Utilizing AI to Reach and Free Victims of Violence, with Megs Shah

I'm really pleased to have with me today Megs Shaw, who is the CEO and founder of the Parasol Cooperative. Megs has spent 25 years building technology products from apps and data analytics platforms to AI platforms for startups and for a number of Fortune 500 companies. She's also a survivor of child sexual abuse. Her personal and professional experiences have led her to found the Parasol Cooperative.

Welcome, and tell me a little bit about the Parasol Cooperative. Absolutely. So the Parasol Cooperative is a technology nonprofit that is building technology to protect lives of individuals that have affected by trauma of any kind. That could be from domestic violence to human trafficking, tech-facilitated abuse, and even other mental health type issues that people encounter on a daily basis.

You know, I founded the organization. It was kind of a culmination of my personal and professional lives that collided during COVID, during the lockdowns, because the initial onset of lockdowns, which created the surge of domestic violence cases around the world, it was almost triple digit increases in countries. And it got me wondering, you know, we use technology every day, why can't we use it to actually help people that are in these really difficult situations, and do it in such a way that it actually giving them the right information to navigate.

Yeah, so one of the things that it does address child sex exploitation and also domestic abuse, right? And obviously, there was a spike in that during COVID. And it's kind of a side question a little bit. But has that continued?

Or we've seen that come down a little bit since COVID? I mean, I think the numbers are still pretty high, I would say. It's not completely come down. But I do think that there's been a lot more knowledge about what the issue is in many different people's lives.

And any number would be unacceptable, wouldn't it? Yes, exactly. Yeah. To the degree that you're comfortable, how did your own story shape when you, how you started to do this AI outreach, this AI tool?

Yeah, absolutely. So, you know, when we started with Parasol, it was just initial sort of idea of using tech. But, you know, being a product developer, I really wanted to make sure we were solving for the right problems. And so the first two years of the organization, we literally just spent talking with survivors, with advocates, organizational operations, individuals, to see where the gaps really were and where can we fill those gaps with technology in a way that's making it efficient but more effective.

And so we uncovered two primary issues. One was access to the reliable information and resources and also safety planning, which becomes critical for those individuals that are still in the homes that are violent. And then the second aspect of it was really trying to figure out how do we coordinate those services in a safe and discreet way. And the second issue, you know, we touched on a little bit, but over a period of time, we had thought about how do we solve for these issues so that it's not invasive, that it's still very private, confidential, and informed enough so that people can take action.

Because there are times, I will tell you from a survivor perspective, I didn't disclose what happened to me until I was well into my 40s. A large part of the reason was because shame was weaponized in this instance. And that's actually a large part of the population of people that don't report. It's because they're worried about judgment or they're just shamed of what happened to them.

And it's not their fault of what happened to them. So how do we make it so that it's a safe place for them to get the right information and not be re-traumatized telling that story over and over again, which is really how Ruth AI came to be? So what does that seem like for the person calling in? What is the assurance?

I mean, you kind of described an AI kind of warm, safe place that's quiet or something. And what is the experience of it that you achieved those things of privacy, a sense of privacy and security? Yeah, so I mean, artificial intelligence is everybody's using it right now. And the way I differentiate general AI chatbots versus what Ruth AI is doing is several folds.

So language plays a huge part in this. Language barriers is like spoken language. We can actually use technology to make it much easier for us to reach people that previously were hesitant to reach out because they couldn't speak that language. But the second part, which is the more important part, and that's where we need to make sure we have the supportive language that motivates them to reach out to get the assistance that they need and do it in such a way that it's safe.

And then when we built Ruth, we had to really consider what are the ways that the conversation could be compromised? What are the ways that their safety could be compromised? And so we don't require a login. We don't require a password.

You cannot put in an EPII into the chatbot. we don't store conversations for more than seven days. It's permanently deleted. So in the case of even subpoenas, we don't even know which conversations to send as part of the subpoena because we don't collect that information.

So it's given this sort of space of being able to dialogue with something that can really help you to understand. And the third piece we really were very, very keen to make sure that we had is we train advocates on a daily basis with certain protocols for different situations. How do you assess where that person is in their journey? All of those protocols were things that we incorporated into the chatbot in such a way that it became another channel for them, especially for those that weren't safe to actually reach out but also ones that weren ready to reach out And so now we reaching early indicators of people that previously wouldn have even had the courage to call the hotline for instance So you describing how Ruth helps you suggesting it but it is a sort of safe quiet intake of people who are having these or being traumatized.

But it sounds like along the way, how do you help people? I mean, in addition to just listening, which, and maybe giving them courage and some of the prompts back or encouragement of some kind? How do you eventually connect them if they need help getting out of their situation? So right now, the way it does it, there's access to the resource list that we keep.

So for instance, we have one of our partners, which is the National Domestic Violence Hotline. They have a resource database that's fully vetted of contacts, providers, and online resources that people can use. So we will tap into that database and provide those that are known to be helpful. And Ruth kind of connects people to that information.

It gives them that information, but the agency is always on to the user to do it themselves. We don't connect directly just because that would create traceability. And so we're very mindful of how we do that. But there is a way that we're working towards doing a live chat transition.

So one of the things that the National DV Hotline and us are ideating at the moment to do it safely and with the right privacy and dignity for the survivor is if they've done chatting with Ruth and they're finally ready to talk with somebody, then can we transition them safely to a live advocate on the chat that they have online? Because that would allow for us to do a safer transition. Now, that agency would always still sit with the survivor, but they could initiate that in a way that can summarize their transcript and provide details that they don't want to.

They can edit it and then send it over to the hotline so that the person at the receiving end, who's the advocate that's going to help them, knows the situation, knows what resources were provided. And so this way, they can navigate that conversation in a more effective manner. How do you get the word out about Ruth? How do traumatized individuals who need that outreach, how do they?

So right now, we work with major partners like Polaris and with DV Hotline and a few others. So they usually can go to their website if they want to use Ruth. We have it free on our website, which is ruth.parasolcooperative.

org. The other ways that we're actually looking to partner now is to work with, you know, my vision was always to work with banks and with utilities companies and with apps that we use on a day-to-day basis because it's innocuous enough to go to those places as a survivor and not have alarm bells going off for the abuser or the trafficker. And that gives us a channel of communication and resources where previously it was much harder to figure that out. So you use the term trauma-informed.

How does Ruth become trauma-informed? I mean, you talked a little bit about interviews and particularly, I guess, listening to survivors. But anything more you can say about that? Yeah.

So the term trauma-informed is essentially where we have supportive language that's being used in order to ensure that we don't re-traumatize somebody, right? So if there's a question that generally gets asked by people that may not necessarily know how to talk with somebody that might've been abused, for instance, right? Well, why don't you just leave, right? That actually is accusatory question, right?

Why are you so weak? Exactly. It puts them in that, again, that judgment and shame mindset, right? And so when we dialogue with somebody in a way that actually motivates them to take a healthier action for themselves, that's really what trauma-informed is, is where I'm asking you questions to better understand your situation and then encouraging you to say, here are the healthy things that you have choices of.

You have a choice to pick which one you want to use. Now, the thing that goes away first in any situation that is abuse-related or trafficking related is the control over your own thoughts and decisions, right? And so what we wanted to do was to be able to equip them with knowledge? Because I've always believed that knowledge is power that nobody can take away from you.

So can we give them knowledge enough to know, here are the actions that I can take, here are the things that I know are my rights, and here are some ways that I can actually get the help I need in order to be in a healthy relationship, in an environment, and not go back and regress into old patterns. It's interesting, this idea of, I guess it stirred my imagination, the idea of a trauma-informed AI. And then I started to think, you know, trauma-informed people. We need a lot more trauma-informed people in the world.

And I wondered if you could say anything about what people who might be listening to the podcast need to understand. I guess I can take that in two parts, right? So the first one is really, how can we equip people to be more trauma-informed? Well, they can role play with Ruth, right?

So Ruth actually does coaching on how you engage with somebody. So family members are known to come to it to ask about situations that their family members going through. How do I talk with them about it? What are ways that I can engage?

And Ruth can actually role play to understand what you would do is just give them the personality of the person that you're dealing with. And then it will react in such a way and it will coach them to say, maybe this is a different way to phrase it. Or here's a potential way to ask them that question. And that actually helps us to now have healthier communication.

which is one step closer to becoming more trauma informed. And then the second part, which is really what FIs can do, I think there's an opportunity for us when we have customers, for instance, if we're going to a bank, you know, and we taking out large sums of money the person at the bank that there to tell her you know I had that incident happen where they asked me they like is someone forcing you to take out this money I mean in that moment I won know what to say if I actually being forced to take out that money because I'm afraid that person's going to find out, right?

So are there ways that we can train our banking colleagues to know that? Now for the back end, which is on more of the transactional identification of fraud or like money laundering activities, there's ways that we can work with them through our enterprise product where it actually gives those insights. And that's what we're working towards. So I apologize, we're not giving it yet.

But we are working towards it to figure out what that typology looks like, right? So are there certain indicators? Are there certain things that we need to have better communications around, but also investigations around, right? The challenge, and I was speaking with Lisa Wheeler from JPMorgan Chase earlier today, and what I mentioned to her is that the fact that when you go in and make that case or that SAR, for instance, and then that triggers an investigation, large portions of the time when it's a trafficking incident, a lot of traffickers will get their victims to actually do illicit illegal activities.

And so they use the investigation as a way to prove that, see, you can get in trouble for this. Don't report it. Right. And so now you have individuals that are fearful of that whole thing.

So, you know, getting them that individual that may be part of that equation in terms of that investigation, the knowledge and the tool to say, this isn't your fault. There are ways out. And there were legislators that have really worked hard to pass legislation around these same issues where victims who were in fact arrested and had a record for being in a trafficking situation but doing these illicit things can be getting those forgiven as well. So there's different ways we can work at this both from actually trifecta, right, from a policy standpoint, from a process standpoint, and then from a technology standpoint.

I think it's interesting. I know that one of the things that banks can do is also in the bringing victims back into financial well-being in terms of, as you said, dealing with - I mean, often these are people who have ruined credit and worse even because their accounts, their names have been used by somebody. So that's really interesting. The other thing that you said that was interesting to me was that Ruth can train people to be trauma-informed.

And that reminded me, and I don't mean to make a false equivalency, but it's kind of like an Al-Anon for, you said, family members. But it is actually dealing with someone who's really stuck. And you may be pleading with them to get out of a bad situation. And in some cases, it's obviously the family that is the victimizer.

But interesting. That's really interesting. And I mean, it seems like a great concept that Ruth can help people become more sensitive and help those around them. I think you're spot on.

You know, I'm actually one of the users of Ruth, right? So I had a family member who was going through substance abuse. And, you know, I struggled to figure out how to communicate with them. And I've been in this space for some time now.

And I actually used Ruth as a way to say, okay, this is the situation. How do I navigate this? like what are things I can say to encourage them to go into rehab, to stay in rehab, to continue their journey forward. I actually gave it to them as well and said, as you're going through this journey, you know, this could be a way for you to just understand whether you're taking the right next step for yourself.

So yeah, absolutely. I mean, our goal again is can we build stuff that can actually protect? And that doesn't just mean from a tech standpoint, it really means from the human perspective. Well, so in terms of building stuff that protects, have you thought about other uses of AI?

And I'm sort of saying that because we're certainly in the financial world seeing AI as being used for all sorts of nefarious things to take over people's identities and all of that. And increasingly, financial institutions are considering or actually adopting in some cases, tools that, you know, it's a little bit AI versus AI. And I don't know if you have any thoughts about that, as someone who's spent so much time doing AI. I probably have a lot of thoughts, but I think at the onset, one of the things that AI does really well is pattern recognition, right?

And AI has been around since the 70s. This isn't like a new tech, but the generative AI chatbots are fairly new because of the launch of the problem. They keep telling me this is the new, better AI. I feel like every time we get a new product, it's usually, yeah, this is better.

And it is incrementally, but not necessarily the revolution yet. Exactly. Well, I mean, I think in the case of LLMs, It kind of is, right? The journey has been slow, but once it got here, it was pretty good.

Large language models. You know, from my perspective, I think that there are ways that, for instance, I can give an example. Obviously, this isn't in practice yet. But so if someone is chatting with Ruth and we have the high level aggregated data sets that indicate specific types of conversations that are occurring in a city, for instance, because we don't go down to the specific location of the individual.

So if we know from a city standpoint what exactly is happening in terms of the issues for substance abuse or trafficking or domestic abuse or other forms, right, that gives indication to financial institutions that are using Ruth Plus on their platform, or even us sharing that data as an aggregate in a partnership of sorts, gives them the ability to know the types of fraud that could be occurring in those areas to then proactively start to look for those types of markers and patterns.

with their artificial intelligence trained on those specific markers. Right So there ways that we can start to bridge across organizations with artificial intelligence in a way that is not invasive but very informative So I think you hinted at some of the things that you looking at going forward But Ruth and the whole cooperative are very new. You know, where do you see this going? You've talked a little bit about that already, about certain improvements that you're hoping for and certain other things.

But anything else you want to say about where you're going? Yeah, absolutely. two and three years out and everything. I wish I was Oracle and I could tell you exactly where we're going, but I think we will adjust as the wind changes, right?

So adjust our sails as we go. But that said, there is a vision that I have, like I said, of incorporating a chatbot like Ruth into everyday locations. And that allows for people to be going into places or going into websites to find the answers that they may be looking for. So that's one.

The second is to be able to get this proliferated across the globe. I mean, we already have users across 179 countries, which is phenomenal. But I'm also the product manager and developer that says we're probably the only product company that wishes nobody has to use our product. Creating a cooperative partnership with organizations like for-profits and nonprofit organizations, where we can create a global safety center that allows for people to know that there is a place they can go and get the assistance they need and gives us as organizations data that we can use to inform our operations, but in a way that doesn't invade into the privacy and safety of that survivor that's reaching out.

And that's really my vision for at least Ruth AI. Obviously, there's other things. That second issue that I mentioned around coordination of services, we do have a product called Safe Connect, which is an early beta. It's like the best way I can explain it is a lot of times survivors will reach out through text messages or emails to coordinate services.

It can be very unsafe to do that, especially if the abuser or trafficker has access to that information. So our way of doing this is basically using Calendly type functionality, where you can request a calendar invite, for instance, for a specific activity, but obfuscated enough so that that doesn't raise red flags, and then creating a case management process flow within the application that works with what advocates do on a day-to-day basis and not necessarily in addition to.

And so eventually I'd love for Ruth to be able to connect into SafeConnect, which then allows for us to have an interface for survivors to engage with and be able to request the services they need, and then for organizations to have that data and that effective process in place so that they're more streamlined and do it in such a way that it's not cost burdensome. Because that's one of the reasons nonprofits are not incorporating tech. It's too expensive. And I have a last question.

But before I ask it, I was just going to clarify, you said you're in 170 countries? Yes, 179 countries. 179 countries? End of April, yeah.

As of the end of April. And do you know how many people have used the roof? Oh, because you don't really keep it. It's secret.

So you do not know how many. Well, we know the conversations, so we don't know the number of people. But we have, so right now, the last number I checked was over 100,000 conversations initiated. And then of those, we had about 60,000 to 70,000 conversations where there were at least four or more messages being sent.

So from that perspective, it's really great. I think the number that really surprises me and I love looking at is the helpful or mostly helpful rating. So there's an optional survey at the end of a normal closure of the chat. So we have this button that's quick exit, that quickly exit off of the page and deletes that from the history.

Yeah, you know, someone may be watching this person. Exactly. And there's a normal path of ending and closing the chat, which actually brings up this pop-up of a survey. And they can fill in whether it was helpful, not helpful, give us any feedback in writing if they want, and then tell us if they're okay with us sharing it more anonymously with people.

And that rating actually recently, I checked it, it was at 92% helpful or mostly helpful. And that tells me that what we're doing is actually helping people at the end of the day. And that's what we want. Well, speaking about helping people, is there something you could say to this audience?

I know you're here talking to bankers about making them trauma-informed and getting them on board. In closing, is there something you want to say about how people can help you doing what you're doing, who can help the Parasol Cooperative and also Ruth become friends of Ruth. I think the best way that nonprofits can be helped, no matter what nonprofit it is, is funding, right? So there's definitely opportunities for CSRs to reach out to us to talk about ways we can partner that way.

The second way is to incorporate Ruth Plus into their model, both for internal employees and for external customers. And the way that we've designed it, the liability is very minimal. and actually almost negligible. And we're happy to discuss that with people in terms of how you incorporate that into your chatbot.

It literally takes a week to two weeks to enable. And then the last part would be really figuring out how we can collaborate on those types of typologies or strategies and content and data that needs to be there in order to do the day-to-day work that FIs do, but also to inform how we support survivors better. Megs Shaw, founder, CEO of the Parasol Cooperative. Thanks for being here.

Thank you for having me. This was wonderful.