlog, hope you can get it prioritized. Someday you may get it out of the development team and hopefully, it's close to what you're looking for. And by the way, you're probably going to write three change requests on it because that's just the nature of things. Every time we see what we actually got, we then think no, wait, this would be better if. Instead, what we really need is essentially actual conversation with technology.
Erik: That would be a breakthrough. If you could explain a process and say, hey, Krista, I'm going to walk you through this. I want you to do it and then give me the result. And we check it out and you've then programmed, as you would program a colleague to figure out how to do process?
Erik: So I imagine you're moving into a growth phase here, I’m just asking because we have also some CVCs and VCs that listen in on the show, any raise happening in the future?
John: We closed around two weeks ago. So, [inaudible 53:19], the company, we just did a Series A, it was seed funded to the tune of millions of dollars. So we raised a significant amount of seed money that in the old days, seed rounds where tens of thousands of dollars, and then you raised a million or $2 million, the numbers are totally different, especially in the enterprise SaaS world we’re a cloud based SaaS solution. We're spending an entire seed round every month and infrastructure cost with cloud providers. But we raised a significant Series A that gives us more than a year of runway given our current plan. And we'll obviously be interested in both investment partners and in strategic partners that can work with us or symbiotic products to us that can help finish what the customers are trying to accomplish.
So, all of those business development things are open. It's funny because the growth equity and the private equity guys I get 10-15 emails a day, they've all bought SOC email generation bots and they're driving me crazy. When you're doing well, and Forster and half a dozen other analysts have said good things about you, you end up on lists and everybody's generating email traffic to you. So we're not interested in a race right this moment. We're going to crush our current plan. We're going to set up for probably a Series B early next year.
Erik: Well, if somebody is interested in reaching out to the team about partnership, about some maybe just to learn more about your solutions, what's the best way for them to get in touch?
John: So the easy way is JohnJ@Kristasoft.com or www.kristasoft.com. And obviously, to take a look at the website, I'd be happy to field any requests and forward to either the account folks or the customer success folks or if it's financial, I and the CFO will bring that up. I've really enjoyed a conversation, Erik and hopefully, people do. If you have questions or even those who have a completely different point of view than me, those always end up very interesting conversations for me as well. So if someone has a bone to pick with any of the comments I've made, I am happy to either defend them or to embrace critical feedback. I'm completely open. We learned by getting our suppositions tested. So you would be doing me a favor if there's something about what we discussed that doesn't make sense to you, it would be an honor for me to hear a different point of view.
Erik: John, I wish you all the success in the world. I hope my bank adopts Krista soon. Maybe we can have a conversation in another year or two, I look forward to seeing where you develop.
John: Sounds great. I'd be happy to do that and see what progress we've made and where the how the world has changed.
Thanks for tuning into another edition of the IoT spotlight podcast. If you find these conversations valuable, please leave us a comment and a five star review. And if you'd like to share your company's story or recommend the speaker, please email us at team at IoT one.com Finally, if you have an IoT research strategy or training initiative that you would like to discuss, you can email me directly at E ri k dot W A L E and Z EY at IoT one.com
Erik: Thanks for tuning into another edition of the IoT spotlight podcast. If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend the speaker, please email us at team@IoTONE.com Finally, if you have an IoT research, strategy or training initiative that you would like to discuss, you can email me directly at erik.walenza@IoTone.com. Thank you.