In this episode of Executive Connect, Melissa talks with AI strategist and advisor Kavita Ganesan about why so many AI projects fail, what leaders misunderstand about AI readiness, and how to make AI deliver real business value. Kavita explains what she sees when companies rush into AI without a plan, how to tell whether you have an AI problem or a business problem, and why many expensive AI initiatives never make it to production. This episode is for executives, founders, and business leaders who want to make smarter AI decisions, avoid wasting time and money, and focus on practical tools that improve revenue, efficiency, and ROI.

Chapters:

(00:32) Meet Kavita Ganesan

(02:00) What Brings Companies To AI

(04:06) The Biggest Mistakes Leaders Make

(05:14) Why Teams Get Stuck In Confusion

(05:52) Better Questions Before Implementation

(06:30) Experimenting With AI Vs Real ROI

(07:19) What Successful AI Companies Do Differently

(09:48) Business Problems Labeled As AI

(11:31) How To Find The Real Problem

(12:34) Early Signs An AI Project Will Fail

(15:03) How To Stop Scope Creep Early

(16:55) What To Do When Projects Go Sideways

(18:38) Where AI Can Save Money Today

(23:22) Using AI For Compliance And Training

(26:50) Where CEOs Should Start First

(28:31) How AI Can Support Sales Teams

(30:18) The Metrics That Actually Matter

(31:54) What To Do When Results Plateau

(33:48) The Biggest AI Myths Executives Believe

(35:46) How To Know If Your Company Is AI Ready

(38:06) The Simplest Place To Begin

(39:21) How To Explore Before Overspending

(40:50) Final Advice For Leaders

Kavita

(0:00) I think looking deeper into the problem itself will tell you what really the issues are. (0:07) The thing is now they kind of hear the problem then they label it as automation, then they label it as AI or something else. (0:15) But you need to really peek deep into the problem.

(0:18) It could be that overall it’s a customer service type issue, but under the hood why customer service is struggling is because right now every executive feels the same pressure.

Melissa

(0:32) You’re told AI will change everything. (0:35) Your competitors are quote-unquote doing AI. (0:39) Your board is asking you what your strategy is, but quietly many leaders are thinking the same thing.

(0:47) What’s real and what’s just expensive noise? (0:51) Today’s guest has made a career out of answering those questions. (0:56) Gavita Ganesan is a global recognized AI advisor, strategist, and entrepreneur who knows organizations move beyond pilots and proof of concept to AI systems that actually deliver the business value.

(1:12) She’s the founder of Opinion Analytics, has led AI initiatives across governments, healthcare, manufacturing, and tech, and has worked with organizations like McKesson, GitHub, eBay, and the Nuclear Regulatory Commission. (1:28) She’s also authored the Business Case for AI book, a practical guide for executives who care less about hype and more about ROI. (1:39) This episode isn’t about chasing trends.

(1:42) It’s about making AI earn a seat at the table. (1:46) Welcome, Gavita. (1:49) Thank you for having me, Melissa.

(1:51) Now, when organizations come to you, what usually happens inside the business at that exact moment?

Kavita

(2:00) So, there are a couple of categories of companies that come to me. (2:04) So, one category is they’ve kind of used ChatGPT, they’ve used a few other AI tools, and they kind of know what it is. (2:14) So, they use it in different areas of their business, and they see potential in it.

(2:19) And now, they want to know how they can go further with these tools, because these tools are not customized to their business problems. (2:27) But at the same time, they are really confused, and it’s not really even clear if they need an engagement. (2:36) So, they just kind of want to know what’s at stake for them.

(2:39) So, that’s one category of clients. (2:42) Then, I have another category of clients. (2:44) They come in, they’re all gung-ho, they want AI.

(2:47) They want to become AI first. (2:49) They want to apply AI anywhere and everywhere they could. (2:54) So, that’s another category.

(2:58) And in fact, there’s one funny story. (3:01) A client sends me a message saying, I want AI, I want sales. (3:06) I’m like, okay, if you’re not already having sales, that’s more of a business problem, rather than an AI problem.

(3:14) So, I get a variety of, like a mix of those two clients. (3:19) And in both cases, they kind of think they know AI, but they don’t know AI beyond these ChatGPT, Gemini type of Gen AI tools. (3:28) So, they don’t understand the landscape.

(3:31) So, the expectations from those technologies is also not quite aligned with where things can really go.

Melissa

(3:41) Yeah. (3:42) I love that. (3:43) And it’s such a common thing these days.

(3:45) And it is a funny story because people think AI is going to solve literally everything, including their revenue. (3:52) And often, we got to figure out the revenue part first. (3:57) So, what are some of the biggest mistakes leaders make when they try to, quote unquote, do AI on their own?

Kavita

(4:06) Well, when they’re trying to do AI on their own, they kind of think that AI is going to work in specific areas of their business. (4:14) So, maybe an employee has a cool idea that we can use AI to do all this marketing work. (4:20) So, they start using AI, they start getting really deep into those workflows.

(4:25) But then after a few months, they realize, okay, so what are we really getting from this? (4:30) How much productivity improvement are we really seeing from this? (4:35) In fact, we may be spending more time correcting, iterating over the images being generated, or doing things to correct the AI slot that’s been created.

(4:47) So, then they try to take a step back and think through, okay, maybe this is not really the best use of AI. (4:55) And that’s when some of them start reaching out to consultants to see, hey, how should we be doing this correctly?

Melissa

(5:02) It’s usually funny. (5:04) How does unclear business ownership sabotage this momentum when they’re deploying whatever it is they want to deploy?

Kavita

(5:14) I wouldn’t say it’s sabotaging momentum. (5:17) It just puts them in a loop of confusion. (5:20) One employee thinks that this is the way to go.

(5:23) Another employee thinks that is another way to go. (5:25) So, they are stuck in this loop and they’re unable to get out of it. (5:30) So, that’s when they really need some expert guidance to get them out of that and focus on what they really need to be doing.

Melissa

(5:41) And so, should they start with reframing these kind of questions that are taking them in a loop? (5:48) Maybe it’s just they need to ask better questions.

Kavita

(5:52) I would say they first need to understand the tools that they’re using. (5:58) What are the quirks of these tools? (6:00) Where does it really work?

(6:02) Where does it not work? (6:03) And then maybe have some early ideas. (6:07) And then start framing those ideas into maybe what are the real benefits to the business.

(6:14) So, start jotting down all your ideas. (6:17) What are the benefits? (6:18) And which ones actually would move the needle for the business?

(6:22) So, those are early questions that you should be looking at instead of jumping right in and starting implementation.

Melissa

(6:30) Yeah. (6:30) I think that’s one of the most important keys in growing any business is understanding the vision of what the company is about. (6:37) What are we offering?

(6:38) What do we want our customers to experience? (6:41) And so, let’s talk a little bit about experimenting with AI versus getting an ROI on the money we’re putting into the tools that we’re buying. (6:53) So, what separates the companies that get the real ROI from those that just stay in piloting or experimentation forever?

(7:03) Ready to lead smarter and invest wiser? (7:07) On the Executive Connect podcast, we unpack executive strategies for wealth and influence. (7:14) Hit the subscribe button now.

(7:17) Don’t just watch, act.

Kavita

(7:19) So, there are a few things. (7:20) So, the companies that actually see something valuable from AI, they are not looking for shiny objects. (7:29) So, they have like a real problem and they want to solve it.

(7:33) And then they reach out to the right person to get it implemented. (7:37) So, let me give you an example. (7:38) There was a startup client that I worked with a few years ago.

(7:42) So, he had a really revolutionary idea for grocery e-commerce shopping. (7:48) And he could have used AI in a lot of different areas, but he kind of like separated the different areas that needed AI and the types of AI. (7:57) So, one area was NLP, natural language processing.

(8:01) So, this is an area that I am like, I studied in and I have expertise in. (8:06) So, he reached out to me for the NLP piece. (8:09) And we built out the whole pipeline without any LLMs or any sophisticated model.

(8:17) And it works right out of the box. (8:20) It’s lightweight. (8:21) It doesn’t need any additional infrastructure and it gets him off the ground launching his product.

(8:28) So, he didn’t worry about whether it had LLM or not. (8:33) But a lot of businesses worry about that. (8:35) They want to integrate LLM when it may not even be needed.

(8:40) And so, he got his product launched and he said the algorithm works really well. (8:46) And over the years, you can see he can actually switch things out. (8:49) He can improve what he has right now with something more sophisticated, but this really gets him off the ground.

(8:54) So, he has seen ROI from his AI investment, even though he’s a startup. (9:00) And a lot of companies that follow this same playbook, they see the ROI. (9:07) And another group of companies that see ROI are those that listen.

(9:12) So, some companies, they think that they already know AI, they just want to do what they want to do. (9:17) So, they don’t want to take expert advice and they just hire software developers who may not have like solid AI background. (9:27) And they start building tools over these LLM technologies.

(9:31) And then they start implementing them. (9:32) And then they say, oops, it provided the wrong answer. (9:35) And then it resulted in a lawsuit.

(9:37) Then they start stripping back all the tools that they developed. (9:41) So, it’s companies that listen and companies that are very strategic about how they want to integrate AI.

Melissa

(9:48) And I agree. (9:48) And I love kind of this, I find who has accountability is really key. (9:55) And so, a lot of times with any of these digital transformation or AI projects or technology initiatives, usually the blame is usually on the IT departments or the CTO.

(10:09) But when problems aren’t actually tech problems, they’re business problems. (10:14) What business problems do leaders mistakenly label as AI or tech problems?

Kavita

(10:22) Everything. (10:24) I’ve seen simple scraping being labeled as AI, just search problems, simple search problems being labeled as LLM problems. (10:35) So, anything that kind of looks like automation is being labeled as AI and agents and agentic AI.

(10:44) But really, there is a whole spectrum of software solutions at the very, the most simple version is just a simple script. (10:57) Then at the most complex end is this agentic AI idea. (11:04) So, most companies need things at the other end on the left side, not on the right side, but they try to go for the right side.

Melissa

(11:12) Now, do you find that things are often misdiagnosed or operational bottlenecks are part of the problem, or is it silos? (11:24) Where can they be doing better with not just blaming them as a tech problem?

Kavita

(11:31) I think looking deeper into the problem itself will tell you what really the issues are. (11:38) The thing is, now they kind of hear the problem, then they label it as automation, then they label it as AI or something else. (11:47) But you need to really peek deep into the problem.

(11:50) It could be that overall it’s a customer service type issue, but under the hood, why customer service is struggling is because maybe there are some tools that are too slow and it’s not providing the answers that are needed. (12:05) So, then you look into the tool. (12:07) So, maybe it does have AI in the tool, but maybe it’s introducing friction in the agent’s workflow.

(12:15) So, maybe that needs to be improved. (12:17) So, really going deeper and peeling back the layers will reveal the true problems. (12:22) It might not be a single problem.

(12:24) It could be a series of different problems from the leadership side to a tech problem. (12:30) So, all of these need to be solved in pieces.

Melissa

(12:34) Now, I heard some statistics and I’ve seen many different articles recently on the number of AI initiatives that fail. (12:43) I’ve heard it’s always over 80%. (12:45) So, I don’t know if that’s still true to the case or not, but what are the early signs that an AI effort is about to stall, go backwards, collapse, or lose funding?

Kavita

(12:58) Yeah, so many. (13:00) I would say it’s closer to 90% to 95% of AI projects that go sideways today. (13:07) Maybe a few years ago, it was 80-85%, but now I’d say 90-95%.

(13:14) And the first one is just complete misunderstanding of AI capabilities. (13:22) So, let me give you an example. (13:24) So, I had a client who wanted to analyze their client’s financial statements and then look for gaps and then make recommendations on how those gaps can be addressed, perform reconciliation, and do all sorts of financial-type work using a tool like ChatGPT.

(13:44) So, they want to use the APIs from ChatGPT to do that. (13:48) Now, this is – and they label the whole thing as an AI problem, where, in fact, it’s a series of different problems that need to be solved in isolation, and each one of those problems need to work. (14:04) And you need to do that one by one, phased implementation.

(14:07) But what they wanted was, within three months, they wanted an AI tool that solves all those problems. (14:12) So, that clearly is a misunderstanding of AI capabilities, and it’s very clear to me that that project is going to fail. (14:21) And I did not take that company on as a client because their expectations were completely unrealistic.

(14:30) Yeah, so that’s a big one.

Melissa

(14:32) And I think that’s kind of true to a lot of things too, right? (14:36) I think everybody wants the perfect job, and the perfect waist size, and the perfect children, and we want it yesterday, not today. (14:44) And so, I think having some patience with implementations.

(14:49) Now, how can leaders spot scope creep early, and is there specific red flags that show up during executive meetings that they should be aware of?

Kavita

(15:03) So, when they’re breaking the problems down. (15:07) So, a lot of times, the problems are not scoped well to begin with. (15:11) So, it’s just this whole software engineering problem is labeled as an automation problem, but they don’t know whether there are some AI initiatives within that or not.

(15:23) So, it’s not well defined to begin with. (15:27) So, typically, how software engineering problems present themselves is that there might be a big problem, like maybe eliminating spam on your web platform. (15:39) But within that, there might be a UI thing to be solved, there might be some flagging related software engineering work to be done, and then a very small piece might be the AI component.

(15:52) So, they’re not breaking this problem down enough. (15:56) But if they do that, they can scope each piece appropriately, and then also add padding for delays for each piece. (16:07) And also, if you do a risk assessment on each of these individual projects, you’ll see that maybe on the AI side, you don’t have the data to start with.

(16:14) So, then maybe you need to plan some time to go collect that data. (16:18) So, by just getting really granular, you can avoid scope creep.

Melissa

(16:26) Yeah, and it’s interesting, tongue twister, right, to us. (16:30) It’s funny, I’ve seen a lot of these kind of contracts for projects, and they’re very unclear on what the deliverables are, and the timelines are probably sooner than they should be. (16:44) But really understanding what we’re trying to, what’s a win, I think often helps.

(16:49) And really communicating it with all the stakeholders, and not operating in silos. (16:55) And one of the things I’d love to get your perspective on is what recovery moves should organizations do when they realize, uh-oh, this project is not going like we expected. (17:13) Maybe we spent too much money, we’re not where we need to be.

(17:18) So, what are some of the recovery moves that they should work on once warning signs are there and they appear?

Kavita

(17:27) I would say, start with audits. (17:32) So, let’s say you have an AI project that’s not going anywhere. (17:37) It’s been a year, and I’ve seen this in companies.

(17:40) It’s been a year, and they’ve not been able to put things in production. (17:45) And you really need somebody who’s not part of the team to evaluate what’s going on. (17:52) And usually, you’ll find that it’s a mix of the output of the system, and the scope of the project itself.

(18:02) So, maybe it’s scoped too aggressively, that the AI tool can’t every possible scenario. (18:09) So, it’s not ready for deployment. (18:12) So, when they do that audit, then they can scale back and say, maybe we want to reduce the scope to begin with, so that we can get off the ground.

(18:20) So, that audit is crucial. (18:22) And sometimes it’s just the AI tool is not meeting expectations. (18:27) So, maybe the team does not know how to evaluate the solution.

(18:31) So, maybe they need somebody with a deeper data science foundations to help them. (18:37) So, that could be it too.

Melissa

(18:38) And I love that because I often find in the organizations I’ve been a part of these transformations that, to your point, they take a year. (18:50) And I think about when Copilot first came out, we let everybody have a license and found out that 20% of the people were using the Copilot license. (19:03) And so, they’re like, okay, we’re just going to cancel Copilot for everybody because nobody’s using it without actually teaching people how to use it or how it can save them time in their day or support what they’re already doing.

(19:16) So, let’s talk a little bit about where AI can save you money today in your business. (19:22) So, it’s no secret so many companies are bleeding money today in areas that they can’t control. (19:29) And so, I love to kind of get your perspective on how companies can leverage AI in their normal day-to-day that could actually help them also fix problems that they have today.

(19:43) This episode is brought to you by Summit Ventures. (19:48) If you’re an accredited investor, Summit gives you access to one of the greatest tax advantage opportunities, direct ownership in oil and gas. (20:00) Their projects deliver what they call the triple play, cash flow, equity growth, and powerful tax benefits.

(20:08) And here’s the best part. (20:09) These investments qualify for 1031 exchanges. (20:14) That means you can roll gains from real estate into energy while deferring capital gains.

(20:21) To learn more and get a free white paper, oil and gas demystified, just visit www.summitven.com forward slash executive connect.

Kavita

(20:35) Yes, I think there are two parts to this. (20:38) One is personal productivity type work where you’re doing copywriting, you’re generating images, you’re creating slides. (20:49) So, for those cases, I think people lack training, like you said.

(20:54) So, they don’t really know how to use these tools to its maximum potential. (21:00) Like, I use a lot of AI tools, but my husband doesn’t. (21:05) So, because he feels intimidated and he just uses a bit of copilot and that’s it.

(21:11) So, he needs more training. (21:13) And so, that’s something they need to focus on. (21:16) And then the second part of this is, there’s AI and then there’s AI deep within your workflows.

(21:22) So, if you want some type of gain for your business, like really significant improvement that aligns with your broader business objectives, you need to look for opportunities deep within workflows. (21:39) So, for example, in manufacturing, let’s say you want to improve your top line. (21:45) And one of the ways you can do that is to increase throughput.

(21:49) And one way to increase throughput is to look at how many things you can get out fast, faster than you currently are doing. (21:57) So, then maybe you can have AI in defect detection. (22:01) So, get your test engineers to be a lot more effective.

(22:04) So, because AI can see things that human eye may not be able to see. (22:10) And it can pick up things much faster than the human eye. (22:14) So, try to highlight those defects for the human so that they can actually vet that these are in fact defects.

(22:20) So, then things go out much faster. (22:23) And there are many areas like that where the workflow is so repetitive and there’s actually data coming out from that process, like a lot of data being generated, which can be leveraged with AI to make things faster. (22:38) So, finding those opportunities and it might be one or two opportunities, that’s all you need, but finding those are crucial.

Melissa

(22:49) Yeah, it’s so true. (22:51) And I mean, I’m similar to you. (22:53) I think I use it every day personally and professionally, anything from making meals out of what I have in my refrigerator to figuring out the fastest, quickest way to implement something or getting different insights across different tools.

(23:11) And so, I’m curious from a compliance overhead, now how can AI reduce the compliance function inside organizations?

Kavita

(23:22) When you say compliance, are you talking about like data privacy type issues?

Melissa

(23:27) Yeah, just, I mean, compliance is a big word. (23:31) There’s all different types of compliance, regulatory compliance, there’s different frameworks from different PCI or cybersecurity. (23:41) Just curious, like I know a lot of organizations write and rewrite policies and sometimes the policies continue to grow, yet a lot of the policies are not relevant to where we are in the world today.

(23:55) So, I was just thinking more like, how can people leverage or companies leverage AI to maybe help them with their compliance overhead or their policies?

Kavita

(24:08) Yeah, I think that’s a good use of AI actually. (24:10) So, I think using all the compliance documents, you can create training that’s actually accessible to your employees. (24:20) And I think that’s a very good use of AI, especially if it’s accurate.

(24:25) So, there are ways to make your AI chatbots more accurate than it currently might be, but that’s a different discussion. (24:33) So, one is like, and even like if employees have questions, if they’re doing the right thing in terms of being compliant within the company. (24:42) So, if they had like a chatbot, they could ask questions, which is very narrow chatbot that just answers compliance related questions, then that could be helpful.

(24:52) And training material for sure. (24:54) So, that’s definitely an area I’ve seen companies use AI for.

Melissa

(25:00) Yeah. (25:01) And I think back to when employees start, you have your kind of onboarding process with an employee, maybe that’s a week at a company or a day or non-existent, and then you kind of send them on the way to be their expert or whatever you hired them for. (25:18) But I think my personal opinion is without training people on how to better use a tool or things that they didn’t even think of, that people aren’t going to be creative with it and they’re not going to use it.

(25:31) If they’re told to use it for just cleaning up their email grammar or generating pictures, that’s how they’re going to use it. (25:39) But if you tell them to leverage it for helping their day to day or writing a policy, they’re like, oh my gosh, I don’t have to spend 40 hours rewriting my policies. (25:50) I can spend minutes to clean up the policies.

Kavita

(25:54) Yeah. (25:56) So, the rewriting piece still needs expert in the loop very much.

Melissa

(26:02) Yeah, totally agree. (26:05) And I find writing, I’m a writer too, that I’m like, this isn’t a real person or this isn’t a real piece of information. (26:15) So, to your point, it still needs humans reviewing and editing.

Kavita

(26:18) Yes. (26:18) Yeah. (26:19) So, I would be cautious on that.

(26:22) And also, the training has to be customized to the role. (26:25) So, Gemini can be used in so many different ways. (26:29) But like a compliance officer, somebody in the security role, how are they supposed to be using it?

(26:35) Someone in HR, they can’t really use Gemini for recruiting purposes because of all the bias that might be in it. (26:43) So, how are they supposed to be using it? (26:46) So, it has to be very role specific, I think.

(26:49) Yep.

Melissa

(26:50) Yeah, I agree. (26:51) Now, kind of what you alluded to in the beginning about organizations wanting to make more money, drive more revenue, if a CEO only cared about revenue, money making and speed, what should they focus on first? (27:07) Now, we know they care about more things than that, but what should they, if money is the thing, what should they focus on first?

Kavita

(27:14) Yeah. (27:15) So, if your bottom line is important to you, I would focus on education. (27:21) So, education at the C-suite level, so that’s crucial.

(27:26) So, not just understanding the tools, but going beyond and understanding the breadth of the AI world, like how it actually works and what are the different sub-areas within the field and what are the common pitfalls. (27:41) So, understanding things at a very high level, just a breadth of knowledge will give them perspective and it’ll help them understand the risks better too, like why shouldn’t we just upload all our financial data into ChatGP? (27:54) I’ve seen people do that.

(27:56) So, that’s number one, education at the C-suite level and senior leadership. (28:03) And then starting with the plan, like what are your goals? (28:08) Like what do you want AI to do for you?

(28:09) Now that you know, understand it, what do you want to do for you? (28:12) It can be improve your top line. (28:14) So, if you want to improve your top line, then you should look for opportunities to match that, not go after every opportunity.

(28:22) So, it really has to align with what your company needs and not what the industry in general is doing.

Melissa

(28:31) Yeah. (28:31) Now, where can AI accelerate sales performance?

Kavita

(28:37) In a lot of areas, especially like lead scoring, like even suggesting the next action that the salesperson should be taking. (28:47) So, I feel like in a lot of areas, but in the end, the sales professional has to be in the driver’s seat and their workflow has to be augmented with AI. (28:58) It doesn’t have to be, but it’s nice if it can be augmented with AI, especially in the prospecting end, like that’s very, very tedious.

(29:07) So, AI can do a lot of research and find on the prospect, fill up all the missing information about the prospects. (29:14) That’s, I think, super helpful.

Melissa

(29:17) Yeah. (29:17) And as a salesperson, I find CEOs want their salespeople to not be dabbling and working all day in administrative tasks. (29:27) They keep them busy and really focus on what they’re best at, which is sales and driving revenue.

(29:32) So, if you can offload to your point, some of the sales tasks and really focus on the clients that are your right persona, they have the right buying metrics for what you’re selling, it makes it easier to just talk to clients that want what you’re offering versus trying to sort through thousands of potential customers to find the one that matches you. (29:58) So, if you can help your sales team take 1,000 people and bring it down to 100 or 10 or whatever, it makes it quicker to get sales.

Kavita

(30:07) Yeah. (30:08) And that’s an excellent use of AI. (30:10) You’re not putting it in the driver’s seat.

(30:11) It’s in the back end. (30:12) You still have the human who’s vetting all of this. (30:16) So, yeah.

Melissa

(30:18) Now, is there, and painting a broad brush here, is there one metric that leaders and executives and really just anyone in any industry, how can they track and know if AI is actually working?

Kavita

(30:41) Yeah, I would say from a broader perspective, it all boils down to ROI. (30:48) But a lot of AI projects don’t immediately translate to ROI. (30:53) Some are so niche.

(30:55) So, you need to be looking at the in-between metrics. (30:59) And those will look different by the problem. (31:01) It could be reduction in churn.

(31:04) It could be time savings. (31:06) It could be revenue growth. (31:07) So, it’s very specific to the problem.

(31:10) But in the end, all of these in combination within an initiative should translate to ROI. (31:19) But these in-between metrics need to be tracked from the get-go. (31:23) So, the day you release an AI tool, you’re tracking these metrics.

Melissa

(31:28) Yeah, and I agree. (31:29) I’m a big, I believe in tracking and goal setting and vision and all that. (31:35) But I find sometimes, and tell me if I’m wrong here, what happens is the metrics plateau and then everyone’s kind of like, this isn’t working.

(31:47) So, what happens and what should an organization do if the metrics and the things they’re tracking plateaus?

Kavita

(31:54) Yeah. (31:54) So, in my book, I talk about these three areas of success. (31:58) So, business success, model success, and user success.

(32:01) So, any of these three could be at risk at this point. (32:06) So, you need to investigate what’s really happening. (32:10) Again, peeling back the layers.

(32:12) It could be that the model has degraded and it needs to be retrained and it needs to be re-released. (32:19) And it also could be that your users are not using the tools as they should anymore. (32:25) So, they were initially trained to use the tools, AI tools, but they’re not using it the way you should.

(32:29) So, then retraining is needed. (32:32) And it could be that you did not need the tool in the first place. (32:37) So, that’s why you’re not moving the needle.

(32:41) So, you have to look at these, definitely these three areas of success.

Melissa

(32:46) Yeah. (32:46) I agree. (32:47) I think, I love the tracking, but I also love, I find that we learn the best when we make mistakes.

(32:56) And sometimes, because of those mistakes, we develop misconceptions of tools on whether they’re valuable or not valuable to the organization. (33:07) So, what are some of the biggest executive mistakes that people are believing about AI that is just absolutely, completely wrong? (33:40) Just visit moneyripples.com forward slash secrets and enter the promo code EXEC.

Kavita

(33:48) The one that’s going on trending right now is they can replace entire roles with an agent. (33:55) So, recruiter replaced with agent. (33:58) So, but that’s not really how recruiting really works.

(34:02) There’s a lot of human element to it, as an example. (34:05) So, you’re sourcing candidates. (34:07) So, maybe part of that can use AI.

(34:09) Then you’re making decisions on which candidates to interview. (34:13) Then you have round one, round two, round three. (34:15) And each of those requires human decision-making.

(34:19) So, imagine you’re trying to replace that entire workflow with an AI bot. (34:26) That’s not how it works. (34:27) It’s not just one bot.

(34:29) It’s many sub-bots working to solve different problems in that workflow. (34:35) And you still need human in the loop. (34:37) So, the fact that they think that the human involvement is going to go away, that’s not correct.

(34:44) It might get slimmer, but it’s not going to go away completely. (34:48) So, that’s a trend that I’m seeing. (34:51) Yeah, another thing is that they think they need to use Gen AI for everything.

(34:57) But really, there are much simpler models that are leaner, easier to deploy, and those are being forgotten in the field. (35:06) And I think that’s not doing service to the current projects right now.

Melissa

(35:13) Yeah. (35:13) And I find much like the world we all live in, things change. (35:19) Even before AI, people’s roles evolved.

(35:24) Before we had a cell phone, people’s roles evolved. (35:28) And so, it’s just another way of our roles evolving, but we do need the human to tell the tools what to do. (35:36) And so, how can organizations figure out if they’re AI ready?

(35:41) What should they be asking themselves to know if they’re prepared for it?

Kavita

(35:46) Yeah. (35:46) I think the first is going back to the education piece. (35:51) So, do they understand AI beyond the tools?

(35:54) Do they understand the field? (35:56) Do they know how AI works? (35:58) So, that’s the first one.

(35:59) And this is at the executive leadership level. (36:02) I’m not even saying that your employees need to know AI. (36:05) I’m saying you as a leader, do you know AI?

(36:09) Then comes the landscape of opportunity. (36:13) So, do you understand them sufficiently? (36:16) Do you know what your true AI opportunities are?

(36:18) Or do you just know what peers are doing? (36:21) Because what you will look, what you need will look very different from what peers would need. (36:27) So, having that understanding of your AI opportunities.

(36:31) And the third thing is, have they even used these tools? (36:36) I’ve had some leaders, they come to me, but they haven’t really used AI tools. (36:41) They kind of know that it’s good for business, but they haven’t used these tools.

(36:45) They don’t have experience. (36:46) They don’t understand its quirks. (36:47) So, when you don’t know what it is, then your expectations from it is going to be very unrealistic.

(36:55) So, these are the first three things you need to be looking at to become AI ready. (37:00) Then there’s the implementation level pillars, like data, skills, infrastructure. (37:08) So, all of those come later, but these are like the early readiness things.

(37:13) And in fact, on my company website, I have like a early readiness, AI readiness assessment that they can take to get a score. (37:22) So, I call this the Aura Scorecard, which will tell them whether they’re even ready to look into this journey.

Melissa

(37:31) Now, I find my parents’ generation are often skeptical about AI. (37:39) So, for that generation that owns businesses, they’re not really sure where to begin. (37:45) They just know that everybody’s talking about AI.

(37:48) And so, they probably need to do something with AI, but they’re not sure. (37:53) So, for those business owners and leaders that don’t know where to begin, and they maybe don’t have the education with AI and outside of getting education first, what do you recommend their first step being?

Kavita

(38:06) Well, tinker with the tools. (38:08) So, have somebody set it up for you and start seeing what it is, like understand what it can do for you. (38:16) Have your team members pull out different tools and maybe just demonstrate it to you.

(38:23) Like there are a lot of voice tools, voice cloning, what it does, what does ChattyPD do, how does image generation work? (38:30) Just have your team do it. (38:31) You don’t have to do it.

(38:33) Yeah.

Melissa

(38:34) And I love it. (38:35) I think that’s so true. (38:36) I mean, I just saw last week at a conference I went to the cloning of videos from a picture and just a voicemail.

(38:48) And it’s amazing how good some of this is. (38:52) I’m like, I was a little, I was pretty shocked myself. (38:56) And so, to your point, it’s here, it’s coming, get educated and use it where you can.

(39:02) And so, with all these projects, I find that I’ve heard and seen many seven-figure digital transformation AI projects happening. (39:15) The spend is real. (39:17) And to your point, the 90 plus percent fail.

(39:21) So, where can organizations get smart about how they can start exploring AI opportunities before making these six and seven-figure investments for them to plateau or be pulled?

Kavita

(39:38) Yeah. (39:38) Before jumping in, they definitely need to know the opportunity landscape and the benefits and the potential ROI that comes with each of those projects. (39:50) So, once they see that, then it’s easier to actually plan.

(39:54) And then you also need to have your own business. (39:57) Like what are your objectives for your business? (40:00) Are you looking for growth?

(40:01) Are you looking at streamline operations? (40:03) So, it needs to align. (40:05) So, what people are trying to do today is like they’re trying to do what everybody else is doing, but it needs to align with what you want your business to do.

(40:14) Where do you want your business to go? (40:17) So, I would say have that opportunity landscape understanding, understand your readiness for AI, and then plan before you jump in. (40:28) So, it could be just one or two pilots that’s going to move the needle in a big way for your business.

(40:34) So, go after those.

Melissa

(40:36) Yeah, I agree. (40:36) I love it. (40:37) Thank you so much for being here today.

(40:39) I want to get any final thoughts that we haven’t touched on that you want to leave with the listeners or something that you want them to remember from this conversation.

Kavita

(40:50) Maybe I would say something a bit odd. (40:54) Like I would say, stop following the crowd because the crowd is following the hype and the hype is not going to get your business where you want it to go. (41:04) So true.

(41:06) So, do what’s right for your business and your objectives.

Melissa

(41:11) I love it. (41:12) I think that’s such great, great advice. (41:14) If someone jumps off a clip, are you going to follow them?

(41:16) Probably not. (41:17) I hope not. (41:18) But no, thank you so much for being here.

(41:20) Please share a little bit about the good work that you’re doing and where our listeners can learn more about that work.

Kavita

(41:27) So, for my consulting website, if you want to learn more about what the type of work that we do, you can visit company website at opinosis-analytics.com. (41:37) And if you want to learn more about me, you can visit my personal website, kavita-kanesan.com. (41:43) And I’m also on LinkedIn.

Melissa

(41:45) Thank you so much for being here today, Kavita. (41:48) That’s the Executive Connect podcast.