The Future of Community Action: AI, Data & Digital Transformation with Nadeem from Wayne Metro

0:01Hello everyone and welcome to the Fortell podcast. Today I'm joined by one of my favorite people in technology and community action. Um Nadine, thanks so much for being here. >> Oh, thank you for having me here and I apologize for background sounds. I'm in downtown Detroit, so you might hear all the sirens wailing every now and then. >> Well, we can't see any of that and it looks like a very tranquil background, so that's fine. >> Beautiful. [laughter] >> But what I wanted to do is I know we've met on a number of occasions. I loved talking a little bit about you wanting to get into nonprofits and community action. So why don't you just tell me a little bit about your background because it's probably a little atypical I think. >> Sure. Um so coming out of grad school uh I did my PhD in finance uh up in Toronto and then coming out of grad school I joined uh KPMG back then it was called KPMG which transformed to KPMG consulting which transformed to bearing point and went through all these iterations uh and then one of my final clients at KPMG was a bank Lasal Bank and they got me into the world of banking and from

1:04there I spent what 2 odd years in banking but all Throughout my life, back in college days and even while I was working, I was very much active in the community, trying to give back to the community, doing youth mentorship programs, uh various different activities. And I always wanted to do something a little bit more missionoriented, if you will, >> and uh I'd have my day job, which was, you know, fulfilling, but then I'd be doing my evenings and weekend type stuff. And uh this opportunity with Wayne Metro came about and uh this was a nonprofit agency. Uh it's we are an anti-poverty organization based out of Wayne County in uh Michigan and hence our name Wayne Metro. Um we serve uh folks who are underserved. Uh you know we have about 40 50 different types of programs and services. We serve about 100,000 people a year. uh rent support, food support, child care, uh financial counseling, uh roof repair,

2:08weatherization, home repair. There's a whole bunch of different programs and services that we provide. So, when this opportunity came up, it was relatively unique that I could use my background in technology, data, analytics for a nonprofit organization. and I was like, you know what, this sounds like a good fit. So, I gave it a shot and seven years later I'm still here. >> That's great. And I mean to give people a sense of scale. So, you know, there are obviously a thousand community action agencies across the US, but Wayne Metro is one of the biggest. So, it's, you know, it's really quite a sizable operation. and tell me about um the way you came in and maybe what you've changed from a sort of technology perspective because I assume you didn't have a very big budget when you joined and you had to figure out what to do and so what are what are the things you did and what do you see as the future opportunities as well from a tech perspective. >> Sure. So when I came into Wayne Metro, we we are one of the largest community action agencies in the country as you mentioned. When I came in, this was

3:11precoid, we had a budget of about 40 $45 million and we had around 400 staff. Uh COVID hits uh we had all the support money come in. We zoomed up to about $313 million was our peak budget. Now we've come back down to our steady state at a hundred million. So our annual budget year over year is about $100 million. to about 600 staff members. So that gives you size perspective. When I came in, they hired me in as the director of data strategy, which was a newly created position. I had a staff of one, me, myself, and I, and I had a big old budget of zero dollars, right? So that was my starting point. Um, but I came in to a place that's probably very familiar to many nonprofits that are grant- driven that when you take money from granters, your funders dictate many of the systems you use because they want you to use that instance of Salesforce. And so you might have three or four different instances of Salesforce

4:12running or they have X database or Y database. In our case, we we are largely funded through various different government entities. federal, state, county, city, uh different levels of government grants. So with government grants come government databases as well. So the the big picture was your data was spread out all over the place. Uh there was no easy way for you to compile to dduplicate and to get a true picture of where things stood. So that's where I came in. They said, "Hey, can you help us make sense of this data?" Pre-COVID, we were also very much paper driven. It's like, you know, applications were submitted on paper. We had 40 odd different locations, maybe four or five different service centers where people could walk in. We had a call center where people could call in. But in essence, everybody was submitting paper applications. Co, bad as it was, had a silver lining. It forced everybody to go online because

5:15everybody was at home. And so everybody learned how to use Zoom and all the video apps that you can have WebEx etc. Everybody was forced to learn very quickly. Uh everybody was learned to do things online to go online to various websites whether you're ordering food or groceries or going to the doctor's office you're doing everything online. So it forced uh this migration which we took advantage of at Wayne Metro. We said, "Well, you know, we can't be serving people directly." So, and my when I first was brought in, I I I gave my CEO a three-year timeline that, you know, a lot of this change is not going to be overnight. People are resistant to change. It's going to take time to develop the solutions. Give me three years. We'll have things working. But then COVID hits. I'm like, "All right." And I have a budget of zero. So, what do we do? Google forms. Beautiful. So we set up a universal application form using Google forms and Google forms feed into your Google sheets. So initially you have one form with six of our most

6:20uh requested services feeding into one Google sheet. Well, that Google sheet comes in and then staff are working on it and you realize like wait volume of data is really high and you can't just everybody be in Google sheet it crawls and you know sometimes when we personally using it like it's fast well it is if you're the only one using it you know you put 10 people working on the same sheet it starts crawling um so we we split the data what I did was I would take the Google sheet that's my primary intake and pass the data onto six different Google Sheets on the back end. So you have six different program services each utilizing a Google sheet. But even then don't realize it but Google sheets have limits. Maybe personally we never use it. It's you know whatever x million cells and that's it. It hits a limit and the middle of the night you're like hey it's not working. Applications can't come through. Why not? Everything's fine. You find out oh you hit a limit. Also you keep having your thing. You take a goal sheet and you break it in

7:22half. Then you break it half and you break it in half. I mean, you essentially start mimicking some level of a database by using Google sheet hacks on the back end, >> right? >> So as soon as we stabilized out of COVID, we say, well, this is not sustainable. But clearly online works and people want this and by then we had a little bit more financial support. So my team had increased. We moved on to a low code database called Caspio. So we developed our version two of our online platform. We called it gateway. This was the gateway into the way metro world, the gateway into all the programs and services that we offered. And and we developed a whole lot of things and now we're on to version 3.0 of gateway where we're moving away from Caspio and into the AWS world. Um, one of the things as I mentioned earlier was you you have a ton of different databases or systems that you have to work with that you're never going to get away from. And so whether it's statemandated or the feds mandated, you're always going to have a

8:23bunch of systems. So our effort right now is creating this effectively a wrapper solution that sits on top of these mandated systems, but from a client perspective, it's totally transparent. Clients come in to one system, they submit their application to one program. Uh staff process our applications in one system. Then transparently on the back end, we'll feed the required fun required systems so that we're meeting all of the requirements for data and reporting on the back end. That's the >> I got it. That's where we're at right now. >> So talk to me about So we're at this interesting time, right? So we're September 2026 right now. We're sort of three three years into an AI boom and but we're in this interesting period of time where we think okay well we're worried about existential risk as you will have been reading about I'm sure but we're also excited about the possibilities that um in in our industry that it can help lift people out of poverty. So talk to me about let's talk about the opportunities because there's

9:26been plenty of dumerism recently. Um talk to me about the opportunities about where you see AI or conversational AI however you want to think about it for way metro specifically. >> Yeah I mean we are cautiously optimistic about AI use. Um our board even asks about how are we using AI and how we're being careful about the use of AI and the guard rails we're putting around AI. We are very excited in the sense that AI uh can help us process large amounts of data quickly, make connections quickly. It can learn and help respond to queries a lot faster and obviously it's an online system that can work beyond our 8 hour workday. So there are a lot of possibilities here with Fortell. For example, we've been exploring about conversational AI, right? somebody calls in and based on just a conversation, the the phone call, the AI engine is able to go into the back end, pull all the

10:29relevant data, provide as much information as possible up front, and then when you get a more complicated system or you need a little bit more help, you pass it to the human on the back end to support. That's sort of a win-win solution that the client gets good responses quickly. Uh but it relieves a lot of the pressure on on human beings so that we can actually spend the time uh to respond to the more complicated queries and spend time doing case management. Sometimes people think this is about uh reduction in staff. At least at Wayne Metro, we are not trying to reduce staff through the introduction of AI. we're simply redeploying our staff. Uh, one of the challenges we've had is as the volume has increased over the years and it keeps increasing year-over-year. Uh, our ability to do case management, our ability to provide proper services to our clients has decreased, right? Just the volume is just too high. And through the use of AI tools, we feel that a lot of the sort of generic uh queries can be responded to

11:33very quickly, thereby relieving us and giving us some free time to look at folks and look at the complications in their life, complicated cases, etc., and be able to provide those case management wraparound services that really deserve. Plus, next time somebody comes into one of our service centers, you know, if you've already been answered to uh automatically through the AI service, I can actually spend some time learning about your family and what's going on and what the challenges in life to see what else we can help you with. >> Yeah, I agree. I mean, there are a few things to pick up on there. So, in, you know, one of our earliest deployments that you've seen the demo of probably three times by now, which is with MDC and Scott, um, you know, there hasn't been any job loss there. there has, you know, people are doing slightly different things because they're not doing the repetitive task. So, I think that's a good use case. I think intake we're pretty definitive is a good use case. The bit where I'm less clear on quite frankly is we're doing this very interesting pilot in South Carolina where we're having the AI almost like a case manager work alongside a case manager. So, almost like a success

12:36coach. So, case manager meets, they get busy, you know, they've got to do their other things for the next week and then they're going to see the person again who is at risk of homelessness in this case. And then what we're trying to do is have the AI nudge or help people, let's say, create a resume or talk to their landlord or whatever it is that they was agreed and almost think of it like a three-way chat. And I like the idea of it, but I'm still not quite convinced on whether it can work in the real world. So that's why these kind of conversations are useful because you're thinking about this all the time as well. >> Yeah, I I think I think that use case will definitely be very useful. Um, I it's augmenting the work that the person is already doing, but it's picking up on things. We're all human beings. We're going to forget. We're going to miss things. And perhaps you're talking to a client on the phone and they mention five different things of some challenges that they might be facing. And you pick up on two of them and you start focusing and start helping them out. But at some point, you might miss the other three. And so it's very useful to have the software nudges like hey they also

13:39mentioned they need some water support because their bills really high. They also mentioned this and they also mentioned that don't forget this and the AI is going to go on the back and say remind them of XYZ program that you guys have offer and based on the information they've already shared they actually do qualify for this. So it's already doing half the work for you but nudging you to to respond to the client in a in a better fashion. So I think that that augmentation function can actually be one of the best use cases where it's not necessarily replacing the person but it's helping the person respond. That's one. A second good use case is uh at least in some of the demos that we've seen is where a client is calling in and the AI is essentially analyzing the tone and the language being used and say okay this person's really upset >> and clearly they're upset about ABC. uh here's one way you can calm the person down. So you're getting live coaching on how to respond to the person. So you deescalate the situation you and it's not just to deescalate for the sake of that but say all right here's how we can

14:43actually help you one two three things and and depending on the nature of the systems one of the things that we're working on in Michigan is is uh community information exchange where we are able to do warm referrals to other providers. So once that is up and running, you can even say, "Hey, you know what? Wayne Metro, we've run out of funds on program X, but our partner, >> yeah, >> ABC or still has funds available through the CI can get that. I can do a warm referral to you. You don't have to re-enter all your application data. They already know who you are. Are you okay with that?" They say yes. Boom. They get transferred. But the this AI augmentation on the side can help me do all that work. >> Sure. I I think the that's a great example of something we we think of that as utopia, right? And it's sort of working a little bit in Washington state where we work with a few organizations that are in the same area but they do different things and so you know they have their AI Amy and the other one has the AI beer and you know they they sort of talk to each other. Um the problem is you have to get organizations to collaborate, right? And everyone in community action talks about collaboration as a wonderful thing, but

15:44I've seen I've seen not always that to be the case. But I I think this idea that someone can come to you and you know they they're coming to you, let's say they're calling out like, which you know we we talk about or and but there's a reason, right? There's there's there's it's multiffactorial. So if you can try and find other programs that can help them or give them a leg up, I think that's great. The other bit that you mentioned is when someone's calling it's because traditionally it was just transcription. So you were just looking at the transcription and that's quite difficult because you know you could say I'm I'm I'm really struggling to keep living like this. That does that doesn't necessarily mean you're suicidal, right? Um you need to sort of delve a little bit deeper. And the things I'm finding interesting about the AI is a it's getting better at that, but b there's this sense for me that it will when is it good to have the AI and when is it good to have the person. The AI is very calm. You know, it's it's not going to get upset. It's not going to have, you know, get upset because it's the 10th phone call in the day, it's burnt out, all the rest of it. But it doesn't you sometimes don't feel heard or seen, right? So it's working out where the use cases are and trying

16:49to figure out how that works in in frankly a busy organization like yours which has got you know thousands of people trying to get help every day. >> I think to me these are step-by-step developments. So initially you have this cold monotonic AI response but I expect in a couple of versions from now you you'll find AI responses that at some level are empathetic. Uh I if if today already the AI is smart enough to be able to recognize the tone of your voice and say you're upset or you're happy or you're angry based on that then why can't we have the AI respond in a light tone instead of having a default tone to respond in. >> And so you you'll instead of having a monotonic response you you you get a little bit of a response that human like that says oh this person really gets me. Oh yeah. Hi. This is really cool. I mean, I I actually had that experience. I was calling in to a restaurant and um uh to a to to a restaurant to to to make

17:52a reservation and I thought I was talking to a person, >> right? [snorts] >> And then I realized because the response was not just monotonic. It actually had some tones etc. >> And then after a couple of questions, I realized I'm stuck in a loop. is a >> yeah I was like >> it's not just a very repetitive person asking the question so it kind of gave itself away but it had at least initially it yeah a pretty good response >> I agree let me ask you a different question then so if you're sitting down with an executive director who doesn't um have a nadem and a technical background and they don't have an IT person they're a bit smaller than weight metro what would be your your advice what would you say these are the types of things you can do with minimal effort that could actually give a big outcome. >> I think the the the first thing I would I I tell people is you have to have a mindset of wanting to improve a process. >> A lot of times we go get tools and we focus on tools and solutions, but our mindset is still stuck in a different way of doing things. And so the tools

18:56might be great, uh, the solutions might be awesome, but in the back of your mind, you're like, you know, it would really be good to have them submit a paper application because, you know, it's safe, it'll sit in my drawer, no mistakes. >> If your mindset is along those lines, no matter what tool comes along, it's be very difficult to push it forward. So the first thing to me with anybody is, do you want to improve the process? Do you want to help more people with the dollars that you have? If that's the case, then the world opens up and then you start experimenting. Yes, you may not have a technical background, but there's so many tools and solutions available now where you need to learn to use it. I mean, we all use a computer. We're not all computer engineers. I don't understand how the bits and bites work inside the computer. I do know how to turn it on and I know Windows I have a PC here. So, Windows boots up and I know how to doubleclick and utilize the system. I think with whether it's AI or any solution, these are to me at some level become basically apps. They're software.

19:59>> I double click, it works, and I just need to know how to use them. The thing for most people would be getting to the point of comfort >> that I am comfortable, it's going to work exactly the way I want it as opposed to, well, it'll it'll work the way you want it, you know, 92% of the time. And then there's that 8% where it might go rogue on you and do things that you don't like. >> Then people like, "Oh my god, can I trust this thing? Can I afford to have that 8%?" >> You know, >> I think that's a part that people are challenged with. But at some point, even somebody who doesn't have a technology background, I'd say, you know, it's just an app. Just learning to use an app. There's a next level app that can really help you out. >> Test it out. >> I agree. And okay, last question then. Um, what is the thing you're most excited about potentially implementing in the next 12 or 18 months with Way Metro? >> I think our gateway 3.0 system that we're slowly building up. It has many different components. That is the one that I am most excited about. Uh, bringing this online will be a humongous effort. Uh, it has many different

21:03components, but it it it sets a foundation for a lot of growth and development for the future. If we can get this base initial version right, then I expect the next three to five years will spawn a whole lot of growth. That will be super exciting. So getting this base level right is super critical to me. >> Yeah. No, that's great. And thank you for coming on an I love talking to you because you're really at the frontier of some of the stuff that's happening in technology and community action. So I hope you're able to do this and talk more about AI and do you know more more things like that. But uh thanks so much for for coming on the podcast. >> No problem. Thank you for having me.