AI For The Rest Of Us & Your AI Acceptable Use Policy

Tony Martignetti
Hello and welcome to Tony Martignetti Nonprofit Radio, big nonprofit ideas for the other 95%. I'm your aptly named host and the podfather of your favorite hebdomadal podcast. Oh, I'm glad you're with us. I'd suffer the effects of gingival hyperplasia if I had to chew on the idea that you missed this week's show. Here's our associate producer Kate to tell us what's going on.

Kate Martignetti
Hey Tony, we've got AI for the Rest of Us. Our coverage of the 2026 Nonprofit Technology Conference continues with Allison McMillan's survey of these AI tools. Claude, Gemini, Perplexity, and Suno. She reviews their use cases, the differences between them, their limitations, and pitfalls. Allison is CEO of Tavlin Consulting. Then, your AI acceptable use policy. Eric Molho, founder of Bon Partners, explains what belongs in your AI acceptable use policy. As you and your AI tools learn iteratively from each other, evolving your policy and culture, you need guardrails around data protection, transparency, accuracy, ethics, and sustainability. On Tony's Take 2.

Tony Martignetti
My updated book title.

Kate Martignetti
Here is AI for the Rest of Us.

Tony Martignetti
Welcome back to Tony Martignetti Nonprofit Radio coverage of the 2026 Nonprofit Technology Conference. My guest for this session is Allison McMillan. Allison is CEO at Tavlin Consulting. Allison McMillan, welcome to Nonprofit Radio.

Allison McMillan
Hello, thanks for having me.

Tony Martignetti
It's a pleasure. Your subject is AI for the Rest of Us: A Nonprofit Professional's Guide to Getting Started. Give us a 30,000-foot overview of the topic. Please.

Allison McMillan
Yeah, I think right now, I mean, obviously AI is a super hot topic. There are lots and lots of sessions about AI, how to use it, how to adopt it. But I feel like until you get really hands-on with the tools and understand different tools, their use cases, et cetera, it's really hard to get to that aha moment or to get to that moment where you're like, oh yes, I see where and how this can be useful. And so AI for the Rest of Us is really an introduction into what LLMs are, what the different tools are. We do sort of a survey of a handful of different tools, and then it's really getting hands-on with them so that you have that practice. You can compare and contrast tools, and you can really get deeper into which tools you might want to actually adopt, become a power user of, or use more once they go back home.

Tony Martignetti
All right, cool. Thank you. And we've got plenty of time to dive into that. So why don't we begin with our survey of tools? How should we approach this? Do you want to just one by one? Should I throw them out? ChatGPT, Anthropic, Claude, what, you know them?

Allison McMillan
Yeah, up to, I mean, I can go through the ones that I went over in the session.

Tony Martignetti
Please do, yeah.

Allison McMillan
So we started with Claude. We just spent a couple of minutes on Claude. Well, I guess we started with what I feel like are the best ways to even think about how to get started with using AI tools. 'Cause oftentimes people, they see all these really interesting, super complex use cases through, you know, social media or different people talking to them, and they think that they should start there, and that feels like a really hard place to get to quickly. And so what I recommend is that folks start with one of 3 things. Either what are they still Googling? So instead of going to a search engine, go to an AI tool. Where do they get analysis paralysis, right? Where is it like, I have so many things to compare, I don't have the time to go deep into each of these. Yeah. And I feel like I need to make a decision, but I can't. And I've been sitting in this need to make a decision space for too long. Or third, the third is— I knew you would. What are the places where you don't feel like it's a good use of your time, right? Where you're sitting there going, why am I spending the next hour doing this? There are so many other ways that I could and should be using my time right now. So those are really the 3 places that people can— that I recommend that folks get started using AI and experimenting with it.

Tony Martignetti
Okay, entry points.

Allison McMillan
Yes, exactly.

Tony Martignetti
Easy, low-lift entry points.

Allison McMillan
Exactly, exactly. So yeah, so then we started with Claude. Okay, let's start with Claude. Yeah, so Claude is the one that I feel like most folks in this audience have probably played around with at least a little bit. But

Tony Martignetti
Claude is the Anthropic tool.

Allison McMillan
Claude is the Anthropic tool, yes. And Claude is really good for sort of general questions, for, you know, a chat tool. One of the things that we talked about that I like to recommend is projects, which are specific spaces that you can set up that have where you can give it a knowledge base. So you can give it a whole lot of background information that it can then pull into any answers. that you're looking for. So we talk a little bit about projects and how to set them up. They're similar to Gemini Gems or custom GPTs in ChatGPT. And so that's Claude. And then folks got a few minutes to grab an idea, work together in small groups or in pairs, and really experiment and get hands-on with the tool.

Tony Martignetti
Now, what are some tips about setting up Claude that you shared? We don't want to be shortchanging Nonprofit Radio listeners here, since we didn't have the value of sitting in the session, what are some of your setup tips?

Allison McMillan
Yeah, my setup tips are— Claude is actually really easy to get— actually, all of these tools are really easy to get started with. And so we don't actually spend a ton of time on setup because there are so many different tutorials and things that talk about setup that you could spend such a long time in that configuration space without actually getting your hands onto the keyboard and typing in prompts and being like, Okay, if this happens, what answer do I get back? What does it look like? So again, I talk a lot about projects as like a really good space to— and you can have many projects.

Tony Martignetti
Like a project like what?

Allison McMillan
So there is on Claude in the top left corner.

Tony Martignetti
Oh, this is a Claude option?

Allison McMillan
Yes, it's a Claude option. Choose projects? Yes, to choose projects. And then so when you choose a project, you can create a project. So let's say I have a project on for social media, right? I want it to write LinkedIn posts. What I could do is I could then give a knowledge base to that project of LinkedIn posts that I like, ones that I think are well formatted, maybe ideas of LinkedIn post formulas, blog posts, a couple of blog posts about LinkedIn posts that went viral, right? Things like that. And then whenever I want to write a LinkedIn post, I go into that specific project and I have the chat within that project because it has all of that context and knowledge already built in. Yeah. And so one of the things that people talk about a lot is, you know, I ask AI to write an email for me or to write a LinkedIn post and it doesn't sound like me at all, or it just, it doesn't, it doesn't land, it doesn't feel right. And so projects are really a great way to train so that you spend significantly less time editing. Because

Tony Martignetti
you're working on specific use cases. Right. Each project is LinkedIn or a blog post.

Allison McMillan
Exactly. Exactly. And one of the things that I suggest is early on as you're training that project, as you're doing that, let's say that you're writing a LinkedIn post or writing a blog post or even writing a newsletter, right? The top section of a newsletter. You can You can co-create. So Claude can give you something and you can say, I like the second sentence. I don't like the opening. The whole second paragraph has to change. Give me something that's a little bit more this flavor or that flavor. It'll give you something new. You can go back and forth with Claude in that way. And then what I usually do is at the end, I say, great. What did you learn about me, my organization, my style, and my tone through this co-authoring of this post, turn that into an artifact or into a document, and then you can add that document also into your project knowledge base. So it will pull all that information in the future.

Tony Martignetti
Oh, so asking it to analyze itself. What did you, what did you learn about me? Right. And then use that iteratively.

Allison McMillan
Yes. Because then it says like, oh, this is the kind of, you like, you use fun tone, or you're very like business, or this is for donor communication. It always has this this kind of angle, or you always end with the CTA. Like, it will come up with all of the specifics about you based on that back and forth when you said, don't like this, like this, don't like this, like this.

Tony Martignetti
All right, all right, cool. All right, Claude, which model should we talk about next?

Allison McMillan
Yeah, so then we talked about Gemini.

Tony Martignetti
This is the Google one?

Allison McMillan
This is the Google one. Yes. And so Gemini is interesting because I think that Gemini lagged behind for quite some time. It was not a go-to option in terms of these AI tools. You heard a lot about Claude and ChatGPT and not so much about Gemini for a long time. And then Gemini really sort of like sped ahead and leapfrogged, I think, ChatGPT, not quite Claude, but it actually became a very, very good The other thing is that with Gemini, so Gemini itself, you use similarly to how you would use Claude. There's that chat functionality. They have gems, which are similar to projects. They also have scheduled actions, which I think are the easiest sort of agent-like setup to set up, where basically you give it, you say, you know, I want to know this thing. Right. And please run it once a week. I have an out-of-the-box one running right now, which is every day it sends me something interesting that happened in history on this day. Oh. You know, so, and that's an out-of-the-box one that actually Gemini offers. So it comes with scheduled actions, Gems, and then Gemini, which you use like a chat functionality. But the other interesting piece is that it comes with, when you pay for Gemini, you also automatically get NotebookLM. and Nano Banana. So NotebookLM is really good.

Tony Martignetti
Nano Banana?

Allison McMillan
Nano Banana.

Tony Martignetti
All right, we'll get— yeah, I know you're going to explain what that means.

Allison McMillan
Okay. NotebookLM is really good for lots of larger research reports, or basically NotebookLM does a variety of things. One of the most common use cases for it are when people need to summarize notes or big reports, they'll put it into NotebookLM and they have an audio overview button and it will turn it into a podcast for you. So if you enjoy podcasts and listening to conversations and it's a little bit more engaging for you, it will turn into a podcast that you can then listen to.

Tony Martignetti
Not going to be equivalent to Tony Martignetti's Nonprofit Radio podcast. Of course, it goes without saying, but I say it anyway.

Allison McMillan
Yes, yes.

Tony Martignetti
It won't be that quality, so you'll suffer a lesser engaged— well, you might have a better host, but the guest experience is not going to be the same.

Allison McMillan
Yes, but if you think about it—

Tony Martignetti
It's going to be much lower quality.

Allison McMillan
But if you have a 30-page report and you're like, okay, I just need a summary of this instead of reading it through, it makes it more interesting, more engaging. If you like to learn that way. Exactly. Okay. So it does that. It also— I know actually a lot of students that use it. There's a flashcard option, so you can turn, let's say, like a study guide or material into flashcards. That's smart. So I know a number of students that use it in that way as well. So yeah, so that is what NotebookLM is. And then Nano Banana is Google's image generation tool. And it's sort of widely regarded as the best image generation tool out there right now. So, and you can enable it from the Gemini chat. If you say create an image, it will automatically use Nano Banana as what it's using to create that image. Sure. Or you can select— there's like a dropdown and you can specifically select the tool Nano Banana to use for image generation. And Yeah, so, and then people got hands-on and they— there were a lot of, there were a lot of images created, which are, which are fun. But Nano Banana is very good at when you describe an image, or sometimes what I recommended also is you can put in a newsletter or a communication that you want to go out, and you can say, create for me an image that would go along with this email, that would go along with this newsletter, etc. And it will You know, read the information, figure out, and at least give you a first draft. You can also tell it styles like in watercolor style, in a sketch style, in realistic animation, in photograph, whatever. And so, yeah, so those are, those are some of the uses for Gemini.

Tony Martignetti
Okay. Are we going to chat next?

Allison McMillan
No, I actually, I did not cover ChatGPT because I don't I don't find that it's as good as a lot of the other tools at the moment. And so, and I find that a lot of people have some experience with it, and so I like to get them sort of like digging into and diving into other tools.

Tony Martignetti
Before we go to another tool, I want to take a little digression to ask you what Tavlin Consulting is. It's T-A-V-L-I-N.

Allison McMillan
Yes.

Tony Martignetti
Tavlin. What is that?

Allison McMillan
Yeah. So my consultancy is about helping people communicate more effectively. I have a software engineering background and a nonprofit background, and I've held executive leadership positions in both worlds. And so a lot of what I do is facilitate workshops, trainings, and conversations that sort of bridge that technical and non-technical and help people come together. So Tavlin is actually a Hebrew word for, like, a Spice mixture. Mm-hmm. And so, that's what I'm always doing is, you know, one spice is fine, but a spice mixture together, right? All those different flavors, all those different people and things coming together, you know, just elevates everything and elevates what it can be. And the way that it's used in Hebrew is very often there's like different tavlin, there's different tavlins, there's different spice mixtures. And I like to think about that also as companies and organizations that no two are exactly the same, right? Even if it's like a little more paprika or a little more— exactly, exactly. And so, yeah, so that's what Tavzin means.

Tony Martignetti
Okay, cool. Thank you. All right. All right. Let's go to another tool. I like distinguishing these tools.

Allison McMillan
Yes. We have two more tools left. So next we covered Perplexity. So Perplexity is really used a lot for research and statistics. You know, what people often find, and something that AI often gets knocked on, is these hallucinations, right? Which are real, which are true, that if you ask for a study or statistics, you know, some of these other tools will make things up and will hallucinate. And even if you say, okay, where are you citing that statistic from, you know, give me a link to the study, it will, it will give you a link. And if you click on it, it's like, oops, that doesn't actually exist. Yeah. Which is a big worry because it then creates more work to actually like get the statistic, chase it down, make sure it's real, etc. So that is not so much the case with Perplexity. Perplexity gives you the list of sources, cited sources. It also, for each statistic, it will tell you how many sources or how many places it's cited in, which really gives you as a user the chance to say, okay, that one looks really reputable, that one not so reputable, right? So you get to choose which statistics you pull in. Or maybe there's a popular report that then gets cited in McKinsey and Forbes and all these different places. You can see that, and so you can say, okay, this is like a very legitimate study, a very legitimate statistic that I want to pull in for a donor communication, for training that you're doing, for conference talks. You know, I often use it in blog posts to, you know, support some of the, some of the things that I'm talking about. So yeah, so Perplexity is great for those research and statistics.

Tony Martignetti
Statistics, data, you can trust it. It's more trustworthy than the other models.

Allison McMillan
Exactly. Exactly. The other big difference that I always say is Perplexity is very much like a science. It's very scientific. So in Gemini or Claude, if you ask it a question, usually it ends with, gives you information, and then usually it ends with, would you like me to create a slide deck for you? Would you like me to do this? It'll sort of guide you into your next step. Right. Perplexity is a little bit more like, I gave you the answers. These are the answers. We're done now. So it doesn't so much guide you anywhere else. It's a great tool. And I also think it's very like— people use it specifically for research and statistics. And so, you know, it's personality. I like to think about some AIs as like what their personalities are. So the Perplexity personality is very much like, I have given you the information and now we're done.

Tony Martignetti
It's straightforward. And maybe not, hopefully not snarky. Yeah, I like to keep that to humans. Yes, but you have it. I just, yeah, it's done.

Allison McMillan
Yeah, we're done together. Exactly. All right. So yeah, and Perplexity is, it's great on a free model. It also has a pro model, and they run a number of different specials. So you can often get a year of Perplexity Pro free through a variety of different avenues. So I always recommend that folks sort of like Google around or check around to see what the offerings are, because it's all— you can often get extended time period to use it, to use the pro version.

Tony Martignetti
Okay. And you got a 4th one?

Allison McMillan
Yes, the 4th one is the most fun. It's called Suno. How's that

Tony Martignetti
spelled?

Allison McMillan
S-U-N-O. And Suno is song generation. So I use Suno a lot for background music, for like, you know, or introduction music for a virtual meeting. I use it often as, you know, sort of like an icebreaker, or, you know, it's just, it's a platform that it just puts a smile on people's face.

Tony Martignetti
If we have a Zoom together, is there going to be welcome music? Do you do that for all your Zooms?

Allison McMillan
No, not all. Not all my Zooms. But yeah, if I'm doing an all-hands or like Like a larger— yeah, exactly.

Tony Martignetti
Just background music as we're entering the room?

Allison McMillan
Exactly, exactly. Sometimes I'll also do like a thank you song for certain— because it really— and it's very, very simple. Like, you can just put in— the example that I gave to this group was the prompt that I put in was an upbeat tempo, a song for nonprofit professionals describing all of the different hats that they have to wear in an upbeat tempo. And it comes up with, you know, a multi-minute song that has lyrics that, you know, are generally sort of entertaining. And so it's just a fun one to play with and to kind of have in your back pocket. Because sometimes, you know, you just need that morale boost. You need a little bit of fun to come in.

Tony Martignetti
I'm sure you disclose that it's Suno-generated.

Allison McMillan
Yes.

Tony Martignetti
You're not copyrighting these songs. No, no, no, no. I'm also not only a software engineer and nonprofit consultant, but I also songwrite and do music and lyrics. No, you don't claim that. Okay. Good. All right. Straightforward honesty at Tavlin.

Allison McMillan
Yeah. So those are the 4 tools that we walked through. And then we walked I always get asked about how, especially nonprofits, especially folks with limited budgets, how they should pick a tool, because there's so much out there. Again, there's so much out there. There's different things that can be used for different specific circumstances and situations. So for that, I talk about it a little bit like AI tools as supermarkets. Where you always have your go-to supermarket, right? The one that you know how all the aisles are laid out, that it basically has everything you need.

Tony Martignetti
I'm picturing right now. I know where the produce is.

Allison McMillan
Right, exactly.

Tony Martignetti
In the produce, I know where the bananas are.

Allison McMillan
Yeah. You probably have even like a specific time or times during the week that you go, right? So you might like know a cashier or a couple of cashiers, right? It's the place that you sort of like Have invested in, that you know really well. They know I bring my own bags. Exactly.

Tony Martignetti
This guy doesn't want the plastic.

Allison McMillan
Exactly. And then there are other supermarkets that you go to for specific purposes, right? You might go to Trader Joe's when you want to stock up on really good snacks, or to, you know, an H Mart when you want more variety of different sort of sauces or noodles or etc., right? And those might be supermarkets you don't go to as often, right? Maybe you go once a month, maybe you go once every couple months, and you go, you go for very specific purposes. Yeah. And so I think about AI tools in the same way, that think about your use cases and who's going to use it. Is it going to be one staff member? Is it going to be multiple staff members? What are they going to be using it for? What are the use cases? And then pick one tool, pay for that tool, go all like train that tool, really make it sort of know your people, your business, like, you know, establish projects or gems, whatever you're gonna do, like spend the time there. And then the other ones, you know, all of these you can pay for month by month. So the other ones, you know, for the month that you need it, pay for a pro version, pay for a paid version for a month, and then discontinue it and, you know, go back down to free. to the free version, because those can sort of— you can toggle them up and down easily. So I sort of think about it as like this supermarket.

Tony Martignetti
I like your supermarket metaphor. That's cool. All right. You talk some about limitations, pitfalls around introducing AI, getting started.

Allison McMillan
Yeah. Yes, I talk about a couple of things related to limitations and pitfalls. One is being aware of when you're using a free version versus a paid version. There are differences in the quality of the output. There are differences in if the models are training, right, or using the data that you're putting in to train their model more broadly, more widely, right? What proprietary information you put into a free version of an AI is definitely something to be very aware of. So that's one. The second thing that I talk about a lot is compliance. So a lot of these AI tools, I always recommend that folks try to investigate if the tool that they're using— this is especially with note-takers, because I feel like there are so many note-takers coming out. People are always asking about note-takers particularly. There is a compliance called SOC Type 2, which is basically an independent audit that a company has to get that shows that they handle your data and the data in a specific fashion. And so it just provides a little bit more security, but it can be time— like, it can take a lot of time to get SOC 2 compliance, and it is expensive for companies. And so some of the newer tools don't necessarily have that. They don't have a trust center. They don't have all of that sort of thought through and figured out. So, I always like to say you should always be aware of what the security protocols are for the tools that you're using. The third is just to have a— well, the third is around hallucinations. They are real. So, just always make sure that you're checking your sources and making sure that you're not pulling in information that you're not certain is real and true. And lastly, I always recommend that folks put some sort of AI policy in place. It might change a dozen times over the next year, given the pace at which AI is accelerating, but just a little bit about what's the acceptable use, an understanding of, you know, risks and how you intend to manage those risks, and a little bit about oversight if you intend to have any oversight. Great. It doesn't have to be long. You can actually have AI generate a first draft of a policy for you. But I always recommend, because you'll have in a— in any team, you'll have this spectrum, you'll have this range of folks that are like really, really hesitant and haven't touched it at all, and you'll have folks that are early adopters and they're jumping into a ton of different tools and really trying to connect everything together, etc. And you want everyone to be on the same page about what's okay to do and what's not okay to do in this moment for them and for your organization.

Tony Martignetti
Let's pull a little more on the thread about, you know, your own data protection, making sure that you're— as you're uploading content, it's not being used for general consumption and learning of any of these models. How do we make sure that that's not the case?

Allison McMillan
Yeah, I think almost the On all the paid— on all the free versions, I believe that's just part of— there is no way to sort of guard against that. That's why I always recommend, you know, I know that companies don't have endless dollars, but it is good to sort of pick a tool and pay for it. And then in all the paid-for tools—

Tony Martignetti
It's like your favorite supermarket.

Allison McMillan
Right, right. And then all the paid-for tools, there is a way to toggle off some places. hide it more than others. But there is in settings a way to toggle off so that your materials can't be used to train sort of larger models. You just have to find the setting.

Tony Martignetti
Okay, but it's worth looking for.

Allison McMillan
Yes, yes, absolutely.

Tony Martignetti
Because if you want to get close to one of these models, you're going to be giving it some proprietary data that You may not want released.

Allison McMillan
Yeah, yeah, exactly. I also often tell people that, you know, you can often get what you want out of AI without having to put as much proprietary information as you think you do into the system, right? And so it's worth thinking about, you know, I think it's easy to be like, okay, I have these— the spreadsheet, I'm just gonna put it in and do X, Y, and Z. But I think it's always worth thinking about, okay, for this task that I'm doing with this AI tool, what is the outcome? And I have the spreadsheet. Do I need all the columns? Can the people just be numbers? Do I need first and last names? Like, to, you know, it often is just a couple of extra minutes of thought and really thinking about what's the outcome that I'm trying to get to. And then you can often get to a place where you actually don't have to put a lot of proprietary or sometimes any proprietary information into the system in order to, in order to get the result that you want.

Tony Martignetti
That's good advice. It's worth thinking through. What do you need from it? What's the purpose here? Right. Like, perfect example, do I need first and last name? Do I need any names at all?

Allison McMillan
Right, exactly.

Tony Martignetti
So Allison, share something else that we haven't talked about yet that you talked about in your session, or, or some questions, maybe some questions that Anything that stuck with you from your session?

Allison McMillan
Yeah. I think the last thing that we talked about in the session was I gave a few sort of my top tips and tricks. And so a couple of them are, I, when I'm learning something new, when I'm trying to get ramped up on something really quickly, I will sort of give the tool a couple concept, and I'll say, explain this to me like I'm 5, and then explain it to me like I'm 10, and then explain it to me like I'm a new college grad, and then explain it to me like I'm a professional with 25 years of deep experience in this area. And it's a really great way to just get really deep into, like, into a topic very quickly and understand the jargon and at what level different people operate. So that's one that I really like. I often will use I am as a— if I'm trying to, right, like whether I'm a client or a consumer or whatever, I'll sort of give AI that whatever persona I'm trying to send a message to or connect with, right? So I am, you know, a nonprofit executive director. Right. I'm a director with a team that has varied adoption uses. I really want them to adopt X tool. What plan would you recommend for us to go? Or even, you know, I'm on a team, I'm in a nonprofit, my executive director wants me to be using AI more. I don't really want to. You know, what plan might work for me if it's the, you know, executive director sort of trying to— Yeah.

Tony Martignetti
I had a guest earlier who said he uses his prompts, you are. He actually channels the Backstreet Boys. You are, you know, I don't know the song, but he does. You are. Does it make a difference, you are or I am? No.

Allison McMillan
Same difference. Yeah. I also tell people that sometimes you just have to close the chat and open a new one. Sort of like, you know, with any technology, sometimes you just have to turn it off and wait 5 seconds and turn it on again. So with AI, like, if you— sometimes it can just go in the wrong direction, and then once it has that context and it's sort of traveling in that path, it can be really hard to do a reverse direction or to be sort of like, no, no, no, this is like nowhere near what I want. And so sometimes just Closing that chat and starting fresh is really the best way to go and will limit your frustration level.

Tony Martignetti
Okay.

Allison McMillan
Yeah. So that's a good one.

Tony Martignetti
These are Allison's tips and tricks. Yeah. Pro tips. Go ahead.

Allison McMillan
I'm trying to think which other ones there are. Yeah. I think those are the top ones. Those are really good go-tos to think about.

Tony Martignetti
So just leave us with some inspiration for the rest of us.

Allison McMillan
Yes, I think some inspiration is AI tools are rapidly improving month by month. I also say folks that tried it 6 months ago or a year ago and said, this was all hype, I don't understand why everybody's talking about this, should try the tools again because they are getting better and better all the time. And again, your adoption, your usage, you don't have to use the most complex scenario. You don't have to use what the AI master in whatever host has said and, you know, tying these 8 tools together and doing those sorts of things. It's really great to just start somewhere and get yourself to that first, what I call either like an aha moment or even like an, oh, that's interesting moment, and then go from there.

Tony Martignetti
All right. Allison McMillan, CEO at Tavlin Consulting. Tavlin, the spice, you know, the spice hybrid, the spice mixture. Bringing it all together. I love seeing women in software engineering too, or any engineering. So where'd you do your degree in software engineering?

Allison McMillan
I don't have a software engineering degree. I have a political science degree and then had a startup, taught myself how to code.

Tony Martignetti
Taught yourself?

Allison McMillan
Yes. Oh

Tony Martignetti
wow.

Allison McMillan
Yes, before boot camps were a thing, before— yes, you know, 15 years ago. And then moved into software development and then You know, software engineering leadership, worked at companies like GitHub. So yeah, it's my, my path has been a windy one, but no, no computer science degree here.

Tony Martignetti
All right, that's okay. I still love seeing women in engineering and any of the sciences. All right, Allison McMillan, thank you very much.

Allison McMillan
Thank you.

Tony Martignetti
And thank you for being with Tony Martignetti Nonprofit Radio coverage of the 2026 Nonprofit Technology Conference.

Kate Martignetti
It's time for Tony's Take 2.

Tony Martignetti
Thank you, Kate. My book title is lengthy, but very precise. You're gonna, you're gonna know exactly what the book is about. And you'll see that my personality emerges in the book through the title, that I want you to see that from the title. I had a really long title. But I ran into a constraint at Amazon. They only allow 200 characters for the book title and the subtitle. If— and your book goes nowhere these days if you don't publish through— print through Amazon. Not publish. I'm self-publishing. They're not a publisher. They're a printer. If you don't sell— that's what I— they're a bookseller, really, and a printer, of course. If you don't sell your book through Amazon, You don't get very far. So their constraint pretty much sets the tone for the book titles in the world. So here is the new book title, Planned Giving Accelerated: The Cut-Through-the-Shit, No-Nonsense, Practical, Step-by-Step Guide to Start Legacy Giving Fundraising at Your Small to Midsize Nonprofit Simply in One Week with Bequests. Wow. Now that, that happens to be exactly 200 characters, including spaces. They're very precise about that, includes because the space is a character, but they're, they're, they say it explicitly. So I used to have, uh, it was finally some planned giving accelerated. Finally someone wrote a cut through the shit, no nonsense practical. But I had to cut out, uh, finally someone wrote a, but I like the. Yeah. Plan Giving Accelerated, the cut-through-the-shit, no-nonsense, practical, step-by-step guide, et cetera, et cetera. I like the. It's not— there's not one among many. How many, how many cut-through-the-shit, no-nonsense, practical, step-by-step guides to start legacy giving fundraising at your small to midsize nonprofit simply in one week with bequests could there be? There could only be one. That's this one. The. The— so sometimes constraints lead to more precision because you gotta, you gotta tighten it up, you know, you cut out the flab. So, um, so that was— that actually is an advantage. I like that. The Cut Through the Shit, No-Nonsense, Practical Step-by-Step Guide to Start Legacy Giving Fundraising at Your Small to Midsize Nonprofit Simply in One Week with Bequests. Now, I used to have following that, a parenthetical in the title. The title may be longer than the book. Did you need a nap? That was in the title. Now that can't be in the title anymore because I hit this Amazon 200-character constraint. But, but it can still be on the book cover. It's just not part of the title. You can cram and squeeze as much as you want onto a book cover. It just can't all be part of the title. So that's what I'm going to do. So we're not losing. I am not surrendering. The title may be longer than the book. Did You Need a Nap? I think that's important. I think it sets the tone. You know, this is not going to be an academic type text. It's not going to be your average journal type text. This is a— Yeah. It's a book you're gonna have some fun with. So just wanted to keep you apprised of the latest developments. That's a significant one because it's the title. Again, Planned Giving Accelerated, the cut-through-the-shit, no-nonsense, practical step-by-step guide to start legacy giving fundraising at your small to mid-sized nonprofit simply in one week with bequests. And then of course, the title may be longer than the book. Did You Need a Nap? But that's not officially in the title. But that is Tony's Take Two. Kate?

Kate Martignetti
I like how you said you are not surrendering to Amazon's character count limit. You are not surrendering. No surrendering here.

Tony Martignetti
No surrender. No reset. Absolutely. It came out better. It came out better.

Kate Martignetti
I'll be honest, I'll probably just type into my Amazon Planned Giving Accelerated by Tony Martignetti. But your title does sound very, very special.

Tony Martignetti
It is.

Kate Martignetti
And very larger than life, like I know you are.

Tony Martignetti
Oh, thank you. All right. I'm glad you didn't say sounds very interesting. I was afraid you were gonna say, oh, it sounds very interesting, which is kind of dull. No, sounds special. Thank you. Thank you very much. I think it is, and we're sticking with it.

Kate Martignetti
We've got beaucoup buttloads more time. Here is your AI acceptable use policy.

Tony Martignetti
Welcome back to Tony Martignetti Nonprofit Radio coverage of 26 NTC, the 2026 Nonprofit Technology Conference in Detroit. My guest now is Eric Molho. Eric is Founder of Bon Partners. Eric's topic is Step by Step: Creating Your Nonprofit's AI Acceptable Use Policy. Eric, welcome to Nonprofit Radio.

Eric Molho
Thanks for having me, Tony. It's

Tony Martignetti
a pleasure. Is this your first NTC together? Or first NTC, I should say? It is. Yeah, yeah. It's a great conference. This is our 12th coming up.

Eric Molho
Oh my goodness. Yeah, that's exciting. It's been wonderful so far.

Tony Martignetti
And we'll be at number 13 in Portland next year. Just give us an overview of the topic before we dive into some detail.

Eric Molho
Yeah, yeah. So there are 2 things I'm trying to really address here with this topic. The first obviously is this whole notion of the acceptable use policy. So we know everybody is using AI, right? Like I was just scanning some of the— you've probably been talking a lot about it over the days, right?

Tony Martignetti
Yes.

Eric Molho
But a lot of organizations haven't yet caught up in terms of their policies and in terms of providing people with guidance and guardrails and direction about how we use this as an organization. And some organizations may not even be aware that some of their staff are using some of these free tools and perhaps doing some of their work leveraging AI and may or may not be doing that in the way that we want them to. So, the first thing that I'm going to be talking about is actually some of those policies that I think organizations need to be thinking about. The broader issue that I've— that I'm hoping to talk about is just how we build a culture within an organization around usage of AI so that we are really embracing it, so that we're talking about it, so that we are leveraging the expertise that we have in the organization itself as we start down this journey.

Tony Martignetti
Very good. Thank you. So let's dive into these topics. Both the— sorry, there we go— as I unclip a little microphone. So yeah, your policy and also a culture. A culture. All right, so we could take them in that order. Let's talk about what belongs in your policy. Yeah, what belong— what are the key pieces?

Eric Molho
Well, there are quite a few aspects that people need to be thinking about. What I'm going to focus on today is trying to bring that down to a manageable number of things. The first thing that most people probably think about when they're thinking about policy is protecting your data, right? We— nonprofits are using all sorts of data that they need to be thoughtful of when they're using some of these AI tools. So that may be your donor data, that may be client information, it could be grant information. Whatever it is, you've got— Right. stuff out there. Yeah. That you have to provide some guidance to your team as to what they can do and what they can't do. In other words, AI is a fantastic tool for segmenting your donor data or taking a look at donor trends or trying to understand when and how gifts are coming in. But we can't do that just by doing a download of our Salesforce information and chugging everything up into ChatGPT and saying, tell me what the answer is. So we have to have some rules in place around our data to make sure that people understand what is personally identifiable information, what our tools and processes are for gaining approval, and when we're going to use some of that. Whatever that looks like, we've got a lot of data that we have to be thoughtful of, which is one piece of that. Another piece is transparency, thinking about when are we disclosing, when are we talking about AI. And this comes about You know, in very simple aspects, right? Many of us use AI meeting note-takers, or we use record meetings or record information, and then we're leveraging AI to help us maybe generate notes or whatever it might be. We probably as an organization need to be clear as to when we're using this, when we aren't, when we're telling people we're using it, when we aren't telling people we're using it. Another area that we that I think is valuable for folks to be aware of is, of course, accuracy. So, data make— AI tools make mistakes. They have hallucinations. Yes. They make stuff up. What's our process? What's our approach to making sure that we aren't feeding up wrong information, incorrect information within the organization? Many organizations probably need to have some sort of formalized 2-proofing process, 2-person proofing process, or another set of eyes taking a look at outputs and the results that people are generating when they're moving forward. There are ethical and sustainability concerns. So, you know, I've worked with organizations that are environmentally conscious. I've worked with other organizations that are really focused on content creators. Well, those organizations may have very distinct rules and may have very distinct policies that they want to develop that match their values and match their mission because AI has a particular role to play in some of those areas. So lots of aspects to consider.

Tony Martignetti
Okay, so these are elements of our acceptable use policy.

Eric Molho
Exactly. And the key is to start the conversation, right? The key is that we want to engage our staff and we want to give our staff, again, kind of the rules of the road so they don't get into trouble and we're not getting into trouble as an organization.

Tony Martignetti
What's your advice around the accuracy part of that, ensuring accuracy? You mentioned, you know, 2-human, maybe 2-human review. Yep. What else can we have in place? Because that is a risk that I think— Huge risk. Well, these are all huge, but that's the one that puts a lot of people, I think, off. Off AI, or correct, will only— they'll only adopt AI for the most mundane tasks.

Eric Molho
Right, right. Well, we, uh, first of all, we can— we can— when we are doing our prompts, and I'm thinking here, I— we should probably make a distinction here. There's AI as a massive topic. I am talking, generally speaking, about the large language models, the generative AI that we are using through a ChatGPT, a Claude, a copilot, something along those lines. Many of us have used these tools. There are lots of different ways that they are embedded in other software and in other programs, but generally speaking, I'm going to focus on that area. One of the things we can do is, as we're working with the AI tools and as we are prompting them, we can say, show your work. What's your source for this data? Where are you getting this information from? And actually clicking on The link that it tells us. So making sure that we are, as an individual, when we're getting some quote of data or information fed back to us, we can take a look at that information. As an organization, then, we can also have policies and procedures in place in terms of how we are fact-checking and how we are proofing this. This may be a new policy or a new way that we go about work, but before something goes out on email, Right. Before it goes out in our annual report or in our fundraising appeal, do we have somebody who isn't the writer having eyeballs on that and highlighting, hey, you mentioned this fact here, tell me where it came from, how did you come about doing that? So there are things both technologically and within our processes that we can do to make sure that people are protecting themselves.

Tony Martignetti
How about the transparency? You mentioned there may be use cases where you're not Revealing that you used artificial intelligence? Well, I mean, that could be something as mundane as, you know, write me an email, right? And then obviously the person reviews the email. Are there other cases where it's not necessary or where you actually have an obligation to disclose that you have used one of the tools?

Eric Molho
I think it's very important. The example I mentioned And I do think it's very important when we have meetings and recordings and that kind of information that we are disclosing that.

Tony Martignetti
Okay, like you mentioned, meeting notes, transcribers, note-takers.

Eric Molho
I do a ton of interviews in my practice, and so the start of every conversation is, hey, wanted to call your attention to the fact that I've got my note-taker in the room. Is that okay with you? And make sure that we are disclosing that. Within organizations, I think I want us to disclose the use of AI to build that culture piece, right? I want people— in my perfect world, more people within the organization would say, hey, I'm working with AI to help me solve this problem. What do you think about this result? Or, I looked at the data, then I went to AI, then I refined it. Here's my final work product. So that we are creating a culture within the organization that says these are valuable tools. That we all need to be using and we all need to be learning from. I think particularly, and I'm seeing less of this, but I think there is a little bit of this that's still, that was still the case when a lot of these tools were launched in 2022. A lot of fear, a lot of, oh my goodness, this is cheating. Oh my goodness, this is going to—

Tony Martignetti
Going to take jobs.

Eric Molho
Going to take jobs and all of that. And I think we've got to really turn that on its head and we've got to share with one How we can be more effective, how we can be more efficient, how we can be more productive when we're leveraging these tools so that it becomes, again, a tool, not replacing somebody or not replacing the critical human element, but one of the tools we use. In the same way that we might say, hey, I took your data, this is the analysis I did in Excel, what do you think, does this look accurate? We need to get to a place where where we're also saying those and disclosing that this is one of the valuable ways that we're accomplishing our work. And that will both send a signal that this is okay, this is how we do things, and hopefully it will help us learn. Yeah. One of the things that I— one of the reasons I love doing these presentations is invariably at some point in the presentation, it's like, well, how are you using these tools? And I continue to be blown out of the water. People are doing crazy, wonderful things with these tools, and the more we begin to be exposed to how people are leveraging them, the more we can look at our own work and say, oh, I wonder if it could help me with this.

Tony Martignetti
What are some crazy things?

Eric Molho
I've heard people talk about how they've leveraged AI to— with dialects of language, that it is actually more effective to be able to say, hey, we're working with this population from Guatemala or something like that that may have some distinct dialects, and the AI tool is pretty darn accurate in terms of translation. I had one of the examples, not to steal my own thunder for tomorrow, but one of the examples I use—

Tony Martignetti
Nobody's going to hear this before tomorrow.

Eric Molho
Oh, all right, all right, that's good.

Tony Martignetti
We're not that fast.

Eric Molho
One of the examples I use is a Food Shelf. Food Shelf, and they do a lot of Meals on Wheels. It's one of their programs. They do multiple programs. They realized they were spending a ton of time answering questions from folks. When's my meal going to get here? Who's my driver? What time should I expect them? Like, a lot of manual phone calls coming in. They're developing an app with AI now that would be like DoorDash that says to these folks, hey, your meal left our warehouse. Julie is your driver. She should be there between 11 and 11:30. I'm like, how cool is that? Like, just amazing stuff that people can come up with.

Tony Martignetti
Okay. Yeah. Yeah. Yeah. So on the policy, let's move from the policy to the culture.

Eric Molho
Yeah.

Tony Martignetti
I mean, folks may be using this and you don't even know.

Eric Molho
Correct.

Tony Martignetti
So they're They're certainly not adhering to any policy because you don't even know that they're doing it. That's the worst case. That is worse than nobody even adopting it.

Eric Molho
I think that's bad on both ends.

Tony Martignetti
Well, I'd rather— wait, I guess it's the lesser of two evils. If I was the CEO, I'd rather nobody be using it than people be using it rogue and we don't know, and they're not disclosing it because we don't know, and we don't know what they're doing with it. I'd rather the former, but in today's world, they're both setting you behind.

Eric Molho
Correct.

Tony Martignetti
Okay. So a culture around the adoption of and responsible use of these large language tools. Yeah. That's a very broad— well, that's why I'm consulting an expert. Yeah. It's a very broad question for

Eric Molho
Yeah, a lot of it we've been— is the theme that I've been mentioning moving forward, is I think it's very valuable for senior leadership to be able to disseminate out to the organization, we're going to use these tools. We see value in them. We see value in you too. We're not going to use these tools to get you out of a job, but we see value in these tools, and so we're going We need to communicate out to people, and we have rules about how to use them, right? Like any tool, they can be used for good, they can be used for evil. So we're going to use them for good, and this is the way we're going to protect our client data, this is the way we're going to protect our donor data, etc. And this is some guidance that you can have as an employee, as somebody who's working, or a volunteer, in order to— Right. In order to use the tools effectively, so first of all, we have to say that we're going to use them. I think secondly, as I alluded to earlier, I think we want to create a culture within our organizations where we're learning iteratively, right? And that's that's what this whole process has been. What that's what AI is all about, right? We're constantly throwing prompts back and forth to the tools to say, okay, now do it this way. Think about it this way. Reframe it this way. I think we want that within. our organization where people are, how are you using it? Oh, I'm using it for this. That sounds interesting. Did you get good results? Didn't you? I tried this prompt. I tried phrasing it this way, and leveraging our own learning so that we're moving the work forward.

Tony Martignetti
I like your analogy of learning from each other the way we want the tool to learn from us.

Eric Molho
Yeah, exactly. Exactly. There are experts. I would say there are probably There are probably experts in any organization, whether you've got 8 people in your organization, 80, or 800. There are already people who are pretty far down this rabbit hole who are experts in one of these tools, and we need to leverage that. We need to take advantage of their learning and what they've discovered.

Tony Martignetti
So we need leadership buy-in for this AI culture creation. How do we I don't know, do we need to make the case? Maybe there's some CEOs that, well, they may see the value, but they also may be fearful of the transparency, the accuracy, the ethics, the environmental sustainability. So how do we make the case that the benefits are going to outweigh the costs and concerns of the CEO?

Eric Molho
You know, one of the things, one of the little cheats I have as a consultant is when I'm in doubt as to how to answer some I'm like, it always comes back to the mission, which is true. It always comes back to the mission. Nobody got into this work because they want to run spreadsheets for their whole life. Well, maybe a few finance people did, but most of the people who are in nonprofit work, most of the people who are here at this conference today, got into this work because they're passionate about the work that they're doing, because of the mission that they're doing. So I would say to a CEO, to an executive director, you've got this amazing mission. to accomplish, and you don't have nearly enough people, and you don't have nearly enough money to get everything done that you want to get done. And now we have this transformative technology, this amazing AI tool that we're all exploring, and we're all on this journey that can help you to become more efficient, more effective, that can leverage your— the talent that you have brought onto your organization to achieve your mission. So I would frame this conversation around the culture, I would always bring it back to the mission and say, we have the capacity to help more folks or to achieve more of whatever it is we're trying to achieve when we are leveraging these tools to create internal efficiencies, to help us think differently, to help us troubleshoot in new ways. So I think it's that big picture That the CEO is, or the executive director is really responsible for disseminating out. I think what's also important on the culture piece, especially this intersection of acceptable use and the culture piece, is that we not get lost in the lawyers and HR and IT, right? These are tools to help people get work done. And so in the organization, We want the focus to be on getting the work done, getting it done effectively, getting it done ethically, getting it done within our values. Our IT partners help us to make sure we're keeping it safe, to help us understand that we've got consistent tools, that people aren't going rogue on us. Our HR and our lawyer folks help us make sure again that we're not violating policies, that we're not violating laws, but the big Big picture, mission, what are we trying to accomplish? That's what we need to keep our eyes on, and the policies and the conversations that we're having have to serve that.

Tony Martignetti
And as we're focusing on big picture and having this conversation with the leadership, we may identify maybe small things we can begin with.

Eric Molho
Absolutely.

Tony Martignetti
I had a guest call them pilots. Yeah. Let's test. Let's see. Hopefully it's going to be successful. The test is going to— the H naught is going to be proven that this does in fact work to the benefit of the mission. We can start with small steps to bring leadership along.

Eric Molho
Absolutely. And I think we can start in small steps, both in that literal sense of pilot projects or someone, You know, what would happen if we rethought our email campaign or we did a deep dive in our email campaign leveraging AI and trying to make that more effective and more efficient? Great pilot project. We could go and run with that. But it's also small steps in terms of, you know, an executive director who's at this conference goes to a ton of sessions. What can they do next Monday? What can they do in 3 days? Well, why don't you Gather together the folks who are already using the tools. Let's hear how they're using it. Let's just have a lunch and learn and figure out where people are at. Let's take some small steps. We don't have to hire a trainer or consultant. Let's all gather together in the conference room and say, hey, how are you using these tools? Where are you finding value? Where are they wasting your time? Where are they adding value to what you're trying to do? So the small steps is both the literal, how do we launch a project or How do we do a pilot? But it's also on the culture side. How do we open this conversation to make sure that we understand where people are at?

Tony Martignetti
Who should be— jumping back to the policy because we're talking about the team, meeting with the team— who should be contributing to the policy?

Eric Molho
Yeah, I think obviously there are some subject matter experts. If you have a larger organization where you've actually got maybe HR and development and IT, you would probably We want all of those folks in there. If you're a small shop, you may be able to develop and leverage a policy within, you know, by leveraging 1 or 2 people within the organization, leveraging AI as a tool to help you develop the policy. We'll be talking about that. Yeah. And one of the, you know, one of the assets that most small organizations have that I don't always think they take great advantage of, your board members. Your board, every single person on your board is going through this exact same conversation in their organization, and they have expertise to share as well. What I wouldn't do, personally, what I wouldn't do is just call the lawyer and say, we need a policy. I think you want to take a look at how people are using this. I think you want to take a look at the existing policies you already have in place. You probably have a data protection policy already in place. We may just need to modify it. Right. Or we may need to clarify or do some training with the staff that says, okay, here's our policy, here's how it applies to AI. But I don't think we want to just make this a legalistic conversation. There's a role to be played there. My brother's a lawyer. I'm pro-lawyers. But we don't want that to be the focus. We want the focus to, again, come back to the work that people are doing to achieve the mission.

Tony Martignetti
You mentioned using AI to create the AI acceptable use policy.

Eric Molho
Yeah.

Tony Martignetti
What are you going to say about that?

Eric Molho
I think it's a great starting point. I think this is, you know, you think about any policy, I don't care what it is, you want to revamp your data use policies. A lot of executive directors might spend 4 hours searching the web, going to other like similar organizations. They might call their executive director friends. They might call a lawyer. They might, you know, gather 5 different drafts and then cobble together their own new tool. That's not an uncommon way of developing a new policy. Well, why not craft a great prompt and start there? 30 seconds of work and you've got maybe 80% of it in place. Take a look at that, circulate that with the team, and then when you're, when you're at a place where you're like, yeah, I think this is it, then check it over with the lawyers and the board to make sure you're, you're in the right place.

Tony Martignetti
Okay. Start with, you are not a lawyer contributing to an AI acceptable use policy. You're not a lawyer. You're the marketing, you're the whatever function, fundraising, your social media, whatever.

Eric Molho
I think the other thing that when you're starting to develop these policies, again, where AI could be a real benefit in them, is the nuance of your values. So in other words, in that prompt, if I were to prompt ChatGPT to develop a policy for me, I would probably give them a lot of— I would give it a lot of context. I would say, I'm an 8-person team. Our mission is to do this. We have really sensitive data from our donors. We have some federal grants. Kind of lay out that. And I would say, and here's our mission and here's our values, because I think it's important and we can create policies that reflect those values. And that's part of why every organization needs their own policy, because we all have different values, we all have different priorities, and we need to make sure those are reflected in the policies.

Tony Martignetti
You want to leave us with something else that we haven't talked about yet that you are going to cover in your session? Or maybe a little more detail than you— don't hold back on Nonprofit Radio listeners.

Eric Molho
Yeah, yeah, yeah, yeah, yeah. I mean, in terms of the content that I'm going to present, one of the foundational frameworks that I'm just going to skirt by, but I would recommend as a resource to the listeners, was created a couple of years ago by the folks at Fundraising They have a comprehensive acceptable use policy. It's got about 11 different elements to it. It includes just about everything you could imagine. And the reason I think it's so valuable is not that every organization needs 11 pages of content for their acceptable use policy, but it's a great thought experiment. It kind of walks you through, have you thought about this? Have you thought about that piece? Have you thought about this particular piece? And so I often, and it's included in my presentation, I hold that up as sort of the— this is the full Cadillac version. This is the high end that you would— that you may someday achieve. And then for our session, we're going to dig into kind of the first 4 or 5 that you need to tackle.

Tony Martignetti
All right. All right.

Eric Molho
The fundraising.ai. That's the resource.

Tony Martignetti
The culture and the plan.

Eric Molho
Absolutely.

Tony Martignetti
All right, that's Eric, Eric Molho, founder at Bon Partners, B-O-N Partners. Eric, thank you very much for sharing.

Eric Molho
Thanks, Tony, appreciate it.

Tony Martignetti
It's been my pleasure.

Eric Molho
All right.

Tony Martignetti
And thank you for being with Tony Martignetti, Nonprofit Radio coverage of the 2026 Nonprofit Technology Conference.

Kate Martignetti
Next week, more from 26 NTC with Cybersecurity on a Shoestring and Disaster Recovery and Incident Response for Accidental Techies. If you missed any part of this week's show, I beseech you, find it at tonymartignetti.com. Our creative producer is Claire Meyerhoff. I'm your associate producer, Kate Martignetti. The show's social media is Marc Silverman is our web guy, and this music is by Scott Stein.

Tony Martignetti
Thank you for that affirmation, Scotty. Be with us next week for Nonprofit Radio, Big Nonprofit Ideas for the Other 95%. Go out and be great.