Discover how AI is not just a tool for automation but a strategic partner that can transform your HR functions, enhance decision-making, and foster a more engaged and productive workforce.
In this webinar, we'll:
- explore tangible AI applications that are reshaping HR, from crafting precise job descriptions to extracting profound insights from employee feedback
- learn how AI can elevate performance reviews, streamline recruitment processes, and
- provide data-driven insights that empower HR professionals to make informed decisions.
Theresa Fesinstine will share her journey from an HR executive to an AI advocate, highlighting how AI tools can be seamlessly integrated into HR practices to yield significant benefits.
Whether you're an HR leader, a business executive, or an entrepreneur, this session will equip you with the knowledge and insights to harness AI's potential in your organization.
Don't miss this opportunity to gain practical, actionable strategies that you can implement immediately in your HR operations.
About Leveraging AI
If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!
Discover how AI is not just a tool for automation but a strategic partner that can transform your HR functions, enhance decision-making, and foster a more engaged and productive workforce.
In this webinar, we'll:
- explore tangible AI applications that are reshaping HR, from crafting precise job descriptions to extracting profound insights from employee feedback
- learn how AI can elevate performance reviews, streamline recruitment processes, and
- provide data-driven insights that empower HR professionals to make informed decisions.
Theresa Fesinstine will share her journey from an HR executive to an AI advocate, highlighting how AI tools can be seamlessly integrated into HR practices to yield significant benefits.
Whether you're an HR leader, a business executive, or an entrepreneur, this session will equip you with the knowledge and insights to harness AI's potential in your organization.
Don't miss this opportunity to gain practical, actionable strategies that you can implement immediately in your HR operations.
About Leveraging AI
If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!
Hello and welcome to a live episode of Leveraging AI, the podcast that shares practical, ethical ways to leverage AI to improve efficiency, grow your business and advance your career. This is Isar Maitis, your host, and we have a really exciting guest. Session today. The topics we usually talk about go around marketing and sales and operations and so on. But the core, the real core of every single business is the people when I do my consulting or when I speak on stages. People ask me, what's the most important thing in AI adoption? And I say, it's a change management process and people are the one that needs to change and adopt and so on. Otherwise it doesn't matter which systems and processes you have in your company, the new thing is going to fail. Now, in general, I'm a huge believer that people are what drives the success of a business. So today we are going to talk about HR aspects. of AI. Now, some of you may think, okay, why does HR needs AI for? Like they're not generating any content or they're not doing any assessments and so on, which is not totally true. But the reality is that there are many tedious issues. AR tasks that AI can do a much, much faster than they're done today, saving the HR department dozens of hours a week and be more effectively than they're done today. So you're winning both on quality and on efficiency. Our guest today is Teresa Fessenstein is an expert in HR. She's been in, she's been in HR for 25 years. I hope she's not angry that I'm dating her, but she's been in VP positions in large corporate and in consulting. So holding various senior HR positions for 25 years. And in this past year, she is running an HR consulting firm that does help companies implement AI in their businesses. So she's the perfect person to have this conversation with. And hence, I'm really excited to welcome Teresa to the show. Teresa, welcome to Leveraging AI.
Theresa Fesinstine:Thank you so much, Isar. I'm so happy to be here. And even just in that introduction, in the context of HR and what HR leaders manage and do on a day to day as well as their teams. I think there's going to be a lot of clarity in our conversation today around the level of work that we do, the type of things that we're doing in the background often, and how AI is going to be a real tool that we can use to leverage getting into the work of supporting our people, which at the end of the day is what all HR leaders that I know really want to spend their time focusing on.
Isar Meitis:Agreed. So really, let's start on a very high level because again, you've been doing this on more or less every seat possible in the HR world for many years now, what are the key things in which AI being able to help and generate? Either a lot of value or a lot of efficiency or time savings. And let's start with the biggest boulders and then we can work our way down. And then we can talk about how to actually do that.
Theresa Fesinstine:Yeah, I think from just a context setting perspective. So everybody that, that works for an employer is a part of the people, team that works within an organization. And all of those individuals are teams that HR leaders are working day in and day out to support. So it's such an interesting and funny question to me that you asked that you posed earlier around why would HR teams find value in AI or utilize technology like AI, the reality is that HR teams house the bulk of the most critical information about an organization within our systems. So whether it's an HR IS system relative to hiring, promotion, payroll benefits, performance, all of that data is data that HR teams own and manage. And so it is certainly the way that I describe it to the people that I talk to, which I do all the time through education sessions or consulting work is that you know our ability through AI and leveraging the tools of AI. is so exponential in two specific facets. One is inefficiency, which we'll definitely talk about. And another is an insights, which not to go off on a tangent, but there's been years and years of transitioning and conversation around HR teams relative to using metrics and using data to validate our role, our function, our value. And the reality is that most HR leaders I know are not data analysts. They're not, we're not experts in, running pivot tables or creating macros and using tools the way that we needed to really get into the meat of data and the insights that will have an impact on our teams with artificial intelligence tools. We're able to do that. So you talk about the big boulders of tools in the, or functional areas within HR that we could use AI. And to me, that includes the area of HR PR, which includes marketing, event communications, management includes talent acquisition and retention, which is recruitment, job descriptions, job leveling, compensation decisions. It includes performance, which is performance management from a potential career development strategy and, succession planning, but it also includes things aligned with poor performance or corrective behaviors and helping validate and put together communication strategies around that, all the way through to engagement, which is one of the most critical processes that HR leaders really organize and launch within organizations. And it has traditionally been a part, a labor of love and just a labor of walking through rock beds and no, with no shoes on to get that data synthesized. And now, I found some really amazing tools beyond generative AI, but some really amazing tools in the last year to make that process so much more effective and actionable. Earlier, it used to take You know, it could be two, two and a half, three months, a quarter to go from a employee engagement survey to data analysis, to review, to implementing change, if that ever even happens. Now you can do that in potentially a week or even less to get from data to analysis. It's amazing. Yeah,
Isar Meitis:I agree with you. I think the And we're going to dive into details. Those of you like, Oh my God, this is a great overview, but how do I actually get started? we're going to dive into that in a second, but the biggest thing with regarding what you said is the hardest thing to do from a data analysis perspective until a year ago was qualitative data analysis. Quantitative data analysis was relatively easy. Like how many employees we have, how many salaries they get paid, who showed up on time, how many didn't, what is our churn rate, how many people respond to our open positions that we post on wherever, like the quantitative data. Yeah. Because every platform you go on has a dashboard that does the main things you want it to do. Qualitative data analysis. What people tell us in the interviews, what our employees tell us, what is the peer evaluation tell us about the employees? That's really hard because you need a lot of people reading a lot of information. That is. personal that they need to somehow analyze and put in different buckets to make any sense in it. And it's just a lot of manual work that is never consistent because you have three different people. They're going to think three different things on the same thing that was written by somebody.
Theresa Fesinstine:And I joke a lot, not to interrupt, but I joke a lot about, and every time I talk to HR leaders, I see that same spark of recognition. It's. It's without fail when I say I've wasted so many months and years of my life sitting on my couch with spreadsheets and a highlighter and like three bottles of wine, just trying to like navigate through, what are the common themes? And to your point, that's completely biased. I look at that data, even if it's, even if it's anonymized figure out who wrote what and why they wrote it like it's ridiculous to think that we are unbiased in the way that we view all of those experiences. There, there are some amazing tools there's one called inqqa that I was introduced to probably eight months ago, six months ago. the team there has created ways that HR leaders can take massive amounts of employee engagement data from I've worked with clients that are, that have 200 to 500 to 2000 question responses. And that was just not like one response or responses to one question. That's responses to multiple narrative questions, not just Likert scale, which. This is a little soapbox for me, but I think as an HR leader myself, I certainly fell in the category of"Do I add more Likert scale questions because it's easier for me to use the quantitative data and validate that?" Versus asking deeper, more complex questions that is just going to take me a hell of a lot longer to actually get through and try to analyze. We don't necessarily, we don't have to do that anymore. Tools like inqqa, even to some extent, generative AI tools, like the pro version of ChatGPT are able to provide those buckets. I think that the tools that are dedicated to that process are better at the outcomes and better to navigate, but you can do this with a lot of different tools and now you're not sitting there with this looming project sitting over your head, you know you go from being excited because you start to see 50 percent 70 percent 80 percent response rate on your survey, and then you start thinking. And pardon my language, bleep that if this, even though it's live, oh shit, I'm not going to be able to get through all of this data. How long is it going to take me? Tools are enabling us to actually get in, navigate and cross analyze the ways that we just couldn't before from incorporating if you're if you include this data, or you're pulling this data from other systems. Eight, 10 year position leveling and starting to truncate all that data combined with the feedback and experience that employees are having that gives you leverage in the areas you need to work on evaluating the mission and values that you work towards as a company, some of your ethos in which you're actually living the culture that you've created and align on whether it's the culture that you have set out to create. And if not, you're now working a lot faster to start to pivot or reinforce than you were able to ever in the past. It is truly one of the, I have a friend who says she calls them her AI aha moments. And so I'm gonna, I'm gonna steal that from her. that moment that I saw. what this tool could do was an absolute like floor drop, unbelievable, amazing. And it is a game changer for what HR teams are hoping to do and gathering insights and really getting to the core feelings and challenges that their employees are facing, especially now.
Isar Meitis:I love that. Great points. I'll summarize it in literally one sentence, which is quantitative data tells you what happened. Qualitative data tells you why it happens. And the why it happens is what allows you to make the right changes, right? Because otherwise you have to guess why things are happening to address them. And by analyzing what your employees, your prospects, like whatever the case may be, are saying to you, you can really start to understand why things are happening either for the good or the bad, right? It doesn't matter because it allows, and then you can immediately make better decisions, but I want to use that as a segue for us to go towards the right""The how", so what you mentioned one tool, but if there are other processes that can benefit from that, whether it's, writing job descriptions or evaluating, people's responses to the job descriptions, CVs or, whatever. What processes do you see that are really effective to use AI for, and what tools do you and your clients use to achieve that and how?
Theresa Fesinstine:Yeah, I think that there, from what I've seen, there are almost there's almost, there's so much authorship, editorial writing from, to your point, job descriptions. Let's just take that as a starting point. Most companies, especially if you're in the small to mid sized business range, you're probably writing about 20 job descriptions in a given year. Now, there are a lot of different tools you could use to write. Job descriptions. I don't believe that you need to, get one in different from evaluate employee engagement surveys with job descriptions in that process. you're even your free version of ChatgPT Claude AI, Perplexity. All of those tools can really provide value. The key for me in the job description process and why using AI is such a game changer is that it used to take easily. It could take days to create a job description. Because it seems like on the front end, wow, that's not that hard. Like it's just putting together some skills and putting together, but the process is. Especially with a new role, a lot of research goes into it. So in the past, we're sitting there going to Google, looking for similar roles, looking for templates on roles. All of that information is now contained in the large language models that these apps have been trained on. And so it can just siphon into and hone in on the key skills or roles that are similar to ones that you're looking to hire for. So building up a template of prompt in ChatGPT is a great way so you can indicate here are the sections that I always include in a job description that could be qualification skills, unique requirements, even projects that have been worked on in the past. We want to hire somebody that has done x type of Project projects in the past. and then being able to use that similar prompt every time you have a new job description, ensuring that you're asking it about comparable job compensation packages across different geographies. What the hiring requirements might be for people in similar roles. So you take these 20 job descriptions that you may have to do in a given year. You siphon that time frame from weeks or days into hours and minutes. For each section of the process and you're talking about saving exponential amounts of time. So I worked with one client where we literally recreated. We worked on it in the midpoint of last year as an experiment where she's I really want to redo all of my job descriptions because they're You know, they're disparate in style. They are, some of them are really outdated because it just takes so long to get them done. So we put together a mini prompt and I'm happy to share that in a future follow up or on LinkedIn in the comments under this event. We used a simple prompt that we crafted together, ensure that it contained the specifics for her company, including things like, what are their values? What is their mission? And so you can then start to ensure that the language becomes consistent around those values in a job description without having to say, here are our values. Like here, you can do that, but you don't, it narrates the tone and the insights around that. And we literally put in, Literally a list by list of just here's every job we have. We need you to develop a job description for each of these, take us through them one at a time so we can revise and edit. And it was literally, it took us a week. That would have, I've done projects for large companies, just a job, job description revision project. That's taken me six to eight months before, easily. And that took us a week to get her entire organization job descriptions in the same format, with the same characters, with the same style and tone, just that idea of creating that consistency of messaging is huge. And the managers appreciated it. They felt it gave them the chance to do quick revisions. On a project that they may not have hired for in a little while, but it's still important to keep things updated. So that's one example.
Isar Meitis:I think it's awesome. I want to pause you for a second and highlight a few really critical things you said from a practical perspective that everybody needs to understand. But to, to the things you said, two critical aspects of what you said that I think are very important. One- is you must have consistency like you can't have five different people write random prompts every time they need something because you're going to get different things every time, which doesn't make any sense in when you're looking at a company and its operation and so on. And there's two really easy ways to do that. Option number one is create a prompt template or a prompt library that you can host somewhere. I use a Chrome extension called magical Chrome extension to host all my prompts and it's awesome. But you can use a word document, a shared thing on SharePoint, a Slack channel, like somewhere centralized that everybody has access to that everybody's now going to use the same prompts. And the benefit from that is you develop that really detailed, long prompt. Once because these tools each and every one of them works much better the more context they have So if you explain to it, who's your company what industry how many people how long you've been in business? Where are you selling to who's your target persona? What are your core values? What are you trying to achieve? How are you hiring people like? Everything you can, but you don't want to reinvent that every single time. So option number one is really to have a prompt library and then do that. Option number two is to develop a GPT. So those of you who don't know what that is within ChatGPT, you can create a closed loop, very well tailored process to anything. And we create more and more of those for ourselves. We create more and more of those for our clients, but in there you can create yourself, you don't need any programming skills. It's literally like talking to Chachi PT, but you can create the process once. And now you don't need those centralized location because that becomes your machine generate in this particular case job description, and you can upload documents. You can upload your employee guidebook. You can upload your core values document. You can upload whatever you want as references for it and define exactly the process. And the cool thing about this, then you can open this to managers within the company. And say, Oh, you want to create a job description for a new position. You want to open, click on this button and follow the steps. It's going to take you. And they will do it perfectly because it won't give them much options. Like it will literally just walk them through the process. Can I add on to that, Isar? Absolutely,
Theresa Fesinstine:that's why we're here. I love, yeah, I love this conversation about prompting. I, there are a few caveats I would make to Isar's point, which is number one, you do have to have a pro or team or enterprise version of ChatGPT in order to be able to build out a GPT. Now, it literally could be, And I would argue the best money you would spend for your company in terms of value that investment brings if you are. And just for those that don't know, team versions are from 0 to 100. 1 to 149 people. Enterprises above that, for teams, it is 25 U. S. Dollars per person. but the value that Isar is mentioning in terms of being able to share those GPTs is exponential. It is, if that were the value alone, it would be worth the money that you're spending because you can build them on anything from how do building your job descriptions, creating guidelines and training for performance management training and building out performance reviews. You could do it for events. One of the things that I thought was magical, and I worked with a different client to put this together is one of the time sucks that we want to do, but is just a time suck and getting through every year is putting together the events and the gatherings and the celebrations and the acknowledgements and all of those different events in a given year for a company. Again, it seems like a simple thing that like,"Oh, we just put an event together". But there's a lot of logistics, checklists, communication, that waterfall outside of just planning what you're going to do for women's history month. I work with one client, we literally created a GPT for events. And in the beginning of the year, we literally built out every event, every communication plan, every checklist for an event that we need, including anniversaries, including major holidays, including fun summer outings that we needed to do for the entire year. Her plan is done. It's booked. She doesn't have to take hours. On a weekly or bi weekly basis to think about,"Oh shit. What's the next event that's coming up." So that's another illustrative example of how you could use generative AI. The other thing I want to talk about it, is
Isar Meitis:really about the prompt. I want to pause you for two seconds for two different things. One, we've done an episode with a person that plans her external events, like marketing and sales events. And she does it, it's episode 31 of the podcast. And she's talking about, okay, planning an event for your clients and prospects is a huge undertaking. So companies, if they do them like twice a year because the headache is insane. And now she's taking the planning process. We literally in an hour off a live show that we did. So again, episode 31, if you want to go back to that, walked us step by step on a complete planning on an event that used to take her weeks. So she's going,"Oh, okay. That's a fake event. It's not real.", but okay. So instead of weeks, it's two days. Not an hour. It's still insane. So yeah, planning events is a huge deal. But I wanna go to questions. there's a question, and I think it's a very interesting question, from, Jarek Rivers, from LinkedIn, and he's asking, Teresa, you said that when evaluating candidates, there's a personal bias. And I, and that's obviously true. it doesn't matter whether it's candidates or internal, reviews and so on. But then he's asking, doesn't using automation reflect the program slash algorithm bias, or in the case of AI, it's training set?
Theresa Fesinstine:Great question. That is not only a great question, but the answer is yes. I think that there is, I can't deny that ISAR can't deny it, that the models that are trained by humans and humans are biased in their nature. One of the things that I spend a lot of time doing is trying to encourage, endorse, amplify, creating more diversity in the world of technology development, certainly in the world of AI. A few things that I'm doing personally is I support, women in it, women of AI, women defining AI, and I'm a mentor for the organization, All Tech is Human, which is mission oriented to help create diversity in the place and space of technology. And so I, there are certain things I, as a practitioner and as a leader and somebody who's talking to organizations and individuals around AI, will not deny AI is biased. People can do bad things with AI. I would argue that humans are, some humans are nefarious by nature. And they will find a way to be implement their bias or implement their nefarious acts, regardless of whether they can do it faster through technology. So I don't disagree with you. I think there are organizations, there's a company from, it's a recruiting, AI recruiting organization called Searchlight, and they, are working very diligently to remove bias from that recruiting process. Other things that are happening in the world, I am based in Norwalk, Connecticut, but worked out of New York City for 20 years. New York City in New York has created a requirement in a law where employers that are using artificial intelligence tools must be transparent about that. I see that happening and expanding across the US. And I also would say that the transparency on AI even in other facets of the way we might use it in the U. S. will probably start to mimic a little bit more of the E. U. A. I. Laws that went into effect at the end of last year. So I think governance is going to be something to watch, but circling back to the main question, I don't disagree with you.
Isar Meitis:So i'll add my two cents and it relates to first of all, I agree with everything you said But I think what we said before Yarek to your question if you add your company's core values if you add what's really important to you If you define the way you want to evaluate it, if you give it a Plan like a blueprint. This is how I want you to work in order to evaluate these. It will work according to the blueprint. And yes, there's going to be some leeway to what it has learned from the X number of gazillions pieces of data that it acquired in the past, but when you give it very specific instructions, and as I mentioned before, the more context you're going to give it, the more it's going to be the results you want it General responses from the model as it comes. And so that helps. It makes a huge difference. So if you provide it, the stuff that you the lens through which you wanted to evaluate employees or reviews and so on, it will use that lens. And so you're moving further and further away from the biases. Off the general model as is, but yes, there's still, yes, there's biases.
Theresa Fesinstine:I love that you said that because I, what I was going to touch on before, has to do with the prompting process and the process of how you build and strategize your approach to talking or chatting with any generative AI tool. So ISAR talked about having, standardized prompts, especially if you're working on a project across multiple people. So there's consistency and output. I am not personally, I've done training and development in HR for 25 plus years. I've been a sales trainer. I was never one for scripts. I'm one for structure. So I use a six point structure for recommending prompting within any of these generative AI tools. The one thing I will say, and this is quickly becoming my favorite story. If you're asking a very simple question, like the other day, my husband and I were having an argument around how many eggs there should be on a fried egg sandwich, like an egg sandwich. And so I asked, I literally went on my app. I asked out loud, ChatGPT. I was like,"How many eggs go on an egg sandwich?" My contention was two. His contention was one. I was like, you're crazy. There's no way. And you don't need to go through a complex prompt in order to find out how many eggs belong in an egg sandwich. But if you're working through something more complex, if you're trying to design a process, or you're creating a new policy within your organization or procedure, or you are launching a new product or a new HR Opportunity or benefit. You might want to use this structure and I'll say it very quickly. And then if you take over chatting, I might just plug it into the chat. I believe personally and have seen studies. There was a report that was done from Microsoft William and Mary, some private researchers and a university in Beijing that did a lot of analysis on emotional prompting, which is it the conversation around how much do you humanize, or anthropomorphize technology. I personally do. I love to see chat as a trusted advisor. I communicate with it as such to do that. I welcome it. I say, thank you for helping me today. I give it a role. So who is the chat and who am I in this conversation? So I'm setting the stage as Isar mentioned. I talk about, the, tone of which I want that conversation to have. you can teach ChatGPT or Claude or Bing or Bart or any Gemini, you can teach all of these tools to sound like you by simply uploading some historic documents that you've written and ask it to define what that sounds like. Copy that, paste it. Every time you write a prompt, if you want it to sound like you, you can also ask it to have the tone of Brene Brown, Priya Parker, Adam Grant, whoever you follow as a, as someone, you look up to from a professional perspective. I also ask it to give me, specify the ask. So what specifically do I look for in the results that they're going to give me? Is it a image? Is it a chart? Is it a structure with certain headings and bullet points? Do I want narrative? Do I want, more clinical responses? I then challenge it to ask me questions. So what questions do you have for me? In that ask, I can also speaking what we were just talking about, ISAR, I can also define that I want it to avoid specific, discriminatory language to be reflective of the, the, demographics of the area with which I'm working in or my organization. And then finally, and here's a big one. I not only say thank you, but studies show that if you encourage Your generative AI tool, it will perform better up to potentially 20 percent better. is what this study came up with. So some people don't want to humanize their technology. And for that, I say, great, just leave off the welcome and the thank you. For me, I find the results so great. And I'm curious about your experience with that same idea, Isar.
Isar Meitis:So I'm very much like you, A, because that's how I communicate and I communicate with it. And when I, it comes naturally to me to say thank you and welcome, and this was great and so on. But also I've seen a lot of research that's proving that when you're giving it compliments, it's going to perform better. So that's one, I will say something that summarizes a lot of what you said. I tell my clients and in the courses that I teach, you got to treat AI as if it's the best intern you've ever had, but it's still an intern. Meaning if you think you can call the intern into the room and say, Hey, go and build me this plan and you're going to get a good result. You're hallucinating and you will never do that. Nobody will ever in the right mind actually come and suggest, right? That, I will bring this intern in, I will give them one sentence as instructions and then I'm sure they'll do like an amazing job. They won't, you will never do that. You will sit with them for an hour and a half. Or have assigned somebody to sit with them for an hour and a half to explain to them. What is the project? What are we trying to do? Why are we trying to do this? Here's where you get the data. Here's our things that we've done in the past that are similar to this. You should probably research these four ways. Like you will give them something to work with. And if it's a very good intern, you will get good results. So if you, and you'll be nice to them, you're not going to go do this. no, you're going to be, Hey, Tom or Lisa or Gina or whatever. Say, Hey, this is your first day. I want to help you get this right. I really want you to have a great time doing this and I think we can get amazing results together. That's how you're going to start. Start the same way here. You're going to be perfectly fine. So if you think about this as the best intern that ever existed, everything else will come natural. So that's just me summarizing all the great points that you said. I want to jump back to, the practical stuff. So we talked about writing a job, job descriptions, what other tasks that you and your clients are doing that are extremely effective and what tools are you using to do those?
Theresa Fesinstine:Sure. So there are one of the Another administrative time grabber for HR leaders is just the volume of questions that we get on a day to day basis around the documents that already exist in our workplaces. And if people were able to just get a quick answer from those documents, they probably would save just, I haven't done a thorough study. I'm sure my friends at, Harriet would be able to help me. but. using tools that can actually serve as the front line when those repetitive questions come in. It is one of the most time consuming parts of the process of that or, working with employees when they find out that they're going to be expecting a baby and what is the FMLA process or what are the leave policies or, caring for a loved one or bereavement time. And so all of these questions are just It's not that you don't want to be interrupted, but they can end up taking a long time and that information is readily available. So there are tools like Harriet, the website is hrharriet, and The team there has crafted a tool that literally can sit on top of your works in your slack channel. So a place where most people, most companies already are communicating and employees can go into that and ask into that channel and ask questions around not only questions on the handbook, because Harriet syncs with different HRAS technologies, it can actually be. answer the question and take action. So you want to book two days off. You want to check how much PTO time you have off and how much you have left. Sorry. And then ask for two days in the future. HR Harriet can go in, check out what your policy is, inform you of how many days you have left. Submit the request for the time off and all of that is happening asynchronously with tools that allow you to then focus on what you need to do. So it saves you the time from the conversation with HR, although everybody wants to take out, talk to HR, so I don't want to take that away from you. But, and it also saves HR the time of going through those same repetitive questions where the information is readily available. The systems are equipped to do it. But. But, to be candid, sometimes people forget, or they just don't want to flip through the handbook or they don't know where in the handbook it is. And it's just easier to do outreach. And I, the other thing with that is the way that I thought about, I could go into a whole conversation with you, Esar, about the fact that open AI has started calling them, custom GPTs, as opposed to chatbots and like the perception of chatbots and the fact that, it's Custom GPTs can do a lot more than just a chatbot, but there is a, I think, negative perception around chatbots. What Harriet is and what some of these other tools do, they're very different. They are conversational in the same way that ChatGPT is. It can sound like and feel a little magical because it does feel like there is a human on the other end responding to you. But it is just that predictive, capability to access data and to communicate and build create conversation that generative AI has that creates this really great interactive experience. Employees don't feel like they're left out talking, their HR team doesn't want to share information or talk to them, and the HR team gets so much of that valuable time back so that they can focus in on what are the common challenges we're facing not sitting on the phone telling somebody they have three days of vacation left and you need to put it into Paycom or whatever system you're using at that
Isar Meitis:given time? Yeah, I think it's awesome. I, in general, I think I'm gonna because you touched on it yourself before and I think it would be a good way to Summarize and put some context to what you said. You can use tools like ChatGPT or building your own internal bot on a GPT, which probably not going to be enough. GPT is limited with how many documents it can upload, but you can use tools like Dante or Chatbase to build chatbots. the biggest difference between these tools are just generic tools like ChatGPT, Gemini and Dante to like a Harriet. Is the level of understanding of the subject matter and the level of integrations it's going to have to the relevant systems. So if you're in marketing, those very tailored systems will connect to your CRM, into your email automation, into stuff like that. If you're in HR, it will connect to your HR systems and so on. And same thing with finance, et cetera. So you can do a lot and at least experiment and test with other Free or almost free tools like ChatGPT or Gemini. But if you want to go, and by the way, I'll say one thing that is very important because otherwise they'll forget if, because you praised getting the team's level of ChatGPT, the other huge benefit of getting the team's level of ChatGPT, which is 25 a month is that it doesn't train on your data. So it keeps your data completely separated from their training, which is not the case if you're just using the regular, ChatGPT level. So it's definitely worth doing. But all I'm saying is if you have the team's version, you can now work with, So you can use that with your data in ChatGPT, just to test it out and see load some of the documents and let people use that like that GPT. And if it works and you're seeing that it's actually helping you now go and look for solutions like Harriet that are more tailored, more time, more money, but will give you a lot of additional benefits. But you can do a proof of concept in a day, like literally give somebody to develop that GPT for you that understands what a GPT is and how to create it. So test a little bit, share it with not all the employees with. 30 and see what happens. and you have a use case and now you can go to your leadership team and say, Hey, this is what we've tested. This is what we found out. It saved us 30 minutes every single day of not answering these emails. So should we invest in this, but now we'll also do one, two, three, and four. And that's a great way to get in the door and get budgets in order to do the slightly bigger, more expensive, but a lot more effective tools that are tailored to a specific thing.
Theresa Fesinstine:Yeah, I love that. I think the other thing that I, that. When we go back to the concept of prompting, it, it also illustrates the more detailed you can give, the more specific of a role you can give your gen generative ai. I look at kind of these tools as almost like a library. That's an analogy I use. And the language model is all the books that continue to come in and train and learn and that knowledge base is growing. What. specifying or using either specified tools or customizing a GPT through a paid version. What that does is it starts to get you into a specific area of the library. So it helps it filter that information and I think that's like a good visual concept of Using really more thoughtfully created prompts, in order to get the result you want. And if you take a few seconds before you get started, and I'm going to do a throwback to Stephen Covey, who I used to listen to in long car rides when I was, on the road, doing training and development many years ago, beginning with the end in mind, what output are you really looking for? The more generic you put in, you will get a generic response and it'll probably be pretty good. But it's going to take more revision more adaptation more work if you just put a few seconds in on the beginning and seconds it's not a long time to craft a prompt, you're going to really get that targeting into the library and look at the books and the volumes that you need and you're going to get an output that's going to be. Pretty, pretty spot on. That being said, please always review and edit if needed your outputs. The last I think the last metric that I saw was that 40 percent of people are taking exactly what they get out of ChatGPT and using it to Directly. And I always laugh that, okay. So I have a funny little analogy story. Sorry, if we have time yesterday, I was running a session for in my local community, our chamber of commerce, which I'm so proud and happy that our little town of Norwalk, Connecticut is interested in AI and growing. And during that session, I was saying, Oh, I can always tell when somebody is using ChatGPT because it uses words like supercharge and that's just not how
Isar Meitis:we like, Who the hell heard of the word delve before ChatGPT? He don't
Theresa Fesinstine:use those words. And then I kid you not like 20 minutes later, the MC used the word supercharge in terms of AI. And I thought, you know what? Okay. There's the exception. Now he either always used it or he started using it after he got familiar with ChatGPT. just be conscious, look at what the output is, even if it looks spot on, give it a real thorough review and then make sure that, it, it really does contain in sounds. The way that you want it to sound and it's just a common thought. I'm not sure if so much of what HR is involved in, sorry, I would feel remiss if I didn't mention it. So much of the data, the insights, the information that we have is confidential. And so the other thing I always like to say is if you're using an unpaid version of any tool, assume that information is going to be Pulled in for training or for insight gathering or for just out on the web. So you want to make sure and, redact any identifying information. That doesn't mean that you can't use a ChatGPT or ChatGPT plus individual version to do a employee engagement analysis. What it does mean is that you'd want to redact last names and just pull a column out. you wouldn't want any, intellectual property that has, isn't already in public domain. pull that out. You don't want to add that into a place where you don't want it to show up elsewhere. Now, I also say that there's a lot of data in there. So the likelihood of it showing up might be slim, but it's better to be safe than sorry. No, I
Isar Meitis:agree with you a hundred percent. Great recommendations. I want to talk to you on one last thing. We have a little more time and it's something that I, that is dear to my heart. And it's what we started with, in the beginning, which is analyzing qualitative data. Now, HR does a lot of that, right? So there's. Whether it's engagement assessments or peer to peer assessments or peer reviews, whatever each company calls it, different things or company reviews, like how much do you like your job or manager reviews to its employees that you want to evaluate and go through as HR and so on. What tools do you use and what processes do you use? Because this becomes. And the bigger the company, obviously the bigger the headache becomes. It becomes a headache because it's hundreds and sometimes thousands of items to read and analyze and review. So what tools are out there that you are recommending or that you are using or your clients are using in order to do that part of the process?
Theresa Fesinstine:I'm going to be redundant because the number one tool that I recommend for doing that is inqqa, which is I N Q A. it is built for. Analyzing narrative editorial data, in the native language of the speaker, which I think is a huge thing, especially for an engagement review or a performance review. When you have a global company where managers and employees may be in different locations and may not. speak natively the same language to be able to get those reviews and to analyze that data in the native language of the speaker. That's may sound like a simple thing, but that is a truly dynamic and important facet for me. I used to manage a small team in India and everything went through a translation tool and you're never really sure if it's translating correctly. So That's something I also like we said, I don't have my finger on the pulse of this is the perfect tool for performance management. My, I understand that a lot of the tools that are really great at the evaluation or the, Sending out the survey. So that could be a lattice of 15 five culture and all of these different tools are starting to integrate AI. That being said, what I've seen of their integrations to be completely transparent has not been that great. And in fact, it's almost a rebranding of the same AI elements that they had before, but now that the conversation has become more relevant, they are expanding on that. which is why there are, Searchlight, inqqa, A few other tools that, tools like free fuse. If you're doing training L and D, and you're having to run a training session on performance reviews, being able to trunk eight larger videos into learning paths, that is an Epic. An epic win for somebody like myself who was in L and D and spent a lot of times re framing or read, recording myself doing something. 11 labs has a phenomenal tool where you can create your own avatar And just write in text and it will do the, build the video for you. So there are a lot of little tools. I have a one sheet or so one slide in my deck that I use that is that I say, if you're an HR and you're exploring AI, here are the tools you want to take a look at. I'm happy to share that as well. Or people can go to my website and find that information.
Isar Meitis:But yeah, that would be awesome. if you can send me a link to that, I'll put it in the show notes. And anybody who listens to this afterwards can find it. And we'll put it also in the description on. You're like, Oh, every place you're going to listen to this after the live, we'll put the link to that. I'll add one more thing that I find. And I use more and more both myself. And again, with my clients, before getting into the more specific tools for HR or any other topic is I use ChatGPT or other links to other, so APIs of A large language models within Google sheets. So you can do this by downloading an extension, GPT for sheets and docs, and it costs a little bit of money to use it every time you use it, or you with very little technical skills, I'm not a techie, I'm a business person and I. Was able to have ChachiPT create a script for me, like an app script that I can run within Google Sheets that then brings the ChachiPT API into Google Sheets without no, and if you Google that, you'll find 50 examples on exactly how to do that. So you don't have to do what I did, which I just experimented back and forth trying to get myself and ChachiPT to do it, but you can get, here's the final product. This is what you need to copy and paste into Google Sheets. And then you have But then the cool thing is you can take 5, 000 rows. And you can ask a question and then copy down to 5, 000 rows or highlight all the 5, 000 rows and say, find me. And again, you need to write a long prompt and describe properly exactly what you're trying to do and so on. Because at the end of the day, it sends an API call. It's like running ChachiPT. But you write the prompt, but the final thing, like the main thing you wanted to do is find me the three most negative things that people say about our company. And he will go through 5, 000 lines. And we'll find the three common things that come back again and again, and it will give you a summary. And it's something that is literally impossible to do by a human. Because understanding the references and the nuances of one person said it this way, one person said it that way, But an understanding how it works within the sentence or the other bigger things that they were saying through 5, 000 reviews is just not doable. And now you do this in Google Sheets, like it's nothing fancy. It costs, again, 20 bucks, not even because you're connecting to the API. So it depends on how much you're consuming it. and you're asking that question on 5, 000 items and you're getting an answer within 30 seconds. It's incredible. So being able to do these kinds of things, as we mentioned, either with the existing tools and you have APIs and I know API is a really scary word to a lot of people, myself included, but if you Google that, you'll see that it's a very simple process. You go to the website, you ask them for an API key. It gives you like a very long string of weird numbers and digits. And you copy that and you. Put it somewhere else where it asks you for your API and you're done. So it's really that simple. So there's no reason to be afraid of that. and the results that you can get on your own without any fancy tool, just by using Google sheets is. Mind blowing. And so for any evaluation process of qualitative data, this is now available to anyone for almost free again, 2025 bucks a month. there's one interesting question from, the LinkedIn group before I let you go. So Tina is asking who won on how many fried eggs belong on an egg sandwich. So I think people really want to know.
Theresa Fesinstine:My husband. One egg. ChatGPT said one egg. How do you, okay, how do you just have one egg on a Friday, which like, it's not enough egg. There's too much bread. I'm going to end on something that because of what you just said, Isar, about, we started with dating me by saying that I've been in HR for 25 years, which is in fact true. I am going to turn 50 in a month and a half. And I was just thinking you were talking about the code that, the API code that you get, and it got me thinking about the first email address I ever got, which was when I was 20 years old, graduating college post mail. My uni no. This was my university's internal Oh. Email service before I even knew of Gmail before I, or Hotmail or any of the other ones. yeah. And it was probably 30 characters long at, Memphis. com or edu. And I, it, the reason that like that long string reminded me of that. And it just, I think it's important to reflect on the fact that number one, if you are watching this podcast for the first time or listening to it for the first time, and you have never gone into ChatGPT, and this term seems overwhelming. I talked, I did a session, as I said, yesterday, probably 30 percent of the room has never opened ChatGPT. 30 percent of the room is in it multiple times a day. We are all on a learning curve. We are all navigating it together just because he saw her and I are talking about it. We've just been in it for a little bit longer. And so from my perspective, as from a mentorship perspective, training and flattening the AI learning curve for people that are non technical and HR teams. It's so important to recognize that you are exactly where you should be. Your learning curve is exactly where it should be. And, the world of, that we live in has changed dramatically in the past 25 years, I've been alive. and it's going to continue to do So the quicker that we embrace the curiosity and the butterflies in our stomach, the more advanced where our, all of our voices are going to be in this really seismic shift of the way we operate. I wanna also say thank you for having me today, Issa.
Isar Meitis:first of all, this was awesome. I really enjoyed the conversation. I know that a lot of people found it valuable, like people are doing, like all the clapping sim symbols and so on, on LinkedIn, every time you're speaking. So that's great. You have a secret admirer. Steve says you don't look a minute over 40, so go look him up afterwards in the comments, so thank you Steve. you said that. If people wanna follow you,
Theresa Fesinstine:Hey Steve, how many eggs belong? How many eggs belong on an egg sandwich? Steve, that might be the kicker.
Isar Meitis:Oh, that's awesome. If people wanna follow you, work with you, learn from you, like what are the best ways to connect with you? So on the wall
Theresa Fesinstine:behind you, LinkedIn? Yep. I am on LinkedIn. I am. I'm linked in a lot. I believe in it as a platform for education, training and connection under my name, Teresa Festenstein, which is in the lower left hand corner of my screen. I have a website www. peoplepower. ai. And most importantly, I'll share the link with you, Isar. I run what I call AI quick clinics every three weeks for 30 minutes, no more. and the whole purpose of that is to get people talking on a more consistent basis and exploring. We go through tools. I have guest speakers. and we talk for about 25 minutes. And share and then leave it open to questions. It is called AI quick clinic. If you want to look that up on LinkedIn, you can jump directly in there actually did a post about it today. because I have a Calendly link where you can just go in and pick the dates that you want to do it. My next one is March 27th. And, I will happily work with you, talk with you, connect with you. If your company is exploring or starting to navigate the conversation around AI, I do join organizations. I do education and consulting and would love to get to know you.
Isar Meitis:Awesome. Teresa, this was. Fascinating and fun, more than I expected and I had high expectations. So thank you so much. I'm very happy that
Theresa Fesinstine:you joined us today. Absolutely. Thank you Isar. Thank you to the team that's here watching.
Isar Meitis:And thanks everyone who joined us on LinkedIn on and on, Zoom, and we'll see you next week with another live session. So those of you who want to join us, Just keep on clicking join to those events. We keep on bringing fascinating people, and having these really important conversations on how non technical people can actually use AI in various aspects of the business. Thank you so much. Have a great rest of your day.