Erine Gray, MPA is founder and CEO of Findhelp.
Tell me about yourself and the company
Findhelp is 14 years old, based in Austin, Texas. We focus on simplifying the way that people find and enroll in social services in the United States.
Our company’s history began with the realization that understanding your options in times of need is not easy. We started by building a search engine that allows people to put in their ZIP code and find available social services, such as affordable housing, food, or even programs that are available through public entities, such as the federal government and state government like Medicaid, Temporary Assistance for Needy Families, Children’s Health Insurance Program, and the Supplemental Nutrition Assistance Program. Our platform supports many of the 1115 waiver programs in the United States, which are experimenting with allowing social services to be paid for with Medicaid dollars. We are doing that work in New York, California, and in other states throughout the United States and plan to be doing more in the coming years.
The goal is that people will find those services and then click a few buttons to apply and get enrolled if they qualify. Once they do, we’re the largest closed-loop referral platform that allows people to circle back and say, yes, I did receive these benefits.
Our customers hope to see whether these interventions improve long-term care. We are a full-stack platform for allowing that to happen. They will be able to see what actually occurred from the very beginning, when they were searching Findhelp.org to find a program, to the very end, once they’ve received the program and interacted with their health plan, the health system, or other community organizations. The goal is to study that information over time with the hope that some of these interventions will help them live a healthier life and get through their difficult moments.
How has the safety net changed recently and how might it change going forward?
Generally it hasn’t changed in the last probably 50 years since the Great Society under Lyndon Johnson. My theory is that it has only changed at the margins. There have been two enormous events with respect to the American safety net. One is the New Deal under FDR and then the Great Society under Lyndon Johnson, where he created all sorts of new programs such as affordable housing programs, expansions, and SNAP benefits and things like that.
For the most part, the safety net stayed the same. It’s only lately where people are beginning to think about, is this an efficient safety net? After indexing every program in the United States, I believe that the safety net is inefficient, with large cities hosting thousands of redundant non-profits. I think the safety net will see more consolidation among these non-profits as technology makes it easier.
You’re already beginning to see that consolidation with mergers of different community-based organizations, different United Ways throughout the country. There was one within the last couple of years here in Austin between two of the neighboring United Ways. For those who don’t know, United Way is a great organization that works on collective impact by looking at the entire community and looking at the entire landscape of organizations that serve others.
What I also think will change at this pivotal point in the future is that the federal government is experimenting with allowing social services to be paid for with Medicaid dollars. But the reality is that it’s still a theory, meaning that despite what vendors may tell you, there isn’t overwhelming evidence quite yet, at least through academic studies, that make the case that these interventions will drive down the cost of healthcare. There are so many variables that it’s hard to manage. It’s hard to study that and there’s not a ton of information.
My gut feeling is that allowing states to have more flexibility around what they spend their Medicaid dollars on will ultimately result in better health outcomes, but it’s going to take years to prove that. If we can prove that, then we as a society, or at least the states, will have the ability to decide what interventions are more relevant. For example, if a Texas Medicaid member is unhealthy and they can’t use their air conditioner because it’s broken or they can’t afford their electric bill, an intervention might make sense, if it’s medically necessary, to pay for that electric bill or that air conditioner repair with the dollars that came out of a Medicaid budget. That might be a different intervention in Maine, where transportation is a big issue, getting to the doctor is a challenge, and those Medicaid dollars could be used in Maine.
I hope that with the change in administration, more flexibility can be given to the states to allow for them to design their own interventions with some flexibility of using Medicaid dollars to design these social drivers of health interventions that make a lot more sense at a local level. To allow local care coordinators to decide that they will help this family with the rides to the doctor or help with their electric bill based on everything that they see about the person. I’m also optimistic about the future that less regulation at the federal level of the Medicaid program, in this case, will lead to more innovation in the future. I feel like the states probably want that flexibility and are competent enough to administer innovative programs these days.
How are health systems using social determinants of health and what role do they play in addressing the needs that they might identify?
We work with a couple of hundred health systems nationwide. They have different requirements to look after the social needs of their patients. In some cases, those are driven by federal and state requirements. In some cases, those are driven by risk-sharing agreements with payers. Their goal is to make sure that the patient gets healthy and doesn’t consume unnecessary medical services.
Our customers integrate our platform into their system of record, whether it be Epic, Cerner, EClinicalWorks, or NextGen. We are integrated with hundreds of organizations so that they don’t have to do double data entry. We have the patient context when they click into our platform. We utilize SMART on FHIR integrations and some deeper API-based integrations. They want to be able to, within their medical record, make a referral to one of the social service programs that we index and that we contract with. To be able to make those outbound referrals and also understand what happened after the person received those benefits. That allows them to have a whole picture of what’s going on with the person from right within their electronic health record.
Different hospitals are doing different things. For example, Boston Medical Center has been innovative over the last five years that we have worked together by implementing all sorts of initiatives. Some are going as far as putting food pantries into their systems. Others are building community gardens. Others are contracting with non-profits to provide additional services for their patients. We’re the software platform that integrates with that and helps make those types of interventions happen.
Do the social services organizations receive the SDOH information in a standardized format and then have it integrated with the systems that they use?
We have indexed about 550,000 distinct program locations in the United States. These are all physical locations that provide social services to people in need. Probably the most common system of record is still on paper. Second to that is Excel spreadsheets or Google spreadsheets.
There’s a long tail of proper case management systems that non-profits use to run their organization. One of the more popular ones is Salesforce. They have a program where they’ll provide up to nine licenses to non-profits that sign up for it. We’re building integrations on that side as well. We have built several integrations, but the critical mass is still in spreadsheets and on paper for tracking that information.
There remains an enormous opportunity to educate these community-based organizations and bring them onto our platform. We will build integrations with these non-profit systems of record, but when they don’t have one and they haven’t made that investment, we provide a free platform that they can use. We’ve done that ever since the beginning of our business, so we have been able to recruit many to come over and use our platform.
It’s a big investment area for us going forward. In fact, just this spring, we acquired a company called Kiip. It’s a case management system that is designed for these community-based organizations to be their system of record. We have since launched in the fall a fully integrated version of Kiip that utilizes Findhelp’s network through our APIs to be able to use the Kiip case management solution, make those searches to find services, and make those outbound electronic referrals that hit the Findhelp network. The information is then stored within Kiip.
We offer this for free, with an optional premium version for non-profits. But because there wasn’t a critical mass case management solution, that was an opportunity for us to create one and to put that out there in the world. We also see this as something that isn’t exclusive to the Findhelp network. If there are other networks that exchange electronic referrals for the purposes of social determinants of health, we think the Kiip solution can be utilized to recognize referrals in other networks.
In the long run, I see this going like the cell phone networks, meaning that you can use your Android phone to connect to a different network if you want to. The same thing should be happening in our space to eliminate the need for one monolith. We can lead by example to provide a system of record to these non-profits and then allow them to add as many networks as they would like so that they can see electronic referrals in one consolidated location.
The challenge is that we have to find the motivation of the community-based organization. Why should they close the loop on referrals? Some might say that it would help the hospital or it would help the health plan in their community, but that motivation just isn’t enough. It takes good software design to build the closing of the loop into the workflow for reasons that the non-profits want themselves.
We have a lead user experience designer named Phil Robinson, a great member of the team, who focuses on understanding the motivations of these non-profits. A big part of what they’re motivated by is having good, clean records of what’s happening. Building the loop closure into that workflow so that they have accurate reports that they can use reports for fundraising, for running their operation, is a much better motivator for the non-profits. That’s where we see a huge opportunity, not only in the short term, but in the long term, to have somebody on the other end of that referral closing the loop on that for their own reasons. We expect to see even more growth in that area.
Who benefits from that? The hospitals benefit from that because they will see a higher closed loop closure rate. They will get information that would allow them to establish better interventions in the long run. If they see that a community-based organization is active in their city, they might want to work together and maybe even reimburse that non-profit for some of the work that they’re doing because it benefits their patients in the long run.
Findhelp is trying to solve the information problem that exists. A big part of that is bubbling up that information so that smarter people than I can do something with it. We believe the key lies in providing good user experiences, Not only at the hospital and plan level, but also at the non-profit level.
Will AI affect what you do or how you do it?
I’m definitely not an expert on the subject, but I think that the number one benefit that AI can have, at least in our world, is to improve the workflow of the user.
I’ll give you a quick example. Our customers are continually telling us about changes in their community that they discover while using our software. They might discover that a program that serves people who are looking for affordable housing has changed their hours of operation. We built something called the Program Manager that allows our customers to make those updates directly. We have a curation team in-house here in Austin and we work together on those situations. But what we’re building into that application is the ability to proactively suggest those changes and to correct errors in real time.
We use a taxonomy called the Open Eligibility taxonomy. That’s a free and open source version that we built and put out there in the world. A user may not understand the tags as well as some of our employees might understand the tags, so we can build AI models that look at the descriptions of the programs and say, I think this tag is probably more accurate, would you like to select it?
Those are the tiny things around the edges that are going to make for a better workflow in the long run, which makes for a better data set in the long run. That will be an important part of decreasing the amount of time that our customers use to interact with our systems.
Another area is using AI models to help our customers understand which patients might need an earlier intervention. We have a lot of data about a patient, coupled with data that’s in eligibility files with the payers and other systems. When you have a large number of patients that you’ve made outbound referrals for, we learn a little bit of information about what occurred with those referrals. If I’m the case manager sitting in front of my computer the next day, sometimes it can be overwhelming. Building features that allow you to see that Hannah could probably use a phone call right now. That’s where we see our organization using AI to increase or improve on the user experience of our users. There’s probably many more that are there.
As we uncover the use of AI for looking at large data sets, I’m interested in that in the long run. We just crossed 50 million users on Findhelp, and over 20 million of those users were within the last year. We want to be able to study that data with the help of some of these models to understand trends that may be happening in a more regional level at the community. It would be more macro, but that’s another area where we think there’s a lot of possibility. The challenge is deciding which ones to focus on first. Right now we are focused on making our workflow better and better by using some of these technologies in the future.
What will affect company’s strategy over the next few years?
The number one thing that affects our strategy is that we are trying to build a new safety net, starting from scratch. When Social Security was first enacted, the business problem that came across was there was not a unique identifier for identifying every American. Hence, the Social Security number came out. If you fast forward to unemployment insurance, it was hard to come up with a rate for unemployment insurance because there wasn’t a centralized way of storing people who were unemployed.
Information problems have presented themselves throughout our history. That same information problem is happening today with respect to organizing the safety net. I was reading this book by Henry Seager called “Social Insurance: A Program of Social Reform.” He wrote, “It is impossible, with our present knowledge, to estimate the extent to which illness and death are preventable.” That was written in 1910. What is pretty amazing is that because of the work of a lot of your readers, that’s no longer a true statement. Today, that information has been digitized. The information problem that prevented people from estimating the extent to which these illnesses are preventable has been solved.
The biggest thing facing us going forward is, how do we do the same thing for the social services sector? Our contribution to the world will be that every American understands what their options are in their time of need, but also trying to ensure that as many non-profits as possible have the tools that they need to solve some of these problems. Just like we’ve solved the unique identifier issue with respect to Social Security, your readers have solved the electronic medical records. Making them electronic in the first place was an amazing feat. We would like to do the same thing for the social services world.
RE the AI GLP1 company, Washington Post has an article today by someone who used one of those compounded products,…