Episode Transcript
[00:00:00] Speaker A: Foreign.
[00:00:08] Speaker B: Today I want to talk about a theme which I call AI Goes Enterprise.
And this is a massive topic about ERP and workflow systems and chatbots and agents. And it has to do with the super worker. But I want to talk about it at an enterprise perspective. And the reason that I want to talk about it is there's a lot of announcements coming out that are very important and I'm not going to preview all of them, but I want to just talk about the theme. So for those of you in hr, and I know a lot of you are not, the last 25 or 30 years have been a very heavy investment in cloud based digital technologies at work. ERP systems, CRM systems, case management and support systems, financial analysis systems, and then all of these tools that we buy in our companies. And the average company, by the way, has 90 systems based on some new data that just came out from Salesforce.
Then there's an analytics system next to it. And so we've been automating the transaction processing workflows in our companies with all these applications. Some of them we built ourselves, some of them we buy, and then we.
[00:01:29] Speaker C: Buy a separate set of tools to.
[00:01:31] Speaker B: Analyze all the data. Now, you know, you would hope that the original systems would do all the data analysis and they sometimes do, but they don't do it in an integrated way because we have, you know, heterogeneous applications that specialize in different things.
The big erps, Oracle, SAP Workday and others have tried to build single platforms.
[00:01:53] Speaker C: To do all this, but nobody ever.
[00:01:55] Speaker B: Can keep up with all the business applications. So regardless of what you buy, you've got this problem of transaction processing in one system, analytics in another. Along comes AI. And as you've seen from our four stage model, the first manifestation of AI was a personal assistant that could analyze all sorts of data and find things and generate content. The second manifestation, AI of AI was automating some of the stuff we do. Writing code, writing documents, analyzing data, analyzing.
[00:02:29] Speaker C: Streams of data, analyzing large volumes of.
[00:02:32] Speaker B: Data and speeding up things that we either could do with the tool or we couldn't do with a tool we had to do by hand to go faster. But the third stage is to build AI driven workflows. In other words, what we call cross functional or cross domain agents that can do multiple things that connect to each other.
[00:02:56] Speaker C: So in recruiting, when you're sourcing a candidate and you have a chatbot that.
[00:03:00] Speaker B: Helps find candidates and people start applying and you run an AI process to qualify them or assess their skills or ask them to Take an assessment or ask them for their application, hourly requirements for wages or whatever. You end up with a certain number of people, and then you start sending them information to be interviewed or to do a video interview. Well, those are essentially steps in the recruiting process that would normally have been done by independent people in the old erp, maybe in an ats, but maybe in other systems, and now can be done by one agent. And so the agent starts to cascade between an agent that does one thing at a time to an agent that does a whole series of things at a time and connects the data between them.
So imagine the recruiting agent that does sourcing, assessment, scheduling of interviews, candidate analysis, background checking, job description or job offer generation, gets somebody on a customized onboarding program, follows them through the onboarding process.
[00:04:14] Speaker C: And sends them a survey after they've.
[00:04:16] Speaker B: Been there for a couple of weeks, monitors their first month of work, does.
[00:04:20] Speaker C: A review of their performance appraisal from.
[00:04:22] Speaker B: Their manager in the first 30 or 90 days, and then uses all that information and stores it. By the way, it can continue for years after that and keeps all that information for everybody that you've hired and.
[00:04:35] Speaker C: Starts looking at the relationship between that.
[00:04:37] Speaker B: Data and goes back and says, you know, all the people that we sourced from this background or this incumbent company or this degree or whatever it may be, didn't turn out too well in our company because our culture is different.
[00:04:52] Speaker C: So let's not source from them anymore.
[00:04:54] Speaker B: That is a whole new chain of applications there that have now been brought together. That was impossible in the ERP systems of the past. It simply was not possible.
[00:05:07] Speaker C: Now. I'm sure the ERP vendors all tried.
[00:05:10] Speaker B: To do that, but they didn't have AI systems under the covers when they built all those things.
[00:05:15] Speaker C: So those were static databases.
[00:05:17] Speaker B: And we hoped that the people, analytics person or somebody in analytics did that analysis. And of course they didn't because they were too busy working on something else. And plus, it took them too long. So all of a sudden, this enterprise AI, I guess you'd call it an application, I hate to call it an agent, it's bigger than that. Becomes an ongoing management tool for this whole chain of activities that go on.
[00:05:44] Speaker C: In this case, in the recruiting, onboarding and talent development and talent management process.
[00:05:48] Speaker B: And you can see that going on.
[00:05:50] Speaker C: All over the company in sales and.
[00:05:52] Speaker B: Marketing and supply chain and procurement, in hr, in finance, in different forms of operations, in research.
And that is where this AI stuff is going now. This week it was a couple of interesting things happened. There's an announcement from workday coming next week, not the layoffs, another announcement that'll be very interesting to discuss when they announce it. One of the things I was listening to is the refreshed description by Microsoft on what they're doing with the Copilot. Now, the Copilot does a lot of stuff. Primarily we see it, most of us as an office system to help us with office tasks. But actually, because it's built on the Copilot studio, it has basically a visual.
[00:06:36] Speaker C: Development environment for building applications. And so the Dynamics Group in Microsoft.
[00:06:41] Speaker B: Which sells essentially ERP systems, has basically developed a discussion of how AI is going to be the core of autonomous erp. And I really think this is where the world is going. You can call it autonomous recruiting, autonomous hcm, autonomous finance and operations, autonomous erp. And by autonomous, what we mean is.
[00:07:05] Speaker C: At the fourth stage beyond level three.
[00:07:08] Speaker B: That I talked about earlier, the system gets smart enough and monitors and manages and we manage it on this integrated process and it starts taking over for you.
[00:07:19] Speaker C: And so in the case of recruiting, instead of us looking at the data.
[00:07:23] Speaker B: With an analyst and deciding that we're not going to recruit from this source or we're going to recruit from that source, the system makes that decision for you and starts to optimize the business. This is what a self driving car does. Ideally, if a good one existed, and I think maybe, you know, Waymo's good enough to do this, it would learn that in this particular intersection, when the light turns green, you should wait 5.
[00:07:49] Speaker C: Seconds or 10 seconds because people are.
[00:07:51] Speaker B: Walking across the street very slowly.
And it could certainly learn that by location or other factors to make the.
[00:08:00] Speaker C: Car better and better at driving.
[00:08:02] Speaker B: That's what AI is going to do in our companies.
Now, it's not going to do this across the whole company because nobody's built anything like that yet, but it's going to do it in the domains of business where we automate initially.
[00:08:16] Speaker C: Now.
[00:08:19] Speaker B: Over the last two weeks, I was in South Africa three weeks ago and then I was in the Middle east last week. We have met with a lot of large companies that are looking at various AI solutions and the question they keep asking is, how do I redesign the processes and workflows that I have to.
[00:08:40] Speaker C: Implement a level three or level four? Stage three, Stage four, as I talked.
[00:08:44] Speaker B: About earlier, automation or AI solution.
[00:08:47] Speaker C: And the answer is it's a little.
[00:08:49] Speaker B: Bit more complicated than it looks. How do you take a whole bunch of people who have standalone jobs in.
[00:08:57] Speaker C: Different roles in your company doing some.
[00:08:59] Speaker B: Business process and take a step back and decide how we're going to automate this as a workflow. Well, eventually there will be off the shelf solutions like what HubSpot or Salesforce is for CRM and we'll be able to buy something from a vendor who has already done that work and we can sort of tweak our organization to.
[00:09:23] Speaker C: Fit into the workflow of the vendor.
[00:09:25] Speaker B: But the vendors haven't done this yet. So rather what we need to do is we need to go through a design exercise in our companies to look at these roles and these workflows and.
[00:09:36] Speaker C: Make some business decisions about where the.
[00:09:39] Speaker B: High value work is and what we want to automate. Now, the risk or the danger of this is we look at a whole bunch of stuff that people are doing and we start to automate the tactical things and we end up with a great automation system that might make people 5 or 10% more productive. But the big heavy, high important stuff maybe didn't get automated because we didn't think about it that way. So what I think we have to do is use some of the new tools for job task analysis and work analytics to try to understand what's going on in people's different roles. Because by the way, the job title means almost nothing.
[00:10:20] Speaker C: You're not going to learn much about.
[00:10:21] Speaker B: What'S going on in the job title, but then take a step back and map it out on a whiteboard and say, of these 89 things that are going on in our little business area, what is the big value add here that we could really automate that would transform the process? For example, if you're an insurance company, the underwriting process is magnificently important in your business. The importance of quality pricing, understanding the.
[00:10:55] Speaker C: Risk of the client or the risk.
[00:10:57] Speaker B: Of the opportunity you're trying to price and then socializing that information in a form that all the risk management has done. I mean, that kind of makes or breaks an insurance company rather. So, you know, if you were trying.
[00:11:12] Speaker C: To automate all the little steps, you.
[00:11:14] Speaker B: Might not see this big, big underwriting AI opportunity. We have a client that's done that in claims and has basically built a digital claims agent that understands the details of what goes on in a claim and is capturing intelligence about claims so that the company can now look at the workflow of claims and what is the upfront delivery of high value or low value or high profit or low profit claims clients and maybe help the company decide. We don't want clients like this because they create these kinds of claims in these kinds of situations.
[00:11:55] Speaker C: An insurance company is kind of a.
[00:11:56] Speaker B: Right place for this because there's so much data driven work here. Second example I want to give you, I was at an energy company last week and we were going through a whole bunch of this education about AI.
[00:12:09] Speaker C: And they were looking at using Galileo.
[00:12:10] Speaker B: For a whole bunch of things. And one of the engineering managers came up to me and he said, you know, I'm responsible for this engineering team in this aluminum plant and we know we need a whole bunch of new technology and new skills and we want to give people new career paths and we want to move people into new roles.
[00:12:29] Speaker C: So we want to do a skills.
[00:12:31] Speaker B: Analysis of this 100, 150 engineers and then we want to map their skills to a bunch of other jobs and other roles we have in the company.
And they don't have Eightfold, they don't have a talent intelligence platform at the moment. They don't have really interested in buying something big like that. And they said, how am I going to do this?
[00:12:53] Speaker C: I said, well, what are you doing today?
[00:12:54] Speaker B: He said, well, we're capturing their resumes, we're giving them tests and assessments, we're looking at the courses they've taken, we're getting feedback from their managers and it's.
[00:13:02] Speaker C: Taking a long time.
[00:13:03] Speaker B: It's a lot of manual labor. And I said, well, let me give you an example of what I would.
[00:13:10] Speaker C: Call a stage two solution.
[00:13:12] Speaker B: Take all of that information that you've been capturing on these individuals, put it into a collection of folders in Galileo. Galileo is based on a very powerful AI engine and ask the AI to list or categorize the top 10 to.
[00:13:31] Speaker C: 15 skills in each individual, put it.
[00:13:33] Speaker B: In a table and compare them against each other.
[00:13:36] Speaker C: Number one. That will give you a sense of.
[00:13:38] Speaker B: Whether the AI is capturing the relevant skills appropriately. And if it's not, you might need to teach the AI what these skills are by giving them definitions of the various skills which are probably, by the way, in your job catalog already. And then go back and take that big stream of data and ask it.
[00:13:57] Speaker C: Once it's come up with a reasonably.
[00:13:59] Speaker B: Good set of skills, ask it to compare this list of individuals against maybe 10 different job descriptions and tell you which of these 100 individuals is best aligned or best fit for any of these 10 jobs. Now, I can't guarantee you the system's going to do that perfectly day one, but that exercise is going to take a few days as opposed to a.
[00:14:25] Speaker C: Few months to a year. And then you're going to build this.
[00:14:28] Speaker B: Data set and you're going to be training this system on the actual skills and the actual manifestation of those skills in your company around real people.
[00:14:39] Speaker C: So the system now is suddenly smarter about your system, about your organization. And that means that now maybe it's.
[00:14:47] Speaker B: Capable of helping you with sourcing, maybe it's capable of helping you with development, planning, maybe it's helpful and capable of helping you with skills based pay. And you can see that this constellation of other applications becomes possible because you.
[00:15:06] Speaker C: Went down stage one, stage two to get to stage three, eventually to stage.
[00:15:10] Speaker B: Four in your business centered AI implementation.
[00:15:16] Speaker C: This was never possible with a transactional erp.
[00:15:20] Speaker B: The transactional ERP was good at storing data, but when you wanted to analyze it and reuse it, you had to actually do an analysis process.
[00:15:30] Speaker C: And you needed to have a human.
[00:15:31] Speaker B: Being that could figure out what questions.
[00:15:34] Speaker C: To ask and what reports to run.
[00:15:35] Speaker B: And what cross tabs to do and so forth. And nobody ever has enough time to analyze all the different business processes in a company.
[00:15:42] Speaker C: So now these systems can do it themselves.
[00:15:45] Speaker B: So I am pretty clear in my mind that something that looks like a simple stage one, stage two assistant becomes a stage three, stage four autonomous integrated workflow solution pretty quickly if you buy the right platform. Now there are vendors that are seeing the world this way.
[00:16:06] Speaker C: I told you, Microsoft sees this in Dynamics.
[00:16:09] Speaker B: This is not the copilot. This is in Microsoft Dynamics, which is the ERP system for Microsoft Paradox and a new company called Maki. People that we just started talking to are doing this. In the case of talent management and recruiting Eightfold is going to be announcing some tools in this recruiting. Sana does this in learning and I think this is where all of the AI driven tools are going to go. You know, we originally thought AI was kind of cool for generating a piece.
[00:16:38] Speaker C: Of content, generating a job description, generating.
[00:16:41] Speaker B: A document, generating a report, you know, editing our own content, creating an image, creating a video. That's great, you know, that's very important for creative work. But I think this bigger scenario of AI creating cross domain and cross functional workflow solutions is way bigger. I mean, I think orders of magnitude bigger. The vendors that are probably going to be the most quick to market in this space are going to be companies like ServiceNow.
Those of you that use the Copilot studio can do some of this yourselves.
[00:17:18] Speaker C: Some of the mature vendors in different.
[00:17:20] Speaker B: Domains of hr, the forward thinking, learning and learning systems. By the way, aorist, which is a really cool little company that does this, I'm going to be putting out a.
[00:17:29] Speaker C: Podcast on them pretty soon.
[00:17:31] Speaker B: Sana Paradox, Eightfold Phenom. You know, these, these specialized companies that have very deep domain expertise in these business processes.
[00:17:44] Speaker C: By the way, there's going to be.
[00:17:44] Speaker B: A lot of tools like this in.
[00:17:46] Speaker C: Compensation and benefits, where we have massive.
[00:17:49] Speaker B: Amounts of workflow data to use to better improve those systems, and in employee experience. By the way, there's a really interesting marketplace of vendors now, including to some degree Pecon from Workday, that are starting to use AI to assess and understand and improve employee experience through conversational interfaces and conversational data that comes back from employees.
[00:18:16] Speaker C: There's also an announcement coming out from.
[00:18:18] Speaker B: Workhuman, which they call human intelligence, which is announcement about using recognition data as a source of valuable information, about skills, about potential, about desires, about engagement, about management and leadership capabilities in the company.
So this AI in the enterprise honestly has the potential. I'm not saying it's going to replace our current transactional systems, but to slowly sort of chip away at them and.
[00:18:49] Speaker C: Turn them into very, very different dynamic systems.
[00:18:54] Speaker B: The question everybody keeps asking me everywhere I go is how do I get there from here? How do I learn more about it? You know, what should I be doing? Here's my advice for all of you. Number one, take some basic courses on what AI is. You really do need to understand the.
[00:19:09] Speaker C: Fundamentals of how this works. You don't need to be a software.
[00:19:12] Speaker B: Engineer, but you need to understand the idea of drift, the idea of hallucinations, how a prompt actually works, what it's doing and so forth.
[00:19:21] Speaker C: You can learn that in the Josh.
[00:19:22] Speaker B: Burson Academy and there's millions of other courses you can take to do that. Number two, explore use cases. Download our free report, the 100 Use Cases for Galileo, and just scroll through.
[00:19:36] Speaker C: It and read through it and it.
[00:19:37] Speaker B: Will expand your imagination to think about the things that are possible.
[00:19:42] Speaker C: And then the third is to get your hands dirty. Get your hands on a tool.
[00:19:46] Speaker B: I mean, Galileo is a tool that can be used to explore all of these scenarios. There will be other, and there are many, many others that are kind of baked solutions out of the box as well.
And explore your job, your role, your business processes.
[00:20:04] Speaker C: And think about this as not just.
[00:20:06] Speaker B: Something for you to make your personal productivity better, but something that can work.
[00:20:11] Speaker C: In your functional area across multiple people. And once you've gone through those three.
[00:20:16] Speaker B: Steps as individuals, I think the next step is you get together in a conference room, not virtually, probably face to face for a couple of days, and.
[00:20:26] Speaker C: You map out of all the things.
[00:20:28] Speaker B: We do in our jobs, where can we add the most value by automating.
[00:20:33] Speaker C: In AI to get us to level.
[00:20:34] Speaker B: Three and level four.
[00:20:36] Speaker C: This is an exceptionally interesting period of.
[00:20:39] Speaker B: Time because the vendor market is very, very active at re engineering and rethinking what they're doing.
[00:20:46] Speaker C: All the big companies are building AI platforms to sit on top of their.
[00:20:50] Speaker B: Core systems and then the small ones.
[00:20:53] Speaker C: Are getting funded faster and faster than ever.
[00:20:55] Speaker B: And it's going to be a very, very dynamic place. We are going to be talking a lot about this at the conferences this year. We're going to have a whole series of activities at Unleash in Vegas in May and also in April and May and also in the Fall.
[00:21:12] Speaker C: And we are starting a series of Galileo users groups. So those of you that are Galileo.
[00:21:18] Speaker B: Customers, keep in touch with us. Read the newsletters we're sending out and.
[00:21:22] Speaker C: You'Ll be able to join a users.
[00:21:23] Speaker B: Group to talk about automation projects in your own company. We also are always looking for great stories to prepare for our conference in May.
[00:21:32] Speaker C: So if you have some really interesting.
[00:21:34] Speaker B: AI implementations or visions that you would like to share with us, just contact us. We'll invite you to come to the conference. We'll interview you on the phone, do a podcast, whatever you'd prefer, and help share what's going on in this really exciting world of enterprise AI. Okay, that's it for now. You guys stay tuned for a big announcement next Tuesday and Wednesday and I will tell you more about this new stuff we're doing then. Bye for now.
[00:22:10] Speaker A: Sa.
