A (3:49)

I'm curious. You had mentioned that AI really is cropping up everywhere on the web. Every interaction that we have with some kind of digital platform. Do you worry and are you concerned about just kind of the noise that AI may produce or is producing already? And is the human voice lost because AI is so prevalent?

B (4:10)

I don't think that human voice is lost, but AI voice, yes, it became noise for some places. But when you use it for to create efficiency, I don't think that it's a noise because I use it in my business every single day. Every single day I do a brainstorming session with it. Every single day. I produce documents with it. Yeah. I'm not using what it produces right away. I of course read everything. But it does give me at least 80, 90% of the work and it's done. It's like you have an assistant. Don't treat it as your colleague, but treat it as your assistant. That is the most important differentiator in today's world. Because when you treated it as an assistant, what do I do? I give a job. My assistant does the job. But it's my responsibility to make sure that it's accurate. Let me ask you that question. You're a business official at your district and if you have an assistant and you need to fill a state report, those reports that we, we love, like every single state, we love those.

A (5:25)

Yeah, yeah.

B (5:26)

And you ask your assistant to fill out the state report. You take that report, you submit it to the state, you have a problem with that report. Who's responsible, your assistant or you? Right.

A (5:45)

It's like a trust but verify kind of model. Right. So. So you're delegating a task, but the final say really should still be on the school business official because not only to your well made point, that the buck stops with you anyway. So when the state comes back and says, hey, all of this data is wrong and is trending in the wrong direction compared to last year, what's going on? Not only do you need to answer that, you should still have an understanding of why that data is submitted that way. So yes, I think I would agree with you. I think that the delegating ED as assistant is the move and trusting that what you get back probably is pretty accurate, but you still need to put your expertise to pen and paper.

B (6:28)

Exactly. If you don't read every single word, that's on you. Right. Because it can do mistakes. It's a toddler right now, it's learning. It doesn't matter how smart the toddler is. The toddler needs world experience to be able to judge. And do we trust 16 year old kid, teenager who is extremely smart. They can be the smartest person in the world, but they may not have an experience of a 60 year old person who have lived in this world for 60 years and who has seen different problems and different issues. They faced it, they faced those challenges and they overcame those challenges. They don't have that right.

A (7:21)

Right. And I wonder too, I mean, I agree with you with treating AI as the assistant and not the colleague. But what is the risk that school business officials run if they do treat it like a colleague? Beyond erroneous state reporting data, what risk is there right now if you're essentially to your point trusting a toddler with most of your work?

B (7:43)

We, as business officials, I think we have the expertise to understand and evaluate data a lot better than AI. Not because we know more than AI, but because we have more context than AI. For example, if you're doing salary analysis with AI and it analyzes and it flags a few salary items, right, A few people, but then it doesn't have a knowledge of historical what happened in the district, why that person is making so much, what was the reason, what happened? That context comes from you. So despite the fact that it will flag and it, it does great job. I'm not saying AI doesn't do a great job. No, it does a great job. But it is still at that 80, 90%. Because when you ask your question, you don't ask everything. You miss some of the context that you have in your brain, but you don't relay that message to AI. And that's why I usually say the framework that we develop and there are many, many frameworks out there and I want to reiterate that framework, the TADA framework that we came up with. It is extremely important to assign a title when you assign your task. Give as much context as you possibly can, define the format when you know what type of thing you are expecting it to give you. Most importantly Let it ask you questions. Because when you let it ask you questions, those are the things that it asks you, those are the things that you missed, but it needs so that it gives you a better response. Now, the reason I'm saying as an assistant, not as a colleague, when I brainstorm, I treat it as a colleague because I'm brainstorming. I'm saying this is what I'm trying to do and this is the new project we're working on.

A (9:49)

But the preface to that is, you've already assigned it a task and a Persona, so you can, with a greater degree of confidence, treat it as a colleague because you've already defined the parameters. You're not asking open ended questions without context.

B (10:04)

Exactly. Okay, let me rephrase. Maybe not as an assistant, but as a junior in your role, right? Not as a senior school business official, but a junior school business official, not as a senior cfo, but a junior cfo, because it does not have the same context that you have. And you cannot possibly relay all of the context that is in your brain to AI in one shot.

A (10:34)

Now, I want to go back to the idea of AI Fluency. And we reopened this segment about it really is a career differentiator. So can you tell me a little bit more about why and how AI Fluency is starting to separate maybe the standout professionals from what could be argued as the average ones? Why is AI becoming really a hot button issue in terms of separating yourself from the crowd?

B (11:01)

Now, we always complain in education sector, especially the administration offices, that we're always short staffed. Correct? We always complain that we cannot complete what we're trying to finish. Sorry, we cannot complete the jobs that are assigned to us in timely fashion. Right now, since we are short staffed, we need to compensate somehow. There are two ways. Either you stay and you work 60 hours a week. You work 80 hours a week, right? And you finish that task, or you automate some of your tasks and finish on time. That's how I look at it. Now, where does AI come in? Let's do a scenario. You, as a school business official, you are looking for an HR director for your district. There are two candidates. One professional HR director, very experienced, comes to to an interview and then you say, do you use AI? They say, I, I dabble in it. I write some emails. Sometimes I create letters for my employees when I communicate with them. And that's pretty much it. That's how much I use AI very similar credentials. Another person comes in to your interview, then you ask the same question, do you use AI and then that person answers saying that of course I do what I do. I use NotebookLM. How do I use NotebookLM? I take all of our procedure manuals, I take all of our policies, all of our guidebooks, and I create video tutorials for my district so that when I onboard someone, they can watch every single document and not read it. Who do you hire? Same experience, same credentials, same qualifications. One can complete tasks in minutes and sometimes in hours, and the other will not have time to do those things. Who would you hire? Right?

A (13:21)

I wonder too, is the, the process of finding talent going to change, especially when it relates to AI fluency? Because I would assume you need to know AI pretty well yourself to understand if one of your prospective hires also knows AI. So how do you feel that that's going to change, being prepared for the, the interview process? Just when it comes to AI fluency,

B (13:47)

I think just asking AI about AI is a huge, is very important question. Right? We always interact with AI and I actually posted an article on LinkedIn saying that. How do you start? What is the first step? Very simple. You just do this prompt as whatever your role is as an HR director, as a school business official, as a superintendent, at whatever district you work for, whatever organization you work for, for example, at any town, school district, tell me how I can use you. What tools do you have and how can I implement your services and your tools on my day to day work? It does an amazing job. I did it with ChatGPT, for example. I asked the same question to ChatGPT, you know, first what it did. It said, let me look at my capabilities and what am I able to do first? So it said, I have a canvas mode, I have research mode, I have this mode, I have that mode. So it went through the whole list, then it took that list and compared it to my position and it said, you are a cfo. You're probably doing some expense analysis, you're probably doing, creating presentations for your board. You're doing this, you're doing that. So, oh, one thing I forgot in that prompt, give me 10 use cases that I can start applying every single day. And those use cases should not exceed 15 minutes. That's a huge starting point because once you apply those 10 use cases on a daily basis, there's this thing called compound effect. It compounds. You can do 15 minutes a day, but after 10 days you would actually realize that you saved a lot of hours and you learned many, many different use cases along the way.

A (15:59)

You know, we, we did a training together months ago and you had said something in that training that really stuck with me because some of, some of the pushback that we got from our, our all day session that we ran together was, well, do I really need to pay $20 a month for, for ChatGPT to take advantage of allegedly these greater skills? And you said something, and I'm paraphrasing, said something along the lines of, well, if you can pay $20 a month and save an hour of your time, you've already paid for it how many times over when you, you know, put your salary into an hourly basis. And I think that really resonated with a lot of people because, you know, we have subscriptions for everything. And it's just one more thing. But if you're able to save that amount of time for $20, just think about what you could do and what, you know, what a strong return on investment that, that $20 is or, you know, what, whatever the price tag is on whatever model you're using.

B (16:52)

Exactly. Because we look at the price at a relevant scale. If I asked, do we buy coffee from Starbucks once a month and do we actually sometimes get hungry and buy a sandwich? A sandwich and a coffee will cost you $20. Close to $20. Yeah. It's not $20, maybe 50 bucks.

A (17:16)

I want to push back. It's going to cost you more than that because I don't know what I did in the Starbucks app this morning, But I paid $7 for a drink. Aziz, I got bamboozled. I don't know, I don't know what I did. I have to go back and look at the details. But, you know, I'd much rather spend that money on some time saving measures than some mediocre coffee or whatever I had.

B (17:38)

Exactly. How long does it take for you to brew coffee at home? Probably about five minutes. And if you brew coffee at home for five minutes and save that $7, multiply that with 20 days, that's $140. If you paid only $20 for any platform, it doesn't matter what platform you're using for AI, it can save you hours and hours if you're using it. Right, right. And that's exactly, that's my approach. And now I actually increased it. Sometimes we need a better model and bigger subscription. So there are $100 subscription, there's, there's another 1, 200 subscription. Evaluate how much you're using it and how much time it's saving you. And that's what measures it. Because sometimes we spend $60 at a restaurant and we didn't we don't even think about it. What if we ate at home that day and saved that $60 and paid it for efficiency? I'm not advertising any AI company. What I'm trying to say is I have subscriptions on all of them and I pay subscription money every month for those. All of that together combined, total subscription. I pay, I make it back in one day. I'll give you an example. We did a proposal the other day and I used cloud coworkers for that to put the proposal project for Cowork. It took me about half an hour because I needed to figure out what skills it has, what skills it needs to write the proposal. I needed to give my branding to it, I needed to explain what my tone is, and I uploaded my documents to that so that it understands everything. Then I had it write a skill and I had it write the instructions for me. Once I created that project with Cowork, I said, here's my client and here's what I'm trying to propose to create the proposal for me, but tailor it to my client. What it did, it went online, it researched my client and it researched my previous proposals, context, it knows about what I do anyways. And it created a 17 page proposal beautifully designed in Microsoft Word that I can go change. I changed probably about 3 to 5% of the context. That's it.

A (20:12)

Now 17 pages, how long would that have happened? How long would that have taken you with opening a blank Word document?

B (20:19)

Right, Exactly. Probably I would take about two weeks to gather that information and put the proposal together. Because I don't know what to put on a proposal if I'm a new business, if I'm starting a business. So sometimes as a school business, let's look at from that perspective. You have expense data, you have been doing some analysis for the past experience, but then you give it to Claude and say, be creative, come up with different analysis that I've never done. It'll come up with new things that you've never looked at. Why not? And it creates the files for you, it creates beautiful Designs, it creates PowerPoint presentations. Why not, if we use one hour? Because we create PowerPoint presentations on a monthly basis. Some of us have templates, we just change the numbers. Even changing the number will take you 20 minutes. Right. And I can do it in minutes. And you don't have to be there. That's the.

A (21:36)

And I think that's part of the beauty too is a lot of these models you can start, you can give it a task and it'll run in the background. It's not like you have to sit there and babysit it. If it's a complex, even like a deep research, if it's a complex task that may take anywhere from 10 to 30 minutes, maybe more, you just let it run and then you come back and you're able to kind of analyze what it's given you. And I think to your earlier point, the more context that you give it, and these models are getting smarter and can go find their own context on the web, but the more context and direction you give it and the more explicit you are in its instruction. I think you said what you changed 3 to 5% of that proposal. I mean, that's incredible. It would have taken you.

B (22:16)

Yeah, exactly. It found what the company is. It found all the branches that company has throughout the world. It found all the information. It found what they're trying to do with AI, because I'm going for an AI training. And it kind of analyzed all of that and put it in the context of the proposal, saying that here are your goals and here are your missions and visions. And this is how we align with your mission and vision. For us to do that, we need to do a huge analysis. So who saved time? I did. And that time that I saved, actually one of the proposals, it was late at night. It was like 11 o' clock at night. And I said, oh, I'm so tired. I will give you the task. Ask me any questions right now because I want to sleep. And then it said, here are the questions. I want to clarify two, three questions. I answered those questions and it said, good night. In the morning I came back, the proposal was ready. Word file in my folder. That's great. I opened it, I analyzed it, and I sent it. So my point is, you don't have to be on top of it. You don't have to stare at it while it's working, but you can give multiple tasks at the same time and it will be completed. Then you come back to it and evaluate it.

A (23:44)

So I want to talk a little bit about kind of practical use cases before we dive into. We wanted to talk, Claude, a little bit, but before we do that, can you just talk a little bit more in depth on how school business officials can use AI right now to become better at their everyday jobs? I mean, you've, you've used some examples. I like the example of interviewing the HR person and just identical candidate. It's just one is a little bit more AI fluent. But when we kind of narrow the focus into the business office and maybe just operations in general. How can school business officials more effectively use AI?

B (24:18)

Most business officials do bargaining negotiations like union negotiations, right? We have bargaining contracts. We have. Let's take a teacher contract. You have a teacher contract and now it's time to do a new teacher contract. You have a proposed language by the union and you have your current. You take both of them and you ask AI to evaluate and compare two of these contracts. You don't have to read it at this point and point out exactly what changed, what are they proposing, what are the risks for school district. And by the way, create an Excel spreadsheet comparing those. Create an executive summary in word and create a PowerPoint presentation I can share with my board. That's one example.

A (25:13)

So you're using one interface to create three to four different documents for three to four different audiences that otherwise would take you hours and hours and hours to replicate and change the messaging and the tone. It was pretty impressive.

B (25:26)

Exactly. And it takes 10 minutes. I did this with a huge school district, with a very big school district, one of the largest in the country. And they just signed teacher contract. We took their new teacher contract, we took the old teacher contract, we put it in a folder. Then I assigned Claude with cowork. I said, here's the folder you need to look at. What I need, I need you to compare these two contracts, create an executive summary so that I can read it. Create Excel spreadsheet because I'm analytical person and I want to compare in tables. And also I want you to create a PowerPoint presentation because I want to share this with my team so they're aware of the new contract and what changes there are going forward. They need to pay attention to those when they pay people, when they evaluate salaries and all that 400 page document. And I want to mention that it's two different documents that both is 400 pages. It took it 10 minutes to create the Excel spreadsheet word file and the PowerPoint presentation. That's great. Now how much time would it take you just to read those 400 pages, let alone that comparing it to the previous one and saying that, oh, we had this in that contract and now it's this. It's. It's impossible to do it in a day. It's impossible to do it with five team members?

A (27:11)

Well, it's almost incomprehensible with everything else that has to get done to be able to dedicate that kind of time to just looking at a 400 page document, hopefully remembering everything correctly and then analyzing it and Putting into a format that someone else is going to consume. I mean, it's just that in itself is almost a full time job.

B (27:30)

Exactly. And we don't have that time. Right. So we need to stay after work after everyone leaves and no one asks us question so that we can sit down, focus and read and we steal it from our families. That's how I look at it. We are not trying to automate it so that you lose your job, but we're trying to automate it so that you go home on time.

A (27:55)

Right.

B (27:55)

If you have time left for like for business, office staff. I'm talking about if you have time left, you have five hours a week. You saved five hours a week. That time needs to be allocated for education. You educate yourself. That's one, two. You educate your staff. How do you do reimbursement? Because people don't and they call you instead. Take your workflow, draw it on a piece of paper, take a photo of it, open Gemini, open ChatGPT, turn on canvas mode. That is very important. Turn on Canvas mode, upload your photo and then open your microphone dictation and start explaining what that workflow is. Then you say, we will start with an outline. I want you to write an outline. Then we will work every section at a time, like one section at a time. Then you start that. It gives you a document, editable document, right inside the platform. So if you're in Gemini, it opens on a side. If you're in ChatGPT, it opens to the side. Then you go section by section, you highlight a portion you want to work on. You highlight it, you open your dictation again and you explain what you want to include in that section. And in half an hour you have a new procedure manual. Now let me continue that workflow. You take that ready procedure manual that you just created, put it in notebook. LM create a video tutorial from that procedure manual. By the way, if you're a district who has a few employees or many employees whose English is their second language, create that video in many different languages, up to 100 different languages, and put those videos in a portal that your employees can watch. You've just automated something that would, would have taken you days in. You, you finished it in what, 45 minutes to an hour.

A (30:13)

And especially with translation services, you're saving money and time as well because you're likely paying a professional translator to, to do all that for you.

B (30:21)

Exactly. And now I usually joke about it, but it's not a joke. People please do it. Buy gift cards, $5 gift cards, put it, set it on your desk, then take a gift card, take a video, go to that person who speaks that language, give the gift card and say, can you please watch this 10 minute video and tell me if it makes sense? Right. It's a lot cheaper than $500 that you will pay, you know. Right.

A (30:50)

And to your earlier point, you still need some kind of human intervention to ensure that it's accurate.

B (30:55)

Right? Exactly. I would definitely do that. It's mostly accurate. And I know from the language I speak, I speak Azerbaijani language, for example. And that's one of the languages that there was no dictation, there was no pronunciation by AI before. Not before AI, the computers wouldn't pronounce the language. But now I can create videos in that language. It's very academic, it's very clear, no accent whatsoever. And it's clear to the point. It's an amazing tool. Yeah, that's really impressive.

A (31:35)

And speaking of some of these tools, we did want to talk a little bit specifically about Claude. Historically we've talked about ChatGPT, Gemini and some copilot, but there's recently been some new iterations and integrations with Claude specifically that could benefit school business officials and how it integrates with Excel, PowerPoint and Word. So can you tell me a little bit about these tools and these integrations with Claude and how SBOs can maybe leverage those?

B (32:06)

I think in the past few years CLAUDE was very famous with its coding. Now it's one of the best, I would say, in coding. And people say that platform died. This platform died. The king is Claude. Because of Claude code. Now looking at cloud code, because it knows how to code, it's very proficient in coding. What cloud did, they said, okay, now it's good with coding, but office people don't know how to code. We as CFOs, as superintendents, as HR directors, we don't know how to code. What can we do to make their life easier with our code skills without them seeing the code, what they did, they introduced these skills, cloud skills and it's skills by Anthropic. One of them is Excel skills, the one Microsoft Word skill and another one PowerPoint skill. And they also have PDF skills. Now with the Microsoft skill, you don't even have to ask Claude if you upload a spreadsheet, Microsoft Excel spreadsheet, it knows that, okay, I need to use my Excel skill to read this Excel, right? And then you say create an Excel for me. Then it says, okay, now I need to use my Excel skill to create An Excel what it does, it actually writes code to create the Excel, but it does it perfectly with color coding, highlighting, structuring different tabs, different charts. I can talk all day about its Excel skills because I like Excel. I started with Excel when I was a CFO before and all of us in school environment, in a business office, we use Excel now going to other organizations, other industries. A lot of analysis is done in Excel because the software doesn't give everything we want. So we download the raw data and we try to analyze it ourselves with Claude. Now you can just feed that into. Actually you don't even need to upload it to Claude anymore because there is Claude cowork available, right? You just put it in a folder, show that folder to Claude cowork and say this is the folder that you need to analyze all of the files in there. Put that Excel in there and then it will create another Excel file from that Excel file. Or if you want to just keep the same Excel and add tabs for you, it will do that. I do give salary projection service, as you know, but I'll shoot myself on the foot. Here's an idea. Salary projection tool. I've been teaching this for many years, I think since 2019. I, I started with salary projection and I still teach that. So think about it. Whatever you envision to have in salary projection in Excel, explain it to Claude and it'll do that for you. And then in the prompt say do not use data, use all formulas. Because I want this to be a dynamic sheet so that when I change the data, everything still updates. I don't have to recreate. Will finish it in a few hours. In a few hours you will have a software quality salary projection that, that you've never thought it was possible.

A (35:53)

Now what if you have something like a lot of school business officials, I'm sure already have their own homegrown Excel sheets and scattergram tools and everything. Do you find it's best to reload Claude or an equivalent with that as an example? Or is it easier just to kind of start from scratch and explain really what your end goal is and what you're trying to accomplish?

B (36:13)

It has advantages and disadvantages for some of the Excel spreadsheets that we have. It's good to human eye, but it's really tough on computers because we have these scattered tables everywhere. You know, we have one table up top, we have one table on the right, we have one table on the side. It can get confused. It may also not get confused depending on what kind of spreadsheet it is. I went, I went both ways. One example I can give you, I started from scratch. I said, this is the data and here's what I'm trying to build. And what I also did, I gave my Google sheet with multiple tabs that I created for a district for Grant Automation. And I said, here's your Grant automation template. What I want you to do, go through each tab, go through each row, analyze every single formula that I wrote there, understand the connections between the formulas and the data, what that formula does, where it reads it from, what kind of result I'm expecting, and then tell me what you understand. So it took that automated sheet, it analyzed everything and it said, okay, I see the summary sheet that does this, I see the grant data that does this. Then we went over its plan of creating a software out of that and I, I approved the plan and I made a few changes saying that no, you didn't understand it correctly. Go read it one more time. And it worked out great.

A (37:54)

But it still can go both ways, right?

B (37:57)

Right. What is it?

A (37:59)

But it still required your intervention and knowledge and expertise. Right. You couldn't just give it three prompts and it gave you this complete software package. It was kind of an iterative process where you had to go back and forth and of refine it. Correct?

B (38:12)

Exactly. Be ready to do that, be ready to go and spend some time on it. This was my approach when I was a school business official. I had this report that I need to fill out that would have taken me one hour, but I would do it manually. I would open that one report on the one screen and the report that I need to do on the second screen and I would copy the numbers from the left to the right instead. I usually chose to do seven hour automation. So I spent on that same task for seven hours. But I said next year I don't have to worry about it. And it was a fun challenge because when you solve problems it actually gives you satisfaction rather than doing a manual task. That's how I approach this. If you have an idea, do that iterative process. Spend some time, but you will love the result.

A (39:09)

So we've covered a lot today and we've gotten into the weeds a little bit with how someone could potentially utilize Claude and its new skills features. But if someone listening feels left behind, where would you suggest they get started this week?

B (39:24)

This is a perfect thing that you brought up because I just released a video on that and it's about a shoe story I would like you to tell. I would like to Tell you that shoe story. I have a sister and my sister wanted to start exercising. And then she said, for me to exercise, you know, you exercise, you know how it feels for me to exercise. I need a treadmill. They bought the treadmill, the treadmill went to the first floor. They live on a second floor. She said, the treadmill needs to be on a second floor so I can see it and I can jump on it. They brought the treadmill to the second floor and then she said, I have a treadmill, but I don't have elliptical. I need to get an elliptical. Because without an elliptical, treadmill is nothing. It will not help me. They got an elliptical. She still didn't exercise. Then she said, I would love to jump on that treadmill and practice and exercise. I don't have shoes. It took her one year to find that perfect shoe and she purchased it just before she would jump on a treadmill. She said, I don't have a step. I need a step to do the strength training. If I did a treadmill, that's cardio. I need strength training. And she's still looking for a step. Apparently this is a new information. I was telling the story and she said that I actually got a step and now I need a bicycle. So my point is this. If you don't start, you will never start. What you do, you start, you take it easy, one thing at a time. You will not get proficient. About three months ago, I wasn't using Claude at all and I started using Cloud one by one. And now I use Cloud every day. So it's not about when to start, it's about where to start. And you start small by asking small questions, then very small tasks. Remember my previous prompt example? Act not act as a school business official at any town, school district. How can I use you? List all the tools you have available to me in my position and day to day tasks. What tools do I use and where do I start? List at least 10 different use cases I can start using today. And I will be applying these use cases every single day. The cases should not exceed 15 minutes. That by itself will help you to start using AI. That's great.

A (42:35)

Does your sister know you tell that story?

B (42:38)

She does and she's very mad at me. But no one knows her.

A (42:44)

In your defense, you haven't given any names or anything.

B (42:47)

Exactly. I have three sisters by the way, so who knows which one.

A (42:50)

Obscure it even further. Well, Aziz, thank you so much again for joining me. It's always a pleasure to have you on and you always impart such an immense amount of knowledge onto not just myself, but our listeners. And you know, if those listening, if you haven't started you heard here first, at least get started with something and AI can help you kind of prioritize that. So Aziz, thank you again. I'm sure you'll be on in the future, and it's been a pleasure.

B (43:15)

Thank you for having me. It's always fun to speak with you.

A (43:19)

Thank you for tuning in to School Business Insider. Make sure to check back each week for your favorite topics on school, cool business.