What We’re Seeing Across Nonprofits AI Trends That Actually Matter

about AI trends that actually matter. All the things AI. So Joe and Cal over to you. Have a great webinar. >> Hi, thank you Artha. Nice to see everyone. I'm Joe D. Javanni. I'm one of the co-founders of Tap Network along with Kyle Barkkins, my other uh co-founder. My background is in marketing. Kyle is our technology chief architect. And our our goal at Tap Network is to help organizations modernize their marketing and technology to accelerate impact. And we're also a partner with uh Tech Soup. So we'd love to uh go a little deeper today on AI. It's one of the things that we offer through Tech Soup. If you go to Tech Soup, click on services, you'll see website services, um digital marketing services, and AI for nonprofits. That is all the work that we do here at uh at tap and we are tech soup's exclusive partner for these services. So excited to have the uh conversation today. So three things that we're going to discuss today for in terms of AI is what's actually happening in the field. We know almost every nonprofit is using it but very few can point to actual results and impact. We'll discuss why and what we could do about it. And then what what works and what's hype, what's really uh happening out there. What are people paying for? What kind of ROI are they getting from the platforms and systems they're using? And how is it really impacting efficiency and impact with the communities you serve? and then where to start and looking at not just the holistic picture of your nonprofit but enabling you to pick one workflow, one number, what can you do in in one quarter to get it done. And this is based on our work with over 180 nonprofits in the AI space, over a 100 recorded um conversations, diagnostic surveys, and an entire index that we did across the nonprofits that we work with through Tech Soup. Great. So, what we're seeing um today, we'll look at the patterns that are happening in a lot of nonprofits that we serve and go through this step by step with you. >> Cool. So, the first thing um we're going to go through is is where we're seeing um seeing AI now. Like we know that everybody says they're using AI. there's it whether you're using chat GPT um or personally or using it at work uh using claude even getting as as uh specific as using the claw code or even building agents um but we're what we're seeing and what the the numbers tell us and what um the industry is showing is a lot of people aren't really getting impact from it at this point so spending a lot of time on it instead of saving time um are actually now you're actually investing more time more money into a new system um but not seeing um strategic, measurable, missionbased impact on this. So, it's really coming down to how organizations are working and where they're looking to improve these things. Um so, we think that, you know, we're getting a lot a lot more um in like people are more interested more recently and hey, I've I've been using AI now for the last year, last two years. I've tried these things, but I'm not getting anything out of it. I'm everybody in my company is using AI, but I'm not really able to show results from that. So we're seeing that that things move that way. One one big thing that we've seen um kind of transition from techn just technology applications before and this has always been a concern not just in the nonprofit space but across the board is you know people are using their own tools when someone leaves a company they're taking that with them. They're taking that knowledge with them. That's happening now um even more with AI and and one of the things that we can we think will um help move you all forward is is being able to address that first. So, you know, when you're thinking about your organization, you might have a program lead, a development staffer, and a comms coordinator. They all have their personal accounts, their AI accounts. Um, that they're setting up for the things that matter to them, but they're not looking at how that matters to the rest of their organization. They're not looking at how that can help them scale and then how it can get them out of the things that are are more remedial and what they what we think that needs to move towards. And we're seeing um positive outcomes from is sharing those across an organization, documenting how what they're doing, documenting what the what um those tools are doing. Um documenting what those processes are and then able to train more processes and add um only the relevant tools in there. So, we think that the next quarter for, you know, end of Q3 and going into Q4, one of the cheapest wins that we will see and you all will see with AI is just writing down your top three priorities, your three best setups, how they're working for you, um, and how they belong to the team instead of just, you know, the one person that was that discovered them and started using them. So, you not just kind of gatekeeping what's what's working. The third pattern we're seeing is leaders, and this kind of builds off that second one, is is leaders aren't seeing how AI is actually used. So there's a lot of shadow sort of shadow AI sort of shadow technology being used in different organizations. Um, and the the executive level um is expecting to see big changes from AI, but they're not able to really glean where that's being used, right? So what they're expecting to come from it isn't actually what's happening from it. Um and then when they're going back to those teams and they're asking them like, you know, where is this being used? What are the outcomes? They're not really seeing those things mess mesh up. So we um saw 47% of nonprofits, they don't have an AI policy in place, which means that that usage nobody can see is also the usage nobody has agreed to. So getting started with setting the foundation in your organization um identifying where where AI can be used and what that policy looks like is a is another good first step forward. So taking a quick you know self p like pulse test self test is you know can you actually name all the AI tools your team used today or yesterday um whose accounts they're running on and if not that's a great place to start. Um so we know just anecdotally um from our experience working with organizations when we do these audits what we'll typically find is you know show of hands like who's using chat GPT who's using claude who's using co-pilot and 100% of the time we find someone that's using something that somebody else in that room is like we didn't approve that or you know we didn't know that you were using that or do we have a paid account for that. Um so it's a very frequent uh piece. So quick poll if you can just share this uh in the chat just answer one of the letters next to it to this where is your organization on our AI maturity ladder. I think this is out of order would go next step but where do you feel that you're that you are using AI currently and it can be all of these things as well. Um A you're using chat so like the just common chat bots to to ask questions of AI. B you're using it to generate content. C it's connecting to your different systems. D, it's configured across the organization to do something. And then finally, E, it's delegated. So, you're actually getting agents that are making decisions based on your data and working for your organization. Got a lot of A's, B's, and C's. A through D. That's good. There's no wrong answer here. E, so delegated. Cool. So, a lot of you answered ABC. Um, that's kind of what we expected to see. So, most most organizations are getting to B, which is two on this. So, the first one's chat questions. Like I said, something you're typing this into a blank window. You're asking questions. You know, how can I what can I have for dinner? How do I lose weight? How do I fix my car? What can I do for my organization? You know, what are other organizations like us doing? Just getting those answers back. So, basically using it as the new version of Google. um but b trained on more relevant data and information. B is content. So you're drafting content, rewriting documents, using it to organize documents for you. Most organizations are in that first phase. We're seeing organizations start to to lean into um the connected phase. So that's where like AI is looking at your email. It's it's viewing your calendar. Um it's looking at files on your computer or files in the cloud with your organization. And it's maybe even tied into your CRM. So it can start to take action. This is where we start to move it more towards the automation phase four or in the other one in the poll D um AI is being configured so you're able to save projects and work on those projects over time build across those. So like a cloud project or a chat GPT project or something more um more organized locally. Um and then you're able to work with other people or other in your organization or outside your organization through things like skills that are repeated um shared setup. so that you know that the things that that what you're seeing is the same thing that they're seeing and you all are engaging with the same information. Uh and then lastly the the top end of of right now of AI maturity would be having that delegated. So things that are running on automations, a agents that are running uh across your your organization and making decisions um based on your data so that you don't have to be in there and then really there's just a person there that's kind of reviewing these things. So steps one to four are very much more managed or micromanaged. Step five is when you're able to sort of almost outsource um pieces of your organization to AI and that's where we're getting towards the the higher end of AI maturity. So what knowing that what we find that's that's typically missing and in the in the audits that we're doing in the implementations we're doing and even in the tools that we're building um we see people using the tools so they're doing the chats they're you know beefing up their own productivity they're creating content for social media um and they're using you know the systems that they've used every day. We're not asking people to switch systems and migrate off of the platforms that they're that all their data and things like that are in there. But what's missing is this intelligence layer. Some people you'll hear it called the brain. Um some people call it like a context layer, but it's really that shared knowledge and memory system across those different connected systems where your agents can tie into where those projects are tied into where things like the skills are are are hosted or managed. Um and that's typically what's missing. you know, you might have SOPs written down in a Google Drive somewhere or a Dropbox somewhere. You might have um playbooks or like knowledgebased articles in other places, but it's outdated. It's not consolidated. It's not in one common layer that's that's built for um these tools to connect to it. Um and that's what's that's where we're seeing the missing piece. So, this isn't something that just happens, you know, with uh in one swipe. It's not the kind of thing that just can be done overnight or in a week or something like that. This is built over time gradually. This is something that you would go through or you have already gone through. Um but at the end of it or you know this constantly building thing uh your organization gets smarter right and then that the tools you tie into your into that get smarter and can make more decisions and that gets you closer to that level of of AI maturity. Um so what you end up with from this is um we're what we're seeing sorry the next pattern that we're seeing is um you know a year into using AI um work is happening faster but we're not seeing anything changing. So we're seeing people put in I sometimes call them like interns. They're putting in like bots that are doing things um and doing the busy work and turning things out but it's not changing the outcome. So they're not you know they're not getting more donors. are just maybe mining through donor data faster and they're not um serving the community more. They're not out there, you know, doing more in that community. They're just able to cut out some of the remedial tasks, but they're not really a not really able to see the outcome come grow. Um so we're seeing maybe costs come down for like if you're paying people hourly or if you have them internally, but we're not really seeing outcomes change here. So just being able to move quickly or do things more quickly um it's not going to move the needle on the outcomes and that's the the approach we want to take is what do we want how do we want this organization to grow what do we want to change in this organization not just how do we you know decrease the time to to do things um what we are able to measure is those is that time savings however so when we do think about what that outcome could be what that means for your organization. We do still measure these things. We're a thing we're able to measure as a result of that is hours. So, um if you look at how long something takes, let's say it's 11 hours a month for one person that turns data into ex like data export into a board report and they're spending a bunch of their time doing that and they're consolidating things from all these different systems. Um that turns into 132 hours a year. So like if we can make that if we can if we can fasttrack that with AI or automation um we can save their time um so that you can get them back to doing those other things and then that leads into that larger goal. So how do we get this this better outcome? >> Thanks Kyle. Yeah. So after looking at all these different patterns, we we want to really explore what what's working now, what's hype, and what's coming down down the pike that we could really uh afford and and and drive some impact. So the first the first thing that we we are seeing as far as trends go is the the tool debate is over. So when AI first came out, there was a ton of different tools that could do managing your grants, managing fundraising, you know, all all types of different tools. And what we're seeing is, you know, people were spending a lot of money on tools for slide decks, transcriptions, summarizing reports, chat bots on the front of your homepage. And now the the platforms the you know the frontier platforms like claude, Google workspace, Microsoft and even HubSpot absorb most of those tools. So you can use those tools and build agents on those platforms and that's saving folks a lot of time and money and and you're able to integrate a lot of these systems and and save that that extra step. So where that really moves the needle is, you know, project plans, keeping stuff current in a sheet without a whole project platform. You you can do that now. You can run fundraising agents over your CRM if you're using a tool like like HubSpot. Um even even reporting if if you're using Clawude to pull in all your data like Kyle mentioned as a second brain, that same data can be used for reporting and publishing it for the board. It could tie into HubSpot as your CRM, but generally the amount of different tools out there instead of having, you know, a dozen of them now, you might only need three or four and that operational intelligence layer to cut to connect them. But most of that uh intelligence lying within Claude and and Chat GPT and and others. Then the next slide. So the first thing that you really need to look at now is is the workflows not not departments. What workflows within your organization can you do two things as mentioned save time but if you're saving time and not driving impact then that's not really going to have much effect on your on your bottom line and ROI. So really fixing the most important workflows in your organization. In this case looking at a board report. So you might have your source data for that report, but then the bottleneck in number two here, a person hand turns one thing into another. So that could be a spreadsheet or taking a spreadsheet into a word document or pulling data out of a system and looking what what part of that workflow is there a bottleneck and where can AI or automation at the very least fix that. So in this case, you know, you can export your data, use AI to grab case notes, put it in a structured record, you could schedule a PDF and assign that to be delivered, and then you have one source of truth where you can pull all this information. Um, and then the human part comes in the the last two pieces at the top here where it says re reviewed and delivered. So once that automation's in place, you know, human in the loop, as they say, comes in, you could review that and then send it out, um, send the report out, saving, as Kyle mentioned, three, it could be three weeks out of the year per per report. But really take a look at the workflows. Fix a workflow within a department and then how can that workflow is there any similarity between that workflow and others and start to map out where to really make make a difference and and start to you know increase your your AI maturity across the board. And then verdict three, sorry, we'll get there back up one more slide. Yeah. So that was verdict two. Next slide. uh put putting this in practice. So let's look at some some different workflows. Marketing is where we're seeing some of the greatest impact with with AI right now. Um to to do marketing, let's say the awareness phase, that could be billboards, banner ads, anything to get get your name out there. uh but just creating that content and using AI within that content and looking at the different steps just to to develop it. And then once you drive someone to your website, of course, you don't you don't want it just to be a brochure website. You want to capture their information. You want to find out what program they're interested in or do they want to donate and then personalize that messaging to that person. And if you had to do all these steps by hand, it would take forever. It would be way too costly. And by the time someone entered one of your programs or made a donation, the the cost per acquisition, per se, would would just be too much. But with a platform like a HubSpot or or any type of CRM, you'll be able to look at that workflow and then be begin to automate all those steps on the back end. And it's very u intuitive. So using one of those systems you can do that andor if you don't have a platform like HubSpot you can look at your constant contact you could look at meta your meta business manager for social media and then look how you can connect those and put those AI like triggers in in there but like we mentioned before a lot of platforms now have all those systems in place and you could do it on one platform but at the end of the day not only will you save time but you employee will be able to drive impact through your marketing and communication. So, it's not just awareness, you're driving people into programs, driving behavior change, and really making a difference. And then taking that data out of your platform, showing it to funders, and that that flywheel continues to turn because you'll then get more funding. So, just one example with marketing that that we're seeing. Next slide, please. And then once once you get someone and you attract them and let's say you know you you you someone's enrolling into a program you're getting them housing for example. So a lot of nonprofits then they have to manage the people that they're serving a whole case management. And if you break down a case management system there's so many steps so many platforms so much of it is in spreadsheets. And what we're seeing now and and we're doing is being able to leverage one of these tools and put all these different workflows in place so you can really drive optimization. So these things all begin to compound the time saved and impact from marketing. You get someone into a system, getting them through that system efficiently, that's a whole other workflow that that you can look at. So taking a look at your case management, where are those bottlenecks and writing that down and then looking at the different systems and automations to really drive that piece of of your uh service delivery. Next slide. And then finally, fundraising. You know, this could be grants. This could be reaching out to individual donors and or partners. And what what part of this can we automate? what part really still needs to be personalized and humanized. So to identify folks that could be fully automated, we you could sync in different a APIs, Zoom info, uh Candid, others, but really being able to grab grab the target audience you want to meet that can be fully automated. and then research that target. Write a brief write what they would be interested in. Compare it to the service offerings that you have and be able to write a a really great donor briefing and being able to automate that and get that out to the right person. So, when you do meet someone, let's say they come to the website, they're interested, you know, they hit the website, you get a text message, it says this person was just on the website. you can recognize that this is a large corporation that's ahead of their corporate social responsibility and there's the phone number. Boom. I I can call them and make that ask. And then the thank you and stewardship relationship building that all becomes, you know, personalized. You're doing that yourself, but it's all being tracked. So, as you're talking to these folks, that goes back in the CRM and that enables you to continue to have that real-time intelligent conversation with folks. And then, of course, all the reporting, all the work that you're doing across your grants, your personalized fundraising uh and development, all that can be pulled from AI, whether you're using Microsoft 365, HubSpot, others, all that's connected to your CRM. And then you could report to the board, not just on one donor, but you could segment it across different donors, different segments. So, you know, two or three years ago, all these blocks would have been the dark blue, but now with AI, you'll be able to do, you know, three out of five right now with with AI. And those ones that are in dark blue, you'll have the intelligence at your fingertips to improve your conversion rates when you're supply, you know, when you're responding to grants or speaking uh to a a donor. >> Great. So, the next verdict is um how we're seeing this in practice. Um we were we sponsored a hackathon last week um which was a three-day event. who partnered with the University of Delaware and the state of Delaware and they and they took um different divisions of the state government and we were able to see them put together um to ship some examples and I'll give you different some of the different examples that they they went after, but the one that that stood out the most was a permitting process. and they were able to ship a working version of a replacement to a permitting process that takes weeks and months now manually even with automation in place um to get permits through in days less you know less than a week um with very little or no human interaction. So the way that government the government permitting was working before they would have an application um somebody would fill out an application they would match it to like one of a few different branches and say okay if it's complete it does this it goes to this next thing um but everything else would wait for a human and you know humans are where the stuff would get stuck because they can only look at so many permits they can only review so many things they have to go back and ask all these questions and of course you can automate pieces of that but they didn't have a real intelligence layer in place um and they didn't build a model for this. This the uh this group of students had built um built it into built AI into it. So you maintain the application. The application doesn't really change. It would simp it would shorten the application if you already had applied before if nothing was changing. It was just like a new permit that needed to be updated. Um but it reads across all these different sources that they helped consolidate and put together and build a model from. Um and then it would use essentially agents to make decisions based on what they need. It would find other re, you know, related permits. It would um review local jurisdiction and and requirements and it so it would no longer need a person to look at those things. So if I'm apply, it was a sewage permit a sewage permitting. Um, so if I was in a certain jurisdiction and I knew it needed these criteria and I knew what things it couldn't have, it could evaluate a lot of that against this um against their application and quickly spit out, you know, everything it needed from the applicant or, you know, a rejection for the different reasons and get back to them. Uh, and then very minimal person interaction would really only go um if there were questions or things that needed to be open. Um but what was fascinating about this is they had you know they had people judge the competition. They had the governor there. They had other people from the state. Um they had part of this AI um AI group that we were that we're part of um all judging this and and there were some great applications that other the other people built but this was the only one that really moved the needle uh and was able to provide outcomes for that organization. So going back to what we talked to earlier. Um so it didn't have anything to do with picking a platform. It all had to it all started as we said earlier with you know finding a process identifying something that people are doing by hand where there's a bottleneck um and things that can be you know if you think of them automated but but needed uh you need to add more um I guess degrees or more complexity to that automation and it needs to be able to make decisions. Um that's what's able to move move this stuff forward pretty well. So, um, when we're doing these when we're doing these consults, when we're talking to people and our primary recommendation first is is if we're going to look at spending time and money and resources on AI, we should be able to quickly with like a, you know, sort of a mental model, um, look at what you're doing now and show it's worth funding within minutes, right? So first look at what you are pulling, you know, what what you're going trying to add AI to or trying to use AI for. So like exported material, your case notes, um like a schedule that you're following, that permitting process. Um it can't be like just some pre-built demo. So make sure anybody who's telling you that you can do this is looking at your actual data, not just giving you a hypothetical like, oh, we've done this for this this company before. um make sure that they can kind of walk through that with you or you can do this on your own um and say, you know, here's where your team is spending this time. Here's where there are bottlenecks. Um here's where there are ways that we can sort of remove these things. This doesn't take again, same thing. It doesn't take weeks to go through and audit through these things if if we already have these processes defined out and the people that you're asking to um replace those can do that. That was like what was nice about this hackathon. Each team was given, you know, a a set of criteria of things that need that could be improved. They were able to pick from those and they already had this stuff. They knew what it was. They could they could come up with like a a solution, so to speak, um quickly. The outcome, the artifact should be yours to keep. So, you know, if that's a blueprint, if it's a, you know, a how-to, a guide, or something like that, that should all all be something that that you can keep and you can take away from this. Um, and then you should always make sure that the person who's telling you that this can be done or demonstrating this to you also is telling you what it can't do. So they can say, "Hey, you know, it can do A, B, and C, but it can't do X, Y, and Z, and here's the reasons why, or here's what would need to happen for that to come into place." Um, and when you're doing this evaluation, we, you know, if any one of these things doesn't happen, I would just, you know, we would just recommend like walking away, basically. Um, we do this, you know, we do this for everybody that we engage with. you just recommend every any vendor or consultant you're going to talk to um you can hold them and keep them honest to this this as well because if if it doesn't pass these like four tests, it's probably not going to actually work. Um it might just be like a hey, we think we can solve this problem. Let's pay us to figure it out. Um two of those things that we think, you know, I think were shiny objects before that are um probably worth, you know, pulling funding from now, um website chat bots. Chatbots have been around forever. you could train chat bots well before AI was mainstream on on a lot of your content. Um, this is one of the most requested things you built and you typically one of the most the least valuable things for a number of reasons. Um, if it can answer questions, that's helpful. If it can route questions to you the right way, that's helpful. But unless you're really like a services type servicesbased company, this gets harder and harder to really prove value. Um, a lot of times it will just be a place where people start asking questions of it that aren't really important to the impact that your organization might have. Um, and it might be on a on a place that you only have a limited time to get someone's attention or to ask for something or to to to drive value to someone. And if you let if you leave that off to a chatbot when you have people that can drive impact, um, you lose that um, you know, human touch basically. Um, the other one is like having an agent for everything. We see a lot of we run into a lot of cases where people like I've created an agent for every part of my day and every part of my job. Um we there should only really be agents. When you think about an agent, agents are at the sort of the end of that maturity ladder. Um they only should really belong where you where decisions are being made and and you need to provide there needs to be some discretion. So an agent kind of pairs with a human. um still keeping a human in in a loop, but that doesn't usually that doesn't usually mean there's like 20 different agents for an organization or even for a single job function. It might be one or two. So like if your function if your job function is um you know development, you might have a development agent that helps you with a package set of things, but you wouldn't have you know one for fundraising and one for reporting and one for all these other likely wouldn't have one for all these other things. Um, so if you do get a vendor that comes to you and says shows you an agent in every box and shows you your whole team with agents alongside them, um, be weary of that. Um, and make sure that it's doing the things that really move the needle as you said, like it's actually the that impact. Um, a lot of that is to be said that especially in nonprofit organizations, it's being able to keep your voice human. So you know when an agent's making decisions when a chatbot is answering questions a lot of times the voice gets lost especially as new models get added to those different things. Um so you know machine can certainly do do the remedial things we said can fill in some of those like speed and time and impact gap or time gaps but like you know reading things um tagging and routing things exportingly based on rules summarizing and structuring you know new things certainly creating presentations and taking you know your existing brand and applying it to web pages or content or whatever it might be. Um, but nothing that's going to really change your public voice or what goes out there. So even even things like what letting it create an auto post social media content and things like that for you would run the risk of, you know, could possibly tarnish your brand or change how someone um engages with you. So we always want to keep thing keep the the big things human. So donor appeals, you know, it can draft this stuff, but making sure there's a person that's reviewing that and saying, "Hey, this sounds like us. This doesn't sound like just more AI slop. Um, any advocacy copy, your mission statement. Please don't have AI create your mission and your values and things for you because then they're AI's values and mission, not really yours. Um, and really anything a supporter you ever expect a supporter to read as you or as your organization. So, if you want them to feel the cause of your organization, that typically doesn't shouldn't come from from AI. And as we all engage and see AI more and more day-to-day, people are getting much better at spotting what is AI. I get I get daily I get RFP respon requests and responses and even filling out forms on our website and asking us questions or we'll go and do an audit with someone. They'll say, "Well, I already did this." And they'll send me like a full AI audit like great, you didn't really do anything. You just asked it the question you would ask a person and had it give you an AI answer. And um I would say 9.9 times out of 10 you can find out where there's gaps or holes or where something hallucinated there. Um being sure that AI is running on your data um and getting your data right from the beginning. And what we've we've noticed and what nonprofits have told us is you know a lot of nonprofit data isn't really ready for that. So we find like there's lots of duplicate records across different systems and they haven't had time to consolidate and merge those. Um you'll see things where emails are the same or spelled differently, funders names are spelled differently in different places and that the system is saying that's a different person, but it really is all the same. Um so that can cause data issues obviously um very frequently, especially smaller nonprofits or startup nonprofits, those are spread thin. Um spreadsheets are the real system of truth for them and they're not kept up to date over time. New spreadsheets are created as new things are needed uh and they don't go backwards. And then there's um like something like the source of truth authoritative document something like that lives on one person's computer. It's in one person's inbox or their own Dropbox. So the first step is really you know cleaning that set up and making sure that there's like a a actual record of truth, a source of truth. Um getting field names cleaned up. So you know what's unique identifiers? What are we saying is how do we talk about people? How do we talk to people? have one place where all these things live and then keep that as that source and then that help you start to build out that intelligence layer. Um and what that'll give you at the end is you know as you bring new people on people in new AI agents in new AI platforms in you can always tie it to that one place say okay our source of of information on emails are and and donors is here and this is the this is how we refer to them and this is you know what we know about them and our source of you know how our organization works is here and the history of our organization is there. um that all becomes that place where you can have these different tools and systems um tied in. We know no matter what the cases that we've gone through so far that data cleanup that um organizational layer is the has the highest return on AI and AI spend for um anybody we've worked with and all the or a lot of the organizations we talked to and it's really the one that pretty large list and it's typically the one that people are like not not ready to fund. they're happy to go spend, you know, $15,000, $30,000 on a new piece of software for the year, but not having the data right to begin with. And then that $30,000 software um license has another $15,000 implementation cost and by the time it gets launched, the data is out of place or it's not not um not accurate and then it's just very much junk. Um so we've dumped a lot of information on you. So next we're going to tell you hopefully point like where to focus um and then what we're seeing successful. Now this is just kind of a practical approach. Nothing here requires any type of budget approval. It's just some next steps we think you can take to to move further down this uh adoption cycle. >> Great. Thanks Kyle. So yeah, when we work with nonprofits, what what did the 7% do do differently? What are they doing right? what gets them to really catapult themselves above above the crowd and and the competition. And the first part really is slow down first. Um it's it's not about looking at all the tools. Saw Debbie put uh a question in in the box and we'll we'll send this out, but what what are the AI guidelines? What is what is your mission? What are your goals? And and having these guidelines and guard rails written down. And then to Kyle's point is the first step really is is the data. And if the data is clean, the data is integrated, um, then you can begin looking at at the AI piece of all this. And then the first step really is picking what's the most costly recurring work. Every every nonprofit we work with, we'll do an audit. We'll look at finance. We'll look at development, fundraising, operations, HR, all the different departments, case management, and look at each one of those and which one what are the workflows within each one that there's a large bottleneck. What could be really what could have the the largest impact if we solve that that one piece? And that's really where where to start. If you try to do this across your whole organization at one time, it's it's just not going to work and things will will break. Um, and then document what works. So, if you are fixing a workflow within your marketing and you're automating your content creation, it's researching, it's publishing, people are coming to the website, you're seeing conversion rates, and you could document that. Then you could take it to your other campaigns and begin to replicate those best practices and and use the tools that you you did to connect those pieces that's already built and you could place it into something else. And then empower a small group. Put together an AI team, have someone lead that team and give them actual goals, rocks, you know, deliverables and KPIs and a timeline. So they can begin to to implement this and do it crossf functionally as well. So you know if you're doing some type of AI and automation within one department it's going to affect others. So they need to understand how that's going to impact them and let them learn from you but also you know share best practices. So you could do this across the whole the whole organization. Next slide. Thanks. So, like we said, pick the workflow. Um, name the number. What What do you want to move as far as are you trying to save hours? How many hours are you trying to save? If it's conversion rates on you know your development is uh your you know from from development from training impact case management what's the number what do you want to elevate and then what's the amount of time you put into this so you can measure your ROI and and fix the worst step the place that's really the most manual. If you look at uh the maturity index of of a nonprofit, we look at each department, you know, and and you go from manual to you have a written down process, things begin to get automated. Then you layer on the AI agents and just getting down the process and automating pieces and you know automation has been around for years. You will see a tremendous tremendous impact. Um, you know, everyone's saying AI, AI, AI, but just getting folks to that automation step would be transformational across any organization. And then and then and then you take a look at the agents and building that um that brain per se to to continue to improve. But that's really the best way to start. Pick a project and try to have it complete in 90 days. With AI, it's not too difficult to get something whipped up in a day, a week, 30 days, and with the right support, you'll be able to launch it and and measure it and really really lower the runway to get something launched. So, start with your processes, look at how you can automate them, and then what type of agents can you layer on top of that. >> Cool. Um so we wrap this up a little bit. Um we'll take some questions after this. One of the as underlying messages you've heard is getting all of your data and your standards into one place. Um we do this this is what we do at scale for large organizations. So we'll do this you know one off. We'll work with their systems. We'll go out and build these intelligence layers for them. what we've done at scale for for nonprofits and um you know for a number of our clients so far um we built a system called Tap HQ um where it's one place with all of your standards. It's an easy way to connect your systems, get this stuff into one place. It's going to be your shared knowledge and memory place, connect to the systems you already own and use and it also allows you to connect out to those systems. So if you want to have an MCP that ties into, you know, all of your data and pulls that out in something like Claude or Chat GPT, you can do that as well. Um we have it broken down across different um we call them HQs. So there's like finance and there's growth, there's grants, there's content, there's support. Um and what that becomes possible from that is you know having agents. So we have an agent builder in there as well as agents that are pre-built in that system that run across your data and answer only from your own information and facts. Um it's got reporting built in. So, as we talked about consolidated reporting for like boards and things like that, all that data being in one consolidated place makes board reporting simple. There's also board portals as part of this. Um, and we it's built to make sure that your organization or your staff can be productive almost immediately in days and not in months. It doesn't take some large, you know, um, implementation to to do that like you would have with, you know, CRM or something like that. We're not asking you to replace those things. Um, we'll share a link to this after this as well. Um, we have our clients on this as a as a client of TAP. You get access to this by default. Um, but we're launching this publicly in September. And if we have a little bit of time after we take questions, I can kind of walk through this. But the goal of this is to make is to be an operational intelligence platform for um nonprofits and and missiondriven organizations. So, it's geared towards the things and the the sort of the departments that you have and the the the work you do every day to really see those outcomes and drive those outcomes forward. So, tracking your grants, building out a grants pipeline, making sure that your team is involved in that grant process, building out your brand, consolidating your brand. Um, and this all runs with what we call a super agent um called Tab IQ, which is your that organizational access that organizational brain that you can ask questions of this anytime. So an executive director or an intern can come here and ask any question across the data that's connected to this system uh and get to learn and get to know more information about that organization. It can also then generate outcomes based on that as well. So I could say hey how many employees do we have right now? Are we on track um to deliver this grant? You know who do we need to bring into this? What's missing? And it can come back with all the recommendations and changes specific to you, specific to that that that market. um or that tool. Um do you want help on this? You know, we've talked through a lot of things you can do yourself, but obviously a lot of things we can help with as well. So there's kind of two things you know we can run the gamut as far as how we work together but some basic ways to get started would be either option ones either we do like an AI capacity consult so this is like a working session sort of that we talk through today um or we better understand like your bottlenecks um what start to pick out what to pilot first we take like this is like a good first step um we look at your tech stack uh and it's really focused on automation uh and then moving you up that ladder towards towards you know full full AI maturity. Um the other option would be us doing a full capacity diagnostic. So this is a full AI diagnostic. This is a paid um option here that's actually in the catalog. Um we do it's four it's a series of four meetings and there's work that happens obviously in between. So there's a quick intake meeting to just understand who and what. Um and then we do a recorded session that we'll share with you all kind of based on like what we think outcomes. So like kind of checking those boxes we showed you earlier on those steps. We'll work with you on a um a road map. We'll come back and present that road map to you and and make sure this makes sense. This is the right path forward. And then ultimately we come back to you with a scope of what this would look like um to be built. Whether that's us building it or you taking it on on your own. Um we do the intake 5 days typically before the session. So we try to consolidate this so it's not drug out over a long period of time. Um so you would this would be you know bringing your stakeholders to these conversations making sure you have um their attention and their engagement so that this can get done um in a consolidated format. Um either way you know you have this you know these options for us. We'll share this with you after the call. You can reach out to us through the website. Um our email addresses are right here as well. Uh, and then I'm going to close this with any questions and if we get through those, I can give you a quick demo on on Tap HQ. We'll also share a link to that. You can sign up for the the beta so you get access to that when it launches. Um, it's free to sign up for that. And then there's obviously levels that you can can grow with. So, I'm going to start in the QA thing. Um, someone asked, uh, what kind of security settings organizationwide that need to be in place before allowing AI to be connected, configured, and delegated for your organization? Um, that's a question I could spend days answering. Um, that is very goes back to like very organizational specific. There's obvious I would say some sort of obvious thing. So, um, especially when you're talking about an AI tool like Claude or Casshatbt, making sure that you're using the paid version, you have the the data retention and pieces like that toggled so that you aren't sharing your data with the whole model. Um, and putting that data back out into almost like public and it's not being trained on that, so to speak. Um, not sharing personal information in there. Um and then making sure that a lot of times one of the other things we'll see a lot of people sh will share login to those accounts. So there's a lot of there's you know opportunity. So if I if I have a login if Joe and I both share a login and something gets compromised on Joe's end Joe can login as me or Joe can log in with my account and you know you can run that level of security risk. Being wary of what systems you're connecting to those different systems and what information is shared as a result of that. So, you know, if you connect your Google Drive to to Claude and you have PII or something like that in Google Drive, Claude has access to that. Now, you've exposed all that information to one place. And that's again why it's so important to have like the consolidated sort of intelligence layer and and the security protocols and things like that in place. So, that is organization specific, but um there are some common things that can be done. Um, someone asked if we can talk more about the intelligence layer that is usually missing, what it looks like, is it an organized folder with reference docs that's connected to AI. That's a great start. I mean, it's it's really source of truth, coming up with the source of truth and making sure that that's that's what it's being trained on. There are levels or examples of um, you know, how to evolve that out over time. Um one item is one item worth mentioning um is that that uh it's depending on what model is reading from that or what tool is reading from that um it might only be able to go one one or two layers deep. So just because you have a folder you stuff in a folder doesn't mean it's going to read from that. There's also risks and I've seen these crazy studies before where something had like a million pieces of data in one drive, but they they intentionally put two like sort of counter pieces of data in that drive to see what the model would find. Um, and it will find the outliers um inevitably. So, it's just making sure that that's kept up to date and that's organized well. And you have sort of you have I'll call it like a table of contents for the different models to read from so they know what to go where to go for what answers to which things. Um so it's not just guessing basically because it will just guess. Um so it's more than that but it's a great start. Um with a deal of lot with a we deal with a lot of professional identify information avoid getting to the intelligence layer because we don't understand if this information will remain confidential. Any suggestions on how to proceed? So yeah just back that. Like that's why it is important to like think like I said not just connect not just connecting claw to Google drive or copilot to to one drive or whatever and having a an intentional layer that um is what you train basically. It's not exactly what happens but what you're basically training and telling that that agent or the model to make decisions based on when it comes to the data you're sharing that's should stay out of these systems. if it's going to go into a system or if you're going to add that to it, um you would need some sort of either offiscation layer, so a way to deidentify the data between the two systems. So if you store social security numbers and personal identifying information in a system that never should be exposed um to like another AI platform or system unless that system is HIPACO compliant and you you know you have a BAA in place and you understand that um where that may or may not be used or trained on you would have another layer in between that could be like just relevant to that. So like you could just give you a list of all the people that h of this age that have donated this much or whatever. Um and it has a a separate unique identifier and then in the middle would be another system that would ident would do the deidentification that would never be accessible right um lock that down so that you don't have that information in that system. Um, someone asked if we had any examples of AI policies or templates. So that tap on tap HQ there actually is an AI policy generator um that will be available um you have to there's not like a one size really a one-size fit. So all it has to be sort of consolid or built for you um and answer some questions and then consolidate what your organization does and needs into a policy that that fits that organization. Someone said, "Our not profit, our non forprofit, we're experiencing fantastic growth and have no systems. We suffer all the issues you mentioned. Any advice for balancing or prioritizing the core work while keeping the operations running full of events, publications, etc." So, just going back to some of those steps that we mentioned um you know, identifying the highest value things, the things that are taking up the the chunk of your time that are repeatable. Um that can be sort of like if you think of it like created an SOP for it and you could document that process. um and it has decisions in there. Like if you can make it automated, that's great. That, you know, that takes the need for really AI out of that. But if it does need to make decisions based on, you know, steps before and steps after and it's not just a simple branch like, you know, someone sends us an email, respond to an email, someone sends us a letter, respond with a letter, like if it's more more sophisticated than that, if it's based on like the content of that email or, you know, who it comes from or something like that, then you could put, you know, build AI into that. If that's a common thing that you are addressing or you're dealing with, document it. Document the process itself. Identify what success looks like and then build build to that. How capable is Zephy automating CRM? That's not something I'm super familiar with, but I could you could if you want to reach out to us. I'm sure someone on our team has the experience there. Um, is this only available in the US? No. Um, Tech Soup serves hundreds of countries. We provide services across those countries as well. Um, with Tech Soup. So, this and even like Tap HQ and stuff is available in those those countries. Uh, and somebody just mentioned that this is overwhelming. We agree. Um, that's why that's why, you know, as we point back to some like getting to the basics is is like getting back to basics is where you're going to find you'll you'll see the most success. Um, uh, someone asked, "Is it a no-brainer to have your organization AI work on a local machine that stores the data? I don't know if we should be cloud only or local in cloud." I wouldn't say it's a no-brainer, but it's that's certainly I'll say it's safer. Um, it also, you know, but with that safety also comes scalability concerns and access concerns. Um, and again, it depends on what that data that information is. Keeping the data separate is helpful, um, as I was saying. So making sure that the the the intelligence layer what the what the organiz what the models and things like that can be can have access to and make decisions based on um in one place and then the actual data in a separate place and then it could make use that intelligence layer to make decisions on how to transform this data but it's not that data is not in the intelligence layer things like that will will I think help Um, I think most of this is all I think we got it. Um, as a housing nonprofit, much of our time is spent with clients, fundraising and grant writing. We're still in the low-level use of AI usage based on but in our upcoming meeting next month. I'd like to introduce using a service will help cut our time, increase our marketing, and decreasing our time inputting data and writing. What products do you recommend that will help and don't have a large monthly fee, but are either low cost yearly or small as needed? fee for use. We will likely not need it as a full package for another year. I need something very secure, easily used without for those without a computer science degree and with just a few people as users. Thank you. Our nonprofit is fully not fully funded and Helen destroyed our community. So, we have great need. Appreciate this great info. Um I would go back some of the tool like I'd start with some of the tools you use every day. So, if you use Google, like starting with Gemini, um it's probably if it's in the nonprofit package, you might get it at no additional cost or like a a version of it at no additional cost and that might be enough to do some of the the busy work and things like that um to automate some of those processes. Same thing with like C-pilot. Um if you want to move it up a level like Grock is doing some pretty crazy things, Grock just or so XAI or SpaceX or whatever company it is at this point bought Cursor. Per cursor was is probably the leading um application for like it was coding agents but now it can do so much more. It built you know they were one of the first like agent layers really and agents that could talk to each other. Now there's um what's called grob bot that can tie into a bunch of your tools and you can have you can build whole little like agent teams. So not an agent for everything was talking about before like little agent teams that can do a lot of these like remedial repeating tasks. Um, so like if you were talking about marketing, you create like a little team of marketing bots basically and those could do a thing. So like you know quickly responding to um incoming emails or um if you know people are registering for events, making sure that they're getting put into like a spreadsheet or something like that. If you don't have a full CRM that's pulling that information in. So it can do a lot of that um automation those automation pieces. Um here um someone asked if we could share the dashboard. I know we've got three minutes so it's going to be quick. Um so this is like a highle view of tap HQ. Uh it's got a configurable dashboard. You log in you can put any of the items any of your data and that stuff right on this dashboard. Um and then on the left side well let's go through a few things. First is like you have this tap a tap IQ across the whole system. You can ask tap IQ. So I can ask this you know information. So you know how many employees do we have? I'm in a test account so this always bites me but um this you can see it sends it gives you a chance to unend it and it will like go through your system and it will it will answer the question based on what what's here based on what tool you have connected to it based on what information is in here that type of stuff. Um it's just pulling this up. I'll let this run here. There you go. So, putting people data tap network has 15 employees. This is wrong. Yeah, this is like in a test account. Um, but this will tell you like where this information came from. It'll give you a direct link to where you can find more information about that. Uh, and it also tells you like where things are wrong. So, it says like, you know, I need to reconcile the number of employees versus number of users. Uh, we have like a all the different HQs on the side here and each HQ HQ has different things. One that's probably more relevant to most nonprofits is, you know, grants and and grant tracking. So, you can it will allow you to search across different grants are available. So, these are the ones that I'm currently tracking in my system. Um, and I can also pull in external sources. So, it shows me there's 2500 open or 2500 total, 1400 are open. They have a thousand that are expected to open um open or be be funded. I can create I can track grants as I go through here too. So I can just pick a grant and save it or I can add a new grant and then once you have one in here it's got a full grant tracking pipeline. So I can add tasks. I can upload documents. I can set budgets. I can get reports based on the grant once the grant is funded. I can also use this system to report back to my funer to show that where my money is going. Um you can have it generate like drafts and things with AI. Let me just do this real quick. So, if you have AI credits in here, you can bring your own AI application in here. So, you bring your own key or you can use ours. Um, you can have it summarize the grant. You can have it prepare grant guides for you. Um, we have compliance HQ, which tracks like all your credentials. So, nonprofits have a lot of turnover. Systems get turned over quickly. You don't know who had access to what. This helps store that in one place. Um, it does it will do your regulatory updates. It will manage like your state filings. It will manage things like your insurance policies. So, um if you want if you're going to get like an insurance audit based on how many employees you had and payroll and things like that, it stores all that. You can have that information tied into the system. It will package that up for you. And then when that comes up for renewal or whatever, it'll create that package um and let you know that that's ready to go. it if you keep your financial data and things in here, it can also consolidate enough of that to start putting together your like your 990 package for renewals each year over year. Um, and then we also have um other pieces like that are more for your internal team. So, we have like learning HQ. You can use this internally and externally, but it's got a learning management system built in. You can build courses. You can upload videos. It has 11 labs tied in. So it will actually um you know build voiceovers and stuff like that for you all. Um it has agent HQ which is a full agent builder and full agent suite. U content HQ is the marketing side of the house. So setting up your brand library. Um building out content so you can have it generate content packages for you. It can generate documents for you. So this could be like appeal letters. It could be um you know donation requests, it could be marketing things like that. So there's like whole content packages as part of this. Um there's a lot way way more that I can go through in five minutes, but sort of certainly geared towards each function in an organization and crosses those those parts of the house. Um so if you're interested in this, reach out to us or you can just go to the the um the website and sign up and like to taphq.com. What you'll see right now is a coming soon page. Just drop your email address in there and it will alert you when it's live. As I said, this is available right now to our me to our users, our members. So, people that are have a subscription with us or one of our retainer clients or ongoing clients, they're all already in using this. They use this day-to-day. Um, and they're like our our beta or alpha users right now. Um, but we're opening this up for beta, I think September 1st. So, anybody who signs up, you'll get a notice on September 1st, and it'll give you an invite into the system. um you can sign up for free and then there's different tiers uh as you scale up. I know we're a little over time, so thank you all today. Um you will all get a copy of this deck with the links and stuff that we mentioned. Um and again, we're here to we're here to help. Feel free to reach out.