Ep 57 - HR and AI: Feeding the Machine Without Starving Ethics

Watch the YouTube video version above or listen to the podcast below!

Ep 57 - HR and AI: Feeding the Machine Without Starving Ethics Podcast and Video Transcript

[Disclaimer: This transcription was written by AI using a tool called Descript, and has not been edited for content.]

Dave Dougherty: [00:00:00] Hello and welcome to the latest episode of Enterprising Minds. We got the whole crew here is fantastic and enterprises. Had the most solid idea when we were generating topics for today, so why don't you take us?

AI HR Policies: To Feed or Not to Feed?

Alex Pokorny: Sure. Yeah. It was just a kind of problem that I was realizing that we're coming up against, which was initial AI HR policies typically at a company was don't feed it personal data.

Don't feed in any company data. Sure. You can use it for making, you know, an Excel formula or helping you to write an email, but, you know, try not to name names and [00:01:00] definitely don't use any kind of, you know, real world data in it. . And then as things have kind of progressed, we've learned more and more about AI tools and making them more useful.

It's basically sharing tons of data with them and if you give them lots and lots of context. Then they are far better at being able to pull together solutions, answer questions, all the rest when it's, you know, business related. So it becomes this problem between the two of them of do you feed it or do you not?

So two feed are not to feed, that is the question. So what have you seen, kind of what are the, the limits? You think should be there? Any solutions that you've found? I've, I've actually just seen some few that might help, but I'm interested in seeing kind of what everybody's experiences are too.

Challenges in Enterprise Tool Adoption

Dave Dougherty: I think for me, at least within enterprises, you have that, that kind of push pull of, Hey, we bought a tool, you should use it.[00:02:00]

And then there's no training, there's no onboarding. Or if it is, they create a SharePoint and just say, all the things you need to know is there. And you go, okay, but that's not effective. Or I had a training recently where it, it was like six hours out of my week and it was so detailed on how to submit certain types of tickets.

That 15 minutes in, I just went, I'm not gonna remember any of this until I actually do it. And I don't know that this applies to me and I have five and three quarter hours to go. So, you know, there's a whole lot of problems in terms of rolling things out and communicating things well. As well as like, to your point, what is allowed and.

The gray area that's created through, do we have something off the shelf that then [00:03:00] consumes everything like a chat, GPT for enterprise or you know, Google, Gemini, Claude, whatever, or do we build a bespoke system? Yeah. Leveraging these things, but isolated to the company data so that you can use company data.

You know and honestly, I think you're gonna have a mix of both, whether official or unofficial. And the, the real question isn't necessarily to feed or not to feed, but what to do when things do get fed, right? Because you're not gonna be able to control everybody. No, no. You really can't.

Balancing Data Privacy and AI Utility

Ruthi Corcoran: I come at it from a slightly different angle, which is when I, when I hear you talking about A IHR policies, and don't feed a personal data, don't feel like company data, feed it company data.

This to me is a pretty [00:04:00] stark line. Of just like, you just, you don't do that. Like you, you know, getting an Excel formula, the sort of generic requests, that's one thing. And I go, yeah, Chachi, Petit's really good at that. Like it can do a lot of things. But when it comes to sort of company data, et cetera and working in, you know, I work in the healthcare space, so it's, it's much more conservative to say, Hey, that's just, we we're just not going there.

'cause we don't wanna open the gates until, until we've sort of got the company policies and such in place. That said, I think your comment about, hey, having the context present and therefore the data is gonna help you have better decisions. And so I imagine other companies are doing something similar to what I experienced, which is we do have an in-house tool.

Both, you know, there's a, there's a series of, hey, maybe in-house proprietary closed loop systems isolated Dave that you mentioned. But also copilot, right? I think this is one of the, the things that Microsoft did early on was [00:05:00] to say, Hey, we're gonna offer an enterprise solution that solves for some of the known enterprise issues within copilot.

Is it as effective and awesome to use as, as Quad as chat, GBT, you know, no, not at all. It can be incredibly frustrating at times especially if you're used to working with the, the sort of more out of the box continuously updated. Tools out there. But one of the things that I, I noticed yesterday I was listening to chat t's announcement for business.

They, they made a, a number of things and essentially what they announced is, hey, chat. BT Dow does the things that Copilot does. Yeah. Which is can access your data, it can record and transcribe everything you do. I wouldn't be surprised if it does a better job in terms of providing better, more contextual answers and information than copilot does.

But that suite of tool is so valuable. The idea that I could say, Hey, copilot, please summarize, I. Yesterday's emails 'cause I was [00:06:00] out of office. Get that, find out quickly who I need to respond to, or, Hey, I've got this meeting coming up. Help me prepare. What are the documents I should be looking at?

This stuff is invaluable. So I think there is a, a huge, a huge benefit to providing the context and the data needed to be able to have those advantages. And part of that is just having, having the enterprise solutions to be able to enable that.

Dave Dougherty: I guess when you, when you refocused on hr, right, of course. My brain goes a thousand different ways. You know, there obviously there's some applicable laws that you have to go and follow right around personal identifiable data and whatnot. I guess for me, where my brain went. In particular, it was, especially with the Google IO conference, they showed this a little bit last year, but then this year especially, they showed it with the Google Glass relaunch [00:07:00] and that real time ai.

And the voice to me is where I get more concerned because if you have your work computer. If anybody's listening, I have definitely not done this, but I have thought of this as a, an issue to discuss on this podcast. If you are sitting on one of these like huge Power bi dashboard with a ton of data and you're reading through it and you want certain questions answered, but.

There's no way for you to copy that over into your proprietary system or whatever else. It is very easy for me to imagine somebody pulling out their personal phone and pulling up Gemini Live, showing, you know, the screen from the camera and then asking questions about it, and then just typing in [00:08:00] whatever they need.

And it's like that kind of use case is something that. In terms of HR policy, I could see that easily being a fireable offense because of the proprietary data piece.

Ruthi Corcoran: A few of the things I think about when you ask that question about providing data and giving it context. Because I am so used to having the enterprise tools available and just having to work within there those sorts of questions I haven't thought about in the business context as much simply because it's sort of a.

It's an, it's a non-issue to an extent. Like I don't, I don't have a choice in the matter, or at least I, I'm choosing to follow all of the guidelines and the, the sort of rules and regulations which I think is an important thing. But what I do think about is in my personal use of the various tools, there's a.

A squeamish factor. I'm not sure. Like [00:09:00] I, I don't know how much, how much information I am comfortable or I'm ready to relinquish to various companies. . And I've always felt this way around, around Google and sort of how much information I'm providing. And I still get a little squeamish about some of the photo applications that are like, oh yeah, we can track your, your face across the things, and then you can search for your face and you can find all the pictures that have you in it, like e even.

Even with the sort of rudimentary versions of this, I found myself hesitating. And so I, as of late, I've leaned more towards, you know what? I don't need to give every last bit of detail about the situation I'm encountering in order to get a good enough result because I just don't know enough about the space yet to know whether or not I, I want.

The various tools to know you know, who my family members are, how old they are, what their preferences are, let alone my own preferences. [00:10:00] And of course, a lot of that comes out with the types of questions I asked. So now, Chad, GPT knows what I had for lunch yesterday because it gave me recipes on, or the amount of baking time I needed for particular set of potatoes and some garlic toast.

And so there's, there's that superficial level. But I, I'm curious what you guys' thoughts are about that layer deeper of where that line is for you, about how much information you're willing to provide these different tools in order to get the better results.

Dave Dougherty: Alex, you've been awfully quiet thus far.

So that's you. First I'll throw on.

Alex Pokorny: Actually had a fun icebreaker at work and. It was Ask chat, gt, or whatever AI tool as you prefer to create a trilogy poster about you. And along with, you know, plot summaries. And if you use like a kind of snarky one, like I prefer to use, which was like a chat BTS Monday, which is very snarky.

You know, it gives you something of like, you know, I, I came, I tried, I went [00:11:00] back to bed or something like that. It's like, it just, it was pretty funny actually because everyone who did it, there was enough data and they all got basically the same result. I switched over to using the standard models that I use and I asked it saying, of all the memory that you have on me across all the different conversations, create this.

And it was kind of fascinating because it gives you this like, you know, summarized glimpse into what it understands from you. And you can always go to like chat GT's memory, and you can read the entire kind of basically transcript of what it knows about you and all these different elements, things that you've asked it to remember, and things that it's picked up on.

. But what it got was basically is I ask a lot of really technical questions. I played a bunch about like random AI games with it, and I've asked some random like trivia kind of stuff. So it basically gave me like, you know, three snapshots. Does that represent me at all? Not really. That's, that's not really targetable [00:12:00] psychographic information.

It's not so much about any kind of specifics. I also have the fortunate opportunity of working with a company that's not as tightly regulated as my prior employers. So we're allowed to do much more, but there still is those hesitation, I guess where my line is to your question. Definitely not sales data.

I know it could be way better at some analysis, and I've given it random figures before and it's pulled out things that I didn't recognize, like just a curiosity. I, I put in a bunch of like, random number charts and I made a chart out of it saying like, analyze this, like, tell me product A, B, C. .

You know, what's, what's the sales volume? Or something like that. Right? And it's all made up into Excel data. But it pulled apart things that I, at first, clients at least def definitely did not recognize. So there's, there's value in it, but I'm preventing myself from having that value because of that line too.

So I'm, I kind of always struggle with that line of, you know, how far do you go with that? Allowing it to know public information. Again, being [00:13:00] fortunate that I'm on the marketing side of things, most information that I have is public information, so that's website, online, PDFs you name it. So extracting data from those, creating images based upon our brand standards, based upon our website.

It can do all those different elements. It can summarize, you know, our tone and voice based upon our text, and that's, that's all public information as well. In a way, I am requesting it train itself and create these things though. So you're still compiling things that didn't exist before. So in some way you are still sharing data that it didn't know before too.

So that's always kind of the back and forth of it too, of like the output is now a created piece of content. Now knows privacy wise, of course, I always turn everything off. I just kind of, that kinda individual in general. So I do try to keep it private.

What about you guys? What's, what's the limit?

Ruthi Corcoran: Bit tricky for me to extrapolate further [00:14:00] simply because we are just using the internal tools. They are sort of closed system. I think you bring up a really good point, which is. Where you're not using an internal system, where you're using something like a chat, GPT a quad even if you're working with public information, say you're working on something that's available through a website, PDFs that are publicly accessible, the questions that you're asking to some degree could relate back to company strategy.

Because you're asking those questions and you're saying, this is what we're potentially, this is what we're trying to accomplish. This is what we're trying to do. And that's where the gray area becomes very interesting because we don't yet know what the negative outcomes of that could be. And what I mean by that is how.

If it was sort of a, a random person listening into the questions you were asking in a vacuum, they might not be able to [00:15:00] glean that much. Like, oh, cool, they're improving their website, they're trying to optimize, or this part is, is of the website is not very good. They're looking into how they can fix that part of the website.

Like in, in theory you could find that out by just looking at their website. But you might not know this is the exact thing they're looking at. Okay, now we're working with systems that have a lot of information and there's a bit of smartness thrown on top. What are the sort of second and third run effects of having that sort of compiled together?

And that's, that's where my mind goes when I hear you talk, Alex is. I go, we don't know what the, what the unintended consequences are or what the, what the novel what the, the novel uses of these tools could be. Because this type of information gathering and ability to access data hasn't existed in this format before.

Dave Dougherty: So [00:16:00] for me, it's interesting, at least with.

Okay. Marketing, marketing use cases with proprietary data. I think, Alex, what you said about the publicly available data, I'm gonna go ahead and use that as much as I possibly can. 'cause then I don't need the legal plu rules, then I don't need the the checks because it's already out there. Right? It's already been done.

AI in Performance Reviews and HR

Dave Dougherty: The use case I've been hearing the most about recently has been with midyear performance reviews, and sorry, that

Alex Pokorny: just imagining how bad that can get.

Dave Dougherty: So let's play that one out 'cause that's gonna be relatable to everybody. Right. So how does, how do most people keep track of it? They don't, I. [00:17:00] For the people who do keep track of what they've done for the, you know, throughout the year you have a Word document, you have an Excel sheet, you have something, right?

So then when the company comes back at you and says, all right, put together a one slide put together, you know this in whatever HR human capital system they've subscribed to then you can, you know, go into that. And I know certain. Certain HR platforms are now using AI to extract data from what you've put into the system to decide whether or not you are going to be ready for a promotion, or you are in burnout, danger, or you are in, you know you know, all these other effects that come in.

Are you necessarily worried about proprietary data at that point? I don't know if it's internal communications, like that's where I'm gonna lean on the expertise of the, the, the [00:18:00] lawyers and the people who are vetting the stuff. But yeah, I mean, just. Some people just sitting down with the AI and saying, okay, here's my gut feeling on what I've done this year.

Here are my frustrations, but, and what I think I'm good at, and clean this up so I can put it to my boss for the, you know, the midterm review so that so that I, I can just keep doing the to-do list.

Ruthi Corcoran: It's just depressing on so many levels. If that is what your midyear review is, then you, your boss is not doing their job.

.

Like . I don't care. And I've told them this, I don't care if people I work with or who work with. On my team are using AI to help them formulate their review slides or, or, here's the recap of what I, because I don't want them spending time wordsmithing, like when I have to do those things.

Historically, I've spent [00:19:00] so much time agonizing about every single word. It's like building a resume. That's not a good use of anybody's time. The whole point of reviews and things like mid your reviews is it opens up conversations. It says, how are we doing? Are we meeting the things we talked about last time?

It's not, it's not just paper crunching. Or if it is, then something has gone significantly wrong within the sort of HR approach.

Dave Dougherty: But how do they show usage on the new tool that they bought that has AI features for, you know, human capital management?

Ruthi Corcoran: Let's see. If you

Dave Dougherty: don't have these very specific things uploaded.

Ruthi Corcoran: Just kills me, Dave kills me. When metrics become a target, they seem to be cease to be useful as a metric.

Dave Dougherty: Thousand percent. But how else are shareholders gonna love you if you're not saying you're the most efficient thing possible?

Ruthi Corcoran: I think this gets at, to this, this, so we're talking about AI [00:20:00] in enterprise and this I think, gets at a really important sort of mindset that we need to take on or we need to adopt.

We need to think about our mindset as it relates to ai. If it's just allowing us to do more of the number crunching that looks good on paper and allows us to do these sort of large scale reports that may or may not be. I don't think we're, we're sort of really reaping the value of and potential that exists within this new suite of capabilities and tools that we have.

When I thought about midyear reviews, when you brought that up, I was like, oh, I've got a fantastic use case for this. Because one of the things that we wanna do is we wanna seek feedback from other people about how, how we're doing. What I, I'm not a survey expert. I don't know the right way to ask questions.

You know, you could give me like two days and maybe I could figure it out. So one of the first use cases I had was to say, Hey. You know what? [00:21:00] You are a questionnaire design expert. This is what you're trying to do. Come up with a series of questions. Okay? Now you're a different questionnaire design expert.

What would you do? Now, you're a different questionnaire design expert. What would you do to create a series of questions? I go, oh, this is gonna help me better understand how I'm doing, how those around me are doing and create better, better room for conversations. And that's, that's, I think the piece that's so important is AI can create content.

From now until forever. And, and no one is gonna read it or it's not gonna be useful unless we, unless we have, unless we add the value to it. I'll stop there because I've been on a bit of a rant and I can see Alex has some, has something to say as well. I'm

Alex Pokorny: struggling. I had, so I value 360 reviews highly.

. You're basically asking to explain those quickly. You, you basically are sending a quick survey to colleagues of an individual, either someone who reports to you or yourself [00:22:00] to, you know, maybe five, six people. They fill out the different fields, they send it back to you, and you basically get a, a nice kind of transparent critique of.

How they see you doing in these different areas and where you can grow. I think it's really helpful because you can craft those questions. And I spend a lot of time, and I even have some like templates that I like to use because I don't like standard ones and stuff like that. Trying to get down to something that's very actionable instead of just like, you know, how did this person, you know, apply themselves for this random, you know, you know, one of our mission statement pillars or something, it's like some random phrase and it's like, okay, no, no, no, no, no.

Let's, let's actually make this real, like make some value out of this. I then I had a coworker confess to me that they used Cha g Bt to fill out all the 360 reviews requests that they ever got because there's too many of them.

And it killed me because it was like, you value it so much and they don't. So they're putting [00:23:00] in not much effort into this and honestly. This being three sixty's and AI and all the rest of that could be swapped out for so many other things too. It is a question of doing the important thing, communicating that it is the important thing, and having people agree that it is the important thing and working on it like it is the important thing, like it's.

Creating that focus, creating that priority on it, and then having that shared desire and focus to see it through. . And if you have just more and more and more, more reports, more pages, longer, slides, longer, more charts. That's not helpful. The analysis is what's helpful. The recommendations are what's helpful underlying data and all the rest is just the proof points.

And then you move on from that. And I think that's the difficulty here is what we're seeing is it is really easy to create the fluff or the proof points underneath, but it's still difficult to summarize it down to being the, the actionable things, the important pieces, and to [00:24:00] communicate that. That's a team thing too.

Dave Dougherty: This is interesting for me because I, in the back of my head, I've been thinking about, we've had generation two generations of people who have had, you know, Google Analytics, Adobe analytics, like realtime data on social media platforms. Like just overwhelmed with data all the time. Right. And I think.

There are a number of instances where I've been looking at the situation and I'm like, oh, because there's no data, or because the data doesn't tell the story that you want it to, you are afraid to make a decision, right? You don't know how to make a decision without having 17 bullet points of here's the thing that proves that you know, this is a safe play, [00:25:00] right?

And that's the thing that is interesting to me with, again, some of these 360 reviews or, or survey questions is like, yeah, it does, AI does the thing well, but if it's not prioritized or, if you're not actually dealing with the human beings correctly, it will still absolutely fail because if you have that 360 meeting, but then the manager shows up to the, the review and just goes, look, things have been on fire, are, are you gonna quit?

You're just a crap manager. Let's start there like,

you know. Because the no system, no AI is going to fix that EQ problem. Right? Yeah. And yeah, I don't, I don't know. I'm gonna step down before I get ranty.

Ruthi Corcoran: So fascinated to watch what [00:26:00] the different cultural norms will evolve around. Mitigating this problem that you just described, Dave, I don't even know what the name for that is, as well as anxiety.

I think the AI bluff, right? Like the here's a 40 page deck that just spews information at you. It says, oh, just do the thing. I said, here's 40 pages that say why you should do this. And it forces the recipient to do the analysis. So I've seen this come up a lot more in the last. Actually a few weeks, but even just a couple of months of just the spew of content as though it was sort of a stand in for an analysis or a recommendation.

It's just like, here's a whole bunch of here's, here's just a whole bunch of things. Here's the 17 bullet points, whether or not they're good. 17 bullet points, we don't know, and just, just like. With text messaging and emojis with social media and who you share and how you share there? There's norms that we evolve, sometimes slowly, sometimes quickly.

Sometimes [00:27:00] the, the younger generations sort of evolved them little bit more quickly. We're, we're so early that a lot of those norms haven't quite, quite come into play yet. And so that's one thing that I'll be, I'll be watching.

Alex Pokorny: . I think one of the key things I. Coming back to is that AI isn't good for creative recommendations.

Creating those, like, here's five different disparate pieces of data. I'm gonna pull this together and summarize it. Some of the deep research ones are okay at it, but that's it. They're okay at it. Really content generation for, you know, a gen AI system is what it's built on and built for. It's great at.

Being really verbose, you know, just word after word after word, and you're creating a bunch of stuff on, not much. . But coming up with the creative solution, the prioritization, the internal, like emphasis in Dave, to your point, the eq. . [00:28:00] Those are all gaps. So it is not some, you know, wonderful tool for absolutely everything.

There are just so many gaps all over it, and. I think you said it really well, you know, not treating people as people. That is a massive problem. The more that we do that, the more that we fail, the more that we don't do that. Treating people as people.

Ruthi Corcoran: One thread I wanna pick out of that is it,

it does, these tools do create a ton of words and. What's your comment about how they don't create things, they don't generate, they don't prioritize, et cetera on the surface? They do, and that's what's so interesting to me, is you don't, unless you know the subject matter well it's hard. It's hard to spot.

The, the issues or it's hard to spot the faults [00:29:00] which is yet another area where we're, we're gonna have to figure how you work around that may, maybe these, maybe it gets so good so quickly that that just goes away. But we still are back to that piece, which is you, you have to recognize either when it's hallucinating or when it just didn't do a very good job.

Like it may be like true, it's just not very good. And, and perhaps some of the value we get outta these is sort of providing us different lenses and ways of thinking rather than doing the thinking for us.

Dave Dougherty: On that note, however, I think there is. Propensity for human beings to just take what the computer says as fact and roll with it.

So that's something that I'm on the lookout for, like use cases for that because I know, I've seen it. I've, I've heard arguments where people [00:30:00] are pushing back on. People with very high expertise saying, no, you're wrong. Here's what copilot told me. It's like, whoa, wait. Do you realize what you just implied?

That the entire team of specialists the company has employed doesn't know the ins and outs and the nuances of your particular situation to the extent where. Copilot the worst of all. Ais will tell you the reality. Like, no. Speaking of hr, we should be able to flag people like that. Like this person's an idiot, please review them.

Alex Pokorny: Sorry, this came up earlier and I wanted to comment and just, I just wanna throw in this kind of side note right now.

Enterprise AI Tools: Copilot vs. Gemini

Alex Pokorny: Google. Gemini not fantastic. Good at some [00:31:00] things. Struggles with a lot. Microsoft copilot struggles a lot. . Two, you know, industry incumbents, massive, you know, industry giants who are absolutely failing because of smaller companies that are more nimble and quick, who are able to move faster.

And OpenAI has stated that they still have annual goals. They still have annual, you know. Basically, you know, setups, they still do quarterly goals. They still do all the same kind of corporate elements that any other corporation has. . But they are moving so much faster and so much better than these two massive companies who have all the people and all the history and all the money and they can't do it.

Just kind of a interesting behemoth versus so. Start. Ask me what, were there billions still open eye with their,

Dave Dougherty: I would challenge that a little bit because, [00:32:00] okay, Chad, CPT was the first to market, so they get a kind of a pass for being first and people are more familiar with them, right? I mean, there are obvious advantages to being first to market.

They have. Specifically stayed with text and memory around text and do they have these other things? Yeah, but for the most part, all of the other features that they have, like, you know, voice mode and whatever else are text related. If you look at what Google has been doing, the deep research lab is doing a lot of vertical things, right?

The alpha fold that won the Nobel Prize for protein folding. . That is a very specific use case, that has massive potential for things. They've also spun out an entire AI thing for drug development. Again, not necessarily [00:33:00] consumer. Massive potential in the business side of things. The VO release that they just did on, you know, videos and stuff, and being able to have that continuity between scenes again is really cool.

That's the most consumery thing that they've, but it's spread out. I feel like if Google had a focus. To the extent that chat t has, it would be better. My frustration with Gemini in the way that I use it compared to chat t is that it struggles with memory. It will tell me, oh, I'm just an LLM, I can't do that when my request was perfectly within what an LLM is capable of.

And then I go. You didn't do what I asked you to, and it's like, oh yeah, you're right. And then it will do it. [00:34:00] So like, why, why is that happening? Right? But you, I don't know, you can't sleep on the amount of data that Google has on the entire world, you know, if they don't make the same mistake. Microsoft did and make, you know, AI for Word and AI for Excel and AI for, you know, but instead, does the Google Gemini multimodal across everything, has the personalization across all of your apps that you allow it into, then it will become a lot better, more quickly because it, you know, it knows you and has your map history and you know, you seem to be enjoying pesto a lot these days like.

You know, whatever else. So that would be my pushback.

Ruthi Corcoran: I think you guys are both being overly pessimistic.

Dave Dougherty: What a surprise. Really. This is our role. Go istic. Go.

Alex Pokorny: Yeah, exactly.

Ruthi Corcoran: I, [00:35:00] I think we're short changing Gemini and copilot and here's. Here's why it in my, they're serving a different role at the moment than .

Something like a chat, GBT or a Claude. And to your points, those chat, GBT and Claude are a little bit more, more niche. They're a bit more focused. They're doing a thing. But also the thing that I see is. Microsoft and Google, they've got this suite of productivity applications. . And so it would be silly if they didn't connect this technology into those productivity applications, which presumably just takes a slower amount of time.

To, and to your point, I think it's a good one, Dave. Google's not focusing in part because they've got a lot more potential and so it's like, let's try a bunch of different things. 'cause some of them might add value that only Google could add your comment about the data and poor copilot. I get, I know it's, it's frustrating and, and the pro creating prompts [00:36:00] for copilot sometimes takes many more iterations just to get it to do the thing that you're trying to do.

And it is frustrating. But I think the reason for that isn't because Microsoft doesn't know what it's doing or it's particularly ill-equipped. It's because it has to solve for all the problems that companies like mine are demanding, which is . Well, we, we gotta keep our data secure. We gotta have these enterprise legal and IT security restrictions.

It's solving for a bunch of those. Not to mention each company can place a layer of additional restrictions on top of it. But they are releasing things like I can in a couple weeks here I'll be able to go and create a custom agent that works within teams that uses all the different company data and stuff.

I can populate it with, here's the types of things that I, I wanted to focus on, talk about, and now it's incorporated into teams. So the flexibility by built into applications is gonna be much more robust. Then something like chat, GBT does because chat, GBT, although they announced it yesterday, [00:37:00] they're now starting to do integrations, has to build all those integrations from scratch because it doesn't have this series of applications.

. So that's kinda where my thought is. It's, I. These tools are all doing really fantastic and cool things. It's just they're optimizing for a different set of features and benefits based on where they're sitting in the market. So I dunno, let's, let's cut copilot a little bit of slack for those of us who have to use it.

It's still pretty cool.

Alex Pokorny: Better than

Dave Dougherty: clarity. The potential is there. Do I have any faith based on historical. Data on Microsoft being able to execute a great idea? No. No, I don't.

Ruthi Corcoran: Have you used Loop?

Dave Dougherty: Yes.

Ruthi Corcoran: It's amazing.

Dave Dougherty: It is. It's cool if I had time to set them up for everything that I have to do, but I don't. Google kills [00:38:00] products about

Alex Pokorny: like every week, it seems.

So the, the lifespan of a Google cool thing is also not long, right?

Ruthi Corcoran: I think priority solving for enterprise is a level, level of complexity that's just, it's just much greater . Than the sort of off the shelf, direct to the consumer. It, it's just more difficult and we've all, we've laid out all the reasons why.

It's more difficult. Yeah. Because you've got all these data considerations. How much context, is it a closed loop? Have we considered this? Have we, like what are the HR implications? I mean, here's a, here's another one just to toss in for fun. You in order some, sorry, I'll back up. In order to bring in an AI tool.

Some companies have policies that say we've got a test to make sure that the text generated by, by the AI [00:39:00] isn't out of compliance with our HR policies around how humans treat other humans. Right. They, they have to be right. They have to be compliant. They, I, I am forgetting all of the words, the HR buzzwords, I'm sure you guys can fill it here in a minute, but they have to make sure the hr, that the AI tools are compliant within HR policies that they set for humans.

And, and so that's where we go. Okay. Like Enterprise does, does come with a different set of ex considerations. And that's, I think, one of the. The things that we're good at is saying, Hey, this is how you operate within an enterprise and this is how you've gotta think differently.

Alex Pokorny: Microsoft Teams and Microsoft Outlook should be the same piece of software debate.

Ruthi Corcoran: I. They just made an update so that when you try to open a event within, within teams, it's more of the calendar view. I don't know if it's [00:40:00] exactly the same or if they're just mimicking it, but agreed.

Alex Pokorny: In fact, they will a calendar and kills me and the one that has more options and then if you open a dip one that has more options, outlook, it'll open up teams to be able to display the teams meeting that's been added to the Outlook meeting that this is ridiculous to me.

They're the same at this point.

Dave Dougherty: But it's easier to ignore people on different one.

Alex Pokorny: I want two different levels of unread. You here and you're under there.

Dave Dougherty: Yeah, no. The most useful feature that they did across both platforms was the automatic replies. Finally working on both so that you weren't out of office on Outlook, but still available in teams, which was horrible because then you get all the onesie twosie chats. Like, Hey, got a minute? No, I'm a thousand miles away.

Okay, cool. 15 minutes. Wait, how'd we go from one [00:41:00] minute to 15? I said no, like,

yeah. So this is interesting, and Ruthi, to your point, you know, the, the complexity of enterprise is a lot more, which is why I've always been fascinated, like as sort of third party companies started adopting AI and adding them into the tool set. Watching the difference between HubSpot and Salesforce, you know, since their markets are very targeted to either enterprise or small and medium businesses I do feel like HubSpot is a little more nimble, but through that also creates way more variability.

So, yeah, there's something, definitely something to keep your eye on as this develops. And I think we've, we finally got there. You know, we finally found the passion and the, the stupid things to comment on, which is fantastic. So thank you [00:42:00] for listening everybody. We appreciate your time, like, subscribe, share, also.

Another shout out for the Pathways email newsletter. Subscriptions will be link to describe, there will be in the description, you will get sort of summaries and hot takes on each episode as they come out every other week, as well as a deeper dive into particular topic that is either.

Relevant in the news or a deeper dive in a topic from an episode that we had, or like Ruthi's always bringing up really good frameworks. There's I. Explorations of those frameworks in those newsletters. So a lot of value there to get. So go subscribe. Thank you, and we'll see you in the next episode [00:43:00] of.