Upstream by Design: Integrating Investigation, Engagement & Technology in Child Welfare

In Episode 2 of the CSI Intel™ Brief, we’re zooming out.

Child welfare systems often treat investigation, engagement, and data as separate functions. But when those components operate in silos, leaders make high-impact decisions without full family visibility.

This episode aligns with an upcoming APHSA leadership conversation focused on upstream integration: how early family reconstruction, supported by intelligent data architecture, can reshape decision-making across jurisdictions.

Key Takeaways:

• Where systems lose family visibility
• How technology can responsibly support relational practice
• What governance and oversight structures must be in place
• And how upstream integration improves fiscal and workforce stability.

Find something interesting or would like to learn more? Visit our website at csiinvestigators.com to meet our team, schedule a consultation, or contact us directly via phone.

Transcript

SPEAKER_00 0:09

All right, welcome back everyone to the CSI Intel Brief: Providing the Facts to Act, a podcast exploring how investigation, intelligence, and human-centered engagement help organizations bring clarity to complex situations. Today is episode two, Upstream by Design, integrating investigative engagement and technology in child welfare. In our last episode, we explored Family Connect and how early structured family identification changes outcomes at the case level. Today, we're zooming out because this isn't just about a program, it's about system architecture. Child welfare systems often treat investigation, engagement, and data as separate functions. But when those components operate in silos, leaders make high-impact decisions without full family visibility. This episode aligns with an upcoming APHSA leadership conversation focused on upstream integration, how early family reconstruction, supported by intelligent data architecture, can reshape decision making across jurisdictions. Today we'll explore where systems lose family visibility, how technology can responsibly support relational practice, what governance and oversight structures must be in place, and how upstream integration improves fiscal and workforce stability. So joining me today are Ann Schlegel with CSI Corporate Security and Investigations, and Travis Larson with Smart Systems Technologies, who work directly in the child welfare and family services divisions of CSI. Welcome to you both. Welcome back, Ann.

SPEAKER_03 1:53

Thank you.

SPEAKER_00 1:54

Ann, I know uh people may remember you from episode one, but uh for folks that are tuning in for this episode, can you provide a little bit of your background uh and let people know who you are?

SPEAKER_02 2:03

Sure. My name is Ann Schlegel. I've been involved in child welfare my goodness since I graduated college. So going on 30 years of child welfare, anywhere from caseworker, supervisor, program specialist, and then leading child welfare agencies as a director, and now joining the CSI team probably about two years ago. And the goal of me joining the team was really to broaden my impact on child welfare and be able to create and implement and provide guidance um to child welfare agencies and to CSI on the needs of child welfare agencies.

SPEAKER_00 2:43

As we've said before, you've seen it all.

SPEAKER_02 2:45

Yes.

unknown 2:46

Okay.

SPEAKER_00 2:47

And Travis, welcome. Thank you, Pete. Could you give us a little bit of your background?

SPEAKER_01 2:51

Sure. Similarly, I've been involved in the uh the public benefit sector uh for state, state um health and human services, uh, but on the technology side for for my entire career, uh been delivering, been building, coding, uh, maintaining, and and then working for uh agencies as they deliver systems like child support enforcement or um uh Medicaid eligibility or integrated eligibility for uh aid-to-needy families, uh food stamp programs uh across the board, uh, multiple states. And um I'm excited I'm part of Smart Systems Technology, working with CSI and really excited to help them reframe early family visibility as critical infrastructure strengthening family systems.

SPEAKER_00 3:41

And as we learn in episode one, we I think people understand how important this mission is for what you both do. And uh I think this is great to understand now how two companies working together can can really bring technology to the human element of what you do and make sure it's as effective as possible. So applaud you both for what you do. So Anne, we're gonna start with a fundamental question. Where do systems most often lose visibility into family networks?

SPEAKER_02 4:11

Yeah, uh visibility is often lost somewhere between urgency and fragmented uh information. So urgency as far as the investigation stage of a case, early on in the case, when crisis is happening, or crisis happening anytime throughout of a case, and the information silos that occur within child welfare. So on the caseworker end, they're in crisis, they're doing what they do best, they ensure safety, um, they are trying to find resources to uh to address that safety and information that could help them in order to learn more about the family, are in so many different uh silos, so many different systems. They either they do or do not have access to.

SPEAKER_00 4:59

And and when we're talking silos, we're using it in the the description as often as that people are are limited to what's surrounding them, and that's the only information they have to make decisions upon.

SPEAKER_02 5:10

That's correct.

SPEAKER_00 5:11

Okay. So, Ann, in in your space, the term upstream by design has a meaning uh for people who do what you do. Can you talk about that for a minute?

SPEAKER_02 5:20

Yeah, yeah. Upstream by design really means being able to build systems to identify supports and the family network sooner, earlier in the case. That's pretty much the overall meaning of that term.

SPEAKER_00 5:34

And as we discussed previously, why is this a leadership conversation and not a product conversation?

SPEAKER_02 5:41

Yeah, so uh products cannot provide information. Products cannot determine when that information is to be used. Leadership does that.

SPEAKER_00 5:52

Okay. And when we're talking about people on the front lines and doing what they have to do, are they held to a certain standard of requirements for their jobs? Is good faith something that they need to adhere to in the performance of their duties?

SPEAKER_02 6:06

Yes, absolutely. When I hear the term good faith in child welfare, I always think of compliance focused. Yeah, child welfare is um responsible for providing good faith efforts on family identification. So whether that be making some phone calls, sending some letters out, making note that you asked asked uh a parent if they had any kin available, um, those would be good faith efforts.

SPEAKER_00 6:33

What is the impact of treating family finding as a good faith effort instead of critical proactive case management infrastructure?

SPEAKER_02 6:41

Yeah, so making good faith efforts, you're not getting that full family picture. You're doing your basic general overview of who might be a debt child or that youth's immediate connection. Um, you're not getting that full family history and that full placement. The impact is really on placement decisions, on court decisions, on permanency decisions. So the decisions that you make going forward with that information you have are impacted by the amount of that information, whether it be good faith information or whether it be the full family connect information that we've talked about.

SPEAKER_00 7:19

And is it fair to say that most agencies, given the constraints that they're operating under, are operating in a reactive environment rather than a proactive or preventative environment? And if that's the case, you know, uh, and the there was a shift into more human-centered tech approaches, what would this do to the overall child welfare industry, if you will, in the next five to ten years?

SPEAKER_02 7:47

Oh my, I think that that would have a tremendous impact. More children would uh be placed with kin. Um, there'd be earlier family visibility, there would be reduction in the length of time it takes for uh permanency to occur, there'd be uh fewer fewer entries into placement or into the system itself. There would be stability uh for kids and and and youth and families if um they had that earlier visibility.

SPEAKER_00 8:18

So it's a shift in thinking, it's a shift in approach. But it sounds like, and correct me if I'm wrong, but this shift potentially could have the impact of better decision making earlier in the process that benefits the child at potentially less cost. That's correct. So, Ann, you mentioned a little bit earlier about siloed data. Um, can you talk for a minute about how this can distort leadership decisions?

SPEAKER_02 8:45

Yeah, so I think child welfare, that's one of the issues that's that is significant, is not having all the full picture or all the information needed in order to make decisions. It is that siloed data. You might have a piece of information available to you in your case information system, or you don't have access to a system that that would allow you to make that decision, or some of that family network or information is in a school system or um a police system that you're not able to easily access. So what's really happening is the resources are focused on trying to find all of these pieces of information rather than being able to quickly look at it, act on it, and make good decisions in that in the time frames necessary.

SPEAKER_00 9:34

So it sounds like all these important potential data sources are pretty disparate and in a lot of different places and county to county, they probably look at or look different. Uh, there's not uniformity to it, and it's just knowing where to look and how to access things probably becomes a big problem. Exactly. And do you have any real-world examples of how this issue has led to a child falling through the cracks?

SPEAKER_02 9:58

Yeah, I think um one of the examples that comes to my mind the most, because I think I've seen it the most often, is the impact that it has siblings in her care. So we might have older siblings who have graduated from the child welfare system who have moved on. And now there's younger siblings who are entering the system. The disconnect between siblings themselves. Child welfare might not even, those cases on the older siblings have been closed. So they might not even have access to that information anymore to know who those siblings are if they weren't identified early. Um, to know maybe who those siblings maybe were connected with or placed with, um, that would absolutely expand that family network for the younger siblings coming into care. It is that disconnect in systems that really makes disconnect in families. And siblings is the ones that I see it the most often in.

SPEAKER_00 10:53

Yeah, and I would imagine that's that's got to be heart-wrenching if in hindsight a case worker realizes there's this information that's out there that they could have acted upon or would have acted upon and just didn't have it.

SPEAKER_02 11:05

It reminds me of a case that we had recently with a youth who was getting ready to leave the system. He was aging out of the system, and the county agency had asked us to find supports for him so he can have some supports as he ages out. And in our work, we found that he had siblings that he didn't even know he he had.

SPEAKER_03 11:25

Right.

SPEAKER_02 11:26

And these were siblings that you know the county agency wasn't aware of as well because they weren't listed in their immediate case management system. So um, through this work and being able to have access to that information quickly, we were able to reconnect this this youth with siblings he did not know he had.

SPEAKER_00 11:46

That's incredible.

SPEAKER_02 11:47

Yeah.

SPEAKER_00 11:48

And Travis like to turn to you a second. So there's something called the process abstraction layer.

SPEAKER_01 11:55

You know, let's let's actually um let's step back a little bit here. A lot of the the challenge facing child child welfare and um the caseworkers is is this you know heavy burden uh of work. You heard a number of tasks that they that they have to do and that they are skilled in doing. Uh, but then there's all this other work that doesn't fall into that skill set. And so we you know, we approach systems design from um a human-centered perspective. What are the what's the hu what are the human, what's the business workflows, what are the human workflows, and uh what do they need to make those decisions? We want the caseworkers to be enabled to to focus on where they're where they're applying their their expertise, their creativity, their understanding of uh the child's needs, you know, child-centered decision making, uh, taking the the research, the investigation, the um, you know, the systems access, you know, the compliance and the reporting requirements, all that stuff that adds up. We all see it in our own lives that we're faced with all this additional work to work with technology and you know, and and we hope to bring a an approach to to child welfare that that frees that caseworker to be able to focus on those those key decision points and um and enables um you know the the majority of their day to be spent in in that value that they bring to the table that that um other people can't do, and that that we can shift the portions of the workflow or those processes either to automation or to uh to other skilled uh resources that can focus in those areas that that can't do what the caseworker uh that we want the caseworker to be doing.

SPEAKER_00 13:51

And it's interesting what you said human-centered technology approach. Right. And it to me in hearing this, is it sounds like that this sort of flips the typical idea on its head of, well, we're gonna place a piece of technology in here and you're gonna have to conform to the technology. It sounds like it's the reverse.

SPEAKER_01 14:09

That that's spot on. So much of technology sales and even solution design, or or uh when I'm when I'm faced as a agency head or a policy lead or a legislator, I want a solution that does child welfare. Or, you know, that's what I that's what I think. But when when that happens, you end up with a a system that the that the team of people have to bend around because it's it's got stakeholders at the federal level, at the state level, at the county level, at law enforcement. And uh those are all good requirements. You know, the taxpayers are paying for these things, but it's not necessarily uh supportive of the outcome. Uh and it and it drives all this extra tasking that uh that distracts or pulls away from uh the caseworker being able to focus in those uh those key valued decision points along the the case life cycle of of actually finding the resources for the child that most uh suit the child's needs and that uh point to the uh the the stability and the the outcomes that we want for that child, which is you know uh you know reconnection with the family and and um you know education and health and safety.

SPEAKER_00 15:27

Yeah, and I and I imagine what Ann described as this silent effect is just it can be amplify the problem of improper technology or technology not being properly used, et cetera, because you you lose sight of the objective.

SPEAKER_01 15:41

Right.

SPEAKER_00 15:41

And then not only on top of that, you spend all your time filing your TP reports to miss miss out on what am I trying to do.

SPEAKER_01 15:49

Right. And and those are requirements. I mean, yeah, we need to do that, we need accountability, we need to build that in, but uh, but we want the technology to be doing or the system to support, not demanding that overburden on the caseworker of having to spend their time doing that administrative type work that that supports the reporting, that supports the uh the accountability of what's going on, but if if uh if it can come alongside instead of uh increasing the demand away from actual case management.

SPEAKER_00 16:22

So, how how do you propose implementing technologies that gets around the siloing issues?

SPEAKER_01 16:28

So that's where our systems design approach shines. The uh that the process abstraction layer is what enables us to bring that that systems design approach to bear, looking to look at the actual human workflows. What's the work that's getting done, what what has to be done? And if you can abstract or decouple the logic from the the demand of that uh specific piece of software or that system that you have to comply with and isolate uh the functional, the functionality or the functions that are happening along the way. I need uh the research to be done for the kin to be identified. And that might mean hitting seven different data sources. What's how how best do we do that? Can we automate some of that? Can we have researchers that do that and then feed into into the casework or the engagement liaison that um, you know, and then processing, you know, reaching out to those individuals, you know, who's best uh able to do that? Appointment setting, uh scheduling, uh, qualifying that person from a uh, you know, are they really uh who we think they are? Do they have an interest? If we have 70 kin identified, uh who's it, who's interested, who's available, or who's got, you know, who's who's overwhelmed already with with the situation that they're in, so that the engagement liaison or the caseworker can can start conversations with the people that are willing and able, you know, changing the balance of effectiveness. You know, instead of making 20 calls today, I'm making three and I'm meeting with people that that have availability and resources and willingness to to intervene in the life of this child.

SPEAKER_00 18:13

And I have to imagine that the more this is implemented, because it sort of flips the narrative on its head where it's technology supporting humans instead of humans supporting that.

SPEAKER_01 18:24

That's the promise, right? I mean, that's always the promise. So with that, this this process abstraction layer, um, then the then the caseworker does does does their work. They do, they they engage with the identified kin. They they determine that, yes, this is this looks like a good thing, but um you still have the the case management system back there that needs to be be updated. Can that be automatically updated? You know, the interface that they're dealing with, just focusing on capturing the interview contents or what have you, and then uh automatically pulling it back in or moving it down, uh generating the reports you know for the different stakeholders, uh taking that off their off their plate, but really encapsulating the steps along the way in such a way that that freeze that you can you can provide the best uh solution for that that task, right? That that granularity versus the the system. Maintain the system or keep the system, have it have these these agents or these these individuals, and that but take it off the plate of the uh the caseworker.

SPEAKER_00 19:26

And I would have to imagine that if if there's an adoption on a bigger level for these types of technology integrations and the use, that the the chance to expand best practices across is is there. Right. It's a natural progression of something like this.

SPEAKER_01 19:41

Right. And without the risk of of having to go through um a rip out of the existing system and a replacement of that system or um you know recertification of that system, the system, the the skeleton is still there, the team that supports the technology is still, you know, that expertise is all you know, is all able to still there. This can just lay on top.

SPEAKER_00 20:02

And Ann, do you see practically speaking, from your experience in this field, that the minute you mention a technology piece, there's this technology fatigue or not another software solution that pops in.

SPEAKER_02 20:15

I think I tuned out halfway through halfway through what Travis was saying. I mean, it's for for Chal, it's about the practice. Right. And how is it going to help support that practice? And we want to make sure that the practice leads and the technology is not leading how it's going. Right. And that has always been a challenge. Um, there, as Travis said, I mean, there's promises that it will remain human-centered and or human-led, but I think that that's always the fear is that now we're going to have to change our practice to make so it can fit into what technology is supposed to do for me.

SPEAKER_00 20:50

But it sounds like the philosophy here is look, we're going to make more resources available to you on a correct dashboard or however you want to describe it. That when you need X, you know where to go. You need Y, you know where to go. And it's supporting you with your critical mission, which is find out what's the best decision to make for this child, not waste hours doing. Absolutely.

SPEAKER_02 21:12

It takes off that administrative burden.

SPEAKER_00 21:14

Yeah.

SPEAKER_02 21:15

I mean, it allows us to do the specialist work of investigating locating and that digital footprint analysis and allowing caseworkers to be able to do what they do best, assess safety, and get out in the field and see their families and and kids without sitting behind the computer all day filling out paperwork in order to be in compliance with and so we talked about this briefly that so one of the big issues is that the data is everywhere.

SPEAKER_00 21:47

It's all over the place. Yeah. And there's so much. It's so much. It is just growing.

SPEAKER_01 21:51

Right. Yeah, you've got to connect all that. Well, that's where technology can shine. We talk about a data tsunami when I mean, you just look at the data that's available out there, you just can't read it all. You just can't process it all. There's there's a lot of AI technology that that can ingest and and uh and interrogate that data and and and be I mean and we experience it like you know, what's the what's the best Italian restaurant in Pittsburgh? I mean, there's lots of reviews out there. Uh you know, you're only gonna go see, you know, three or four things, but you can you can go ping grock or chat gpt and it's gonna quickly ingest all of those out there and come up with this is what the people are saying, you know, kind of in a summary fashion. Similarly, we can target tools that that can that can uh digest and and pull out now, and this is where you know the the rub is in the details of of um how is that accurate? Is there bias in that? And um, and there's a lot of concern there. And so, but the capability is there and um and we need to leverage technology to support this because it's just it's not it's not slowing down, it's only increasing.

SPEAKER_00 23:07

Well, and I mean let's look at AI in our worlds, you know, you go ask a question. It used to be that you know, well, here's 85 sources that might contain your answer, go look at them. Now it's here's your answer. You have to question whether it's accurate or not. But if you have access to the sources where it came from to be comfortable that that is a correct answer, now you you have the facts to act.

SPEAKER_01 23:29

Right.

SPEAKER_00 23:30

And and if if you get to a point where this process of taking lots of information saying, look, it's a recommendation, but here's the backup, and here's the company that that can make sure that you can rely upon this information is available, you you now have that early intervention, I think, has been missing from this whole process.

SPEAKER_03 23:49

Right.

SPEAKER_00 23:50

And do you and find that you tuned out and Travis and I talking our tech speak? How hard is it to suggest a technology implementation into an organization and like extremely difficult.

SPEAKER_02 24:04

Okay. Uh that's mistrust.

SPEAKER_03 24:08

Yeah.

SPEAKER_02 24:09

Mistrust is the first thing that comes to mind. Fear. Fear of, well, if this does work, is it going to take over everything that that we do? But I would say mistrust is is number one.

SPEAKER_00 24:20

And is there, I'm sure you've heard this before, a fatigue of we've tried this before, it doesn't work. Or I'm still learning the system from five years ago that somebody made us buy, you know.

SPEAKER_02 24:30

How many systems have been tried through child through child welfare?

SPEAKER_00 24:34

Right. Yeah. So how do you get how do you get past that?

SPEAKER_02 24:39

Um, so I think for us, it's the it's us still having that, the the specialist who are utilizing the system, who are providing child welfare the information, who they have developed trust with to be able to get past that.

SPEAKER_00 24:56

That there's a human behind the technology that they know are or somebody who understands what they mean.

SPEAKER_02 25:02

We have specialists who are analyzing all of that information that that is coming about. We have engagement liaisons who are verifying, helping to verify that information. So whenever it's at the county's hands, it's their verified information.

SPEAKER_00 25:19

Gotcha. And and I think we started this off by saying, you know, this this is the topic for the upcoming uh remind me the acronym on the convention that's coming.

SPEAKER_02 25:28

A PHSA.

SPEAKER_00 25:31

There you go. Why is this such a big topic?

SPEAKER_02 25:34

I I think that it's because it's so prevalent right now in the world in general, not just in child welfare. I mean, we are becoming a AI technology world. I wouldn't even say becoming, we're there.

SPEAKER_03 25:47

Yeah, right.

SPEAKER_02 25:48

And I think child welfare is usually always the last to catch up on innovation and on these sources that could do that work. And I think we're there now, is the child welfare saying we need to be upstream on this. We need to um have the tools and technology at hand to be able to get family clarity early, to be able to change how that that case is going to turn out down the line. And if we had this information earlier, how do we get that information earlier? We need to enhance our technology.

SPEAKER_00 26:20

So is it fair to say that the child welfare space, if you will, is sort of at a pivotal point to say we know this is valuable to be able to do these things early, and we may finally have the tools and processes that support the people to make this happen.

SPEAKER_02 26:38

I would say the tools, the processes, the innovation research, the innovative research out there. There's been so much research being done on it and the drive to move it forward.

SPEAKER_00 26:50

So understanding the importance of this and that we're at this point in time where we we may have the ability to successfully implement the right technology, supporting the humans and not the reverse, you know, how do we make sure this technology supports and doesn't replace professional judgment and empathy? I mean, empathy is critical.

SPEAKER_01 27:10

You know, I'll take that. You know, how how we how we attempt to to mitigate that and keep you know keep it human-led, keep it human-centered, is in that granularity. There should be no decision made by a machine. Right. All decisions should be uh reviewed and and uh well, actually be made, explicitly be made by the caseworker. Uh we want that human judgment. We need that human judgment there. The technology should be applied to support providing the information to enable that decision to be made earlier in the process, to be made from a more comprehensive set of information that applies to that case and uh and be you know, be be more thorough, get provide a more thorough understanding of what's going on, not making decisions, not choosing placement. Uh so we talked, we mentioned bias a little bit in there. Um, you know, it's just like in like the the daily news, you're reliant on that editor for choosing what you see versus what's not shown. That's a decision. Here you've got to validate the process, you've got to validate that what the technology is giving you, what it's coming from, and and be and be good, be be very comfortable and confident that that's that is a thorough and comprehensive representation of the data that's available, not uh, and not skewed, right? And not skewed for um, you know, either a political um perspective or uh uh or even um economic. Yeah, economic or uh or or um um social. I mean, just you you you want to be dealing with the objective picture of what's going on, you have to be able to trust what it's what it's presenting to you. So keeping things in in at discrete, uh well-defined uh stages along the way, and then being those those being the inputs, but but you have to you have to have a process that you've validated what it's generating. So, what's that look like?

SPEAKER_00 29:14

How would you put guardrails around something like this?

SPEAKER_01 29:16

Very good. I think uh, I mean, there I don't know about you, but but I and I think many of us are well uh justified in in that concern. You know, you've got the White House um blueprint on um you know AI Bill of Rights. You've got uh you've got the executive order on on uh safe, secure, and transparent uh AI. You have um NIST industry standards for transparent AI. I think uh you it's a black box, it can be a black box. And and uh I think we even have um some state uh you've got state, uh some states have enacted laws uh demanding transparency and or mandating uh transparent accountability of what's going on, uh what how is it getting to where it's going? And and so as a vendor, you need to be willing to open the uh the doors to show how it's getting there, uh what the inputs are, how it's processing, and what it's put uh uh what it and what it's producing. There's uh a concept of drift, even. So you do it once, that's not enough. You need to have ongoing maintenance and auditing and accountability to to show that it it's still producing, it's still doing it accurately. Yeah. Um, and so you have to build that into the infrastructure of the project so that over time it doesn't drift uh in a direction that you're not aware of, because it could very easily be uh uh something that you don't notice because you don't know what's not getting shown.

SPEAKER_00 30:48

Well, I I mean, and Ann talked about this a lot that right now, and again it goes back to the silo issue, people are making decisions with no information. Right. So, you know, even if you take those two comparatively speaking, which is the lesser of two evils?

SPEAKER_01 31:03

I I tend to think, well, I mean, they're both bad. Right, right.

SPEAKER_00 31:08

And but one you have the chance of making better.

SPEAKER_01 31:11

Exactly. And so, you know, at for do no harm. So um I think the onus is on the the new technology coming in that that it needs to have appropriate guardrails, oversight, AI review boards, you know, from at the state level or at the county level, however, whatever the jurisdiction is, ought to have control and and over and be able to hold the agency, the agency is accountable for its decisions that it's making. And so it needs to be comfortable and confident that what it's basing those decisions on are uh are trustworthy and ethical. And and as a veteran community, we need to support the establishment of those guardrails and the uh willingness to be transparent, to know what's going on uh and to be accountable for what we're delivering. So review boards, mandated transparency, mandated and uh compliant reporting and ongoing auditing, uh, training of useful. I mean, uh our caseworkers already have to undergo training of what's the ethical use of the data stores that I have access to. I can't go look up my neighbor and see, you know, that's that's that's an unethical, inappropriate use of the technology. Same thing here. They they need to be trained in in the ethical and and appropriate use of of the technology and and and it has to have proper guardrails put in that that prevent inappropriate access, even right. So security around it, you know, you this this would be a data store that that you don't want, you can't let out external, you can't allow external access to. So you have to have the proper safeguards in place. How do you uh how do you ensure that that's the case? You know, we're as an industry, you know, we we need to stay compliant in those those areas too. You can't just go, oh, this is gonna be this is gonna be great. This is gonna save us X amount of hours or or money or free up these resources without the supportive infrastructure.

SPEAKER_00 33:11

Well, and I I come largely from the legal space. Yeah. Um, and legals dealt with this, particularly as a you know, AI has hallucinated cases that says, hey, you this would be the perfect case for your argument if it existed. I wish it exists, right? But but I the and maybe it's the same approach, the way it seems that the legal industry is is tackling that is saying, look, you can use AI, but you're responsible for the output. So it's uh you know, verify and verify, trust and verify, verify and trust, however you want to put it. Yeah, but ultimately, hey, if you know this seems like the right answer, and I have to make sure that how it got me to that right answer is correct, that verification is a lot quicker.

SPEAKER_03 33:50

Right.

SPEAKER_00 33:51

And and sounds like you know, in the whole idea of upstream by design and uh family connect, it's can we get to that information as quickly as possible? I would imagine that verifying things is faster than it is starting from ground zero. Absolutely, yeah.

SPEAKER_01 34:07

And and even um evaluating the outcomes, you know, or the or you're evaluating the outcomes, but evaluating the output redundancy, testing that, changing the inputs, and you know, are you getting, I mean, and just just continue, you know, a regression testing, you know, from um just ensuring that that it is still as trustworthy as you thought you implemented. Right.

SPEAKER_00 34:31

So everything we've discussed, I mean, it it seems to make sense that any agency that had its its opportunity to look at these types of things would certainly take a look at it. But like, how do these overwhelmed agencies who are are spending their entire day just dealing with their caseloads right now and don't have an extra minute in the day look to pilot these things? And how do they even how do they approach their leadership? Like what's even the process for assessing readiness for something like this?

SPEAKER_01 34:59

That's a great question. You know, state agencies are often uh handicapped by their procurement process, which is a process that is put in place to ensure fair competition, that that they're getting the best uh solution for the money, that they're not just handing it to a preferred vendor. But in so doing, they they add burden to the agency to come up with uh the um the RFP, the the request for proposal and and and uh and to manage it. But they have to get funding for that. And and so there's a there's a it's a long cycle and it's a lot, it's a large effort. It's a major lift for the agency. So starting small, the there, there are, you know, going back to that, you know, discrete uh task identification and and definition, picking um lower level, however crit all you know, critical tasks that you can test uh this out that that may be able to um be a proof of concept and and to establish trust and to build a process framework or a um a way of doing it in in bite-sized pieces that that don't put the whole operation at risk or don't uh require as much. Don't change everything overnight. Don't change everything overnight. And could we carve off this task, right? Uh apply automation to it, validate it, and immediately have an impact on the caseworkers' workload. I mean, they're drowning in work, right? The caseloads are rising, the demand of what they have to do is rising, and and uh and so there can be an immediate impact there. And so minimize the risk, and then you'll grow the expertise in internally and within the the procurement cycle and within the the legislative awareness of of how um and then you can prove the um prove uh uh prove the outcomes are or validate that that this is a a good path to be going down without choking on the big yeah. I mean, uh but also but also um there's a uh a phrase human in the loop. Uh and you know, I talked about earlier of never letting the system make decisions, choosing things where it's just taking work off their plate that they wouldn't normally have that now they don't have to do. And so you can you can prove that and you can you can then go to the next one, and you can go to the next one, and then monetarily the risk is lower. And you're not replacing the system, the underlying system in the in this uh approach. You're you're just you're augmenting or you're building the integrations into the system so that that um that the context in which you're doing the work doesn't change. The the rest of it is still is still able to be there.

SPEAKER_00 37:42

And I guess this goes to the idea of assessing readiness is yeah, pick something that we all know is redundant and painful and time consuming and maybe not mission critical on day one, but it's it's something we can test.

SPEAKER_01 37:54

Yep. And often it's below the thresholds where where you have to do a full RFP right or go through the the major uh procurement cycle that you can you you don't have to wait until the next legislative cycle for funding potential. You can be in under those thresholds and and uh and have an it and and and now you can say, hey, this we we followed this process, this worked, this had these the desired income or outcomes, and then we can carve off another piece and carve off another piece uh incrementally.

SPEAKER_00 38:22

And I imagine big picture, you know, in an ideal world, all these things are implemented and implemented prop properly. This is probably an economic savings for the use department. Yeah, massive.

SPEAKER_02 38:33

Like I would recommend leaders do a complete kind of assessment of where they're at. Where are their bottlenecks currently? Where's their data currently living? Where's the family clarity breaking down at? Where are caseworkers overloaded at? And have all those pieces available and compare that onto how the system, the technology, the work that we're doing would solve some of those issues. So when you asked about where a leader should start, like really taking a good look at their current system and understanding where all those bottlenecks and and caseworker crisis is and uh where the work's not getting done at and where the data is currently living or not living, um, I think is the best way to start.

SPEAKER_00 39:19

Well, and and it's true for every organization. You always have high value people doing low value tasks, and you know, where can you where can you clean that up? Right.

SPEAKER_01 39:29

You know, where that's a great way to put it.

SPEAKER_00 39:31

Yeah. And and and in organizations, I'm sure, in child welfare systems where you have less people, bigger caseloads, more being piled on upon you, like you're talking about, that just becomes an inevitable problem you're gonna have to deal with. Do you have any ideas of you know, big picture if this ultimately goes where it could, what this could mean savings-wise for a department?

SPEAKER_01 39:57

Yeah, I think I'm gonna defer to you, Ann. Oh, okay. I mean, uh, you know, obviously, technology, the goal would be to reduce the um the amount of licensing fees and the amount of technical overhead, and and those are the promises. But I think more importantly is um is the the human cost.

SPEAKER_02 40:17

Yeah, the um kids staying at home, not going into high-end placements, concrete care placements, earlier decisions being made where cases aren't open, and then you have the whole system cost, you know, if cases staying open for an extended period of time. I think that um the reduced length of stays, the stability of kids, the amount of administrative time uh to staff, overtime cost, um, and we could keep going of the financial impact.

SPEAKER_00 40:49

Um so it seems like a lot of this is in probably the downstream, even not clearly visible expenses that by doing the process better, by making decisions earlier, it's like, look, you're keeping kids out of the system longer than they need to be, and placing them with people who are better for them that keeps them out of a foster environment.

SPEAKER_01 41:08

Yeah, and what's the value of that? I mean, right, that's that's priceless. And and it actually has their their their dollars associated with, you know, if a if a child ends up in the in the juvenile justice system, I mean, not to mention the congregate care and the the all the all the state systems that have to then go be marshaled to support that child, if the intervention can happen earlier and the stability can be applied earlier, all of those downstream impacts are reduced greatly. And that that I think is where the value is in in considering strengthening the family system earlier.

SPEAKER_00 41:46

Yeah, and the societal value is the biggest one and all it should be.

SPEAKER_01 41:50

Yeah.

SPEAKER_00 41:50

But we're dealing with government organizations and budgets and things like that too. And that's a practicality that ultimately would have to be worked through. Yep. Um, but it sounds like it can work on both fronts.

SPEAKER_01 42:00

It can. No, I think you you can you can well, how much does it cost for congregate care for a year? Right. How much does it cost for the state to have to to support that? That that's a number, right? And so that's a large number, right? How much does it cost to continuously hire new caseworkers and train them because caseworkers are burnout and always working in crisis-driven situations and you know, moral injury of of uh for for the caseworker that that goes in with the best of intentions and and just gets ground down of I'm just not seeing the outcomes that I thought I'd be able to execute. Right.

SPEAKER_02 42:40

Um got into this to help kids and families.

SPEAKER_01 42:45

Yes.

SPEAKER_02 42:46

And I don't feel like I'm doing that.

unknown 42:47

Right.

SPEAKER_02 42:48

Yeah.

SPEAKER_01 42:49

You hear it from teachers too.

SPEAKER_02 42:50

Yeah. Yeah.

unknown 42:51

Yeah.

SPEAKER_01 42:52

It's it's across the board. I think that's where that data tsunami, the the we're all feeling like we're part of cogs in a in an inhumane system. And so it's not just uh uh a trait, an empty saying of of human-centered. Um I mean, that's that's the mission.

SPEAKER_00 43:13

It's humans trying to make decisions for vulnerable humans. Like if you don't make it human-centered, you're missing the whole point. That's great. Yeah, yeah. So given the valuable impact that we all see that these changes could have in child welfare, and that somebody listened to this podcast or or enmeshing themselves in these issues and sees the value of this, what advice would you both give? And Anne, we'll start with you about how does somebody take this back to their agencies to say, I think this is something we need to take a look at?

SPEAKER_02 43:47

Yeah, I think uh somewhat of what I said um earlier about taking a look at where your agency's at, find out where your bottlenecks are, where uh your data is currently stored, what's overwhelming your case. Workers the most? Where's your family clarity breaking down? Um, where are your permanency efforts being delayed? Having a very clear picture of where you're at in your agency, it would be a great starting point on what you need to move forward.

SPEAKER_01 44:16

As you identify those, those constraints, those, those, those places where things are those bottlenecks, then considering what discrete tasks can be identified as being, can we take this off of their plate? Can we I can we automate this chunk? Can we shift it to a more a more specialized uh person that can do that task, that's focused on that, that doesn't that I that that can take the load off the caseworker? And starting to think creatively about your budget and your legislative support or your policy support and your procurement office. That how how can we how can we pilot something? How can we test the the waters here? Because the the reality is you need to uh improve the technology, you need to to leverage the technology that's out there because it's not getting easier. It's only going to get harder and the demand is only going to be higher on your caseworkers tomorrow. And so, but but starting small, start where you can you can build the maturity in the guardrails, you can limit the potential risks, and you can you can grow as an organization as you carve this up into those manageable chunks. And that should be uh more well received than hey, we need to bring in you know the AI system that's going to help us, you know, versus you know, starting the starting small, manageable.

SPEAKER_02 45:45

And where you can get away from just making those good faith efforts to to really having concrete infrastructure that's going to really make an impact.

SPEAKER_00 45:58

When as the saying goes, every journey starts with the first step. Yeah, don't worry about 10,000 steps in front of you. Take that first step.

SPEAKER_03 46:06

Yep.

SPEAKER_00 46:07

Makes a lot of sense. Well, Ann and uh Travis, thank you so much for uh sharing all this information with us today. It's fascinating. Um, I think this makes a lot of sense that this is where the future of things are gonna go and help and promote to get us there as quickly as possible. Because again, as you said, caseworkers got into this, not uh being broiled in technology and push buttons. They got into this to help children, and and the the faster we can get them back to that being their core function, the better.

SPEAKER_03 46:36

So that's right.

SPEAKER_00 46:37

Yep. Thank you, thank you, Peter. All right. Upstream integration is not about adding complexity, it's about aligning what already exists, investigative expertise, engagement practice, and technology so leaders have usable family clarity at critical decision points. When early visibility improves, permanency accelerates, placement disruption decreases, workforce pressure stabilizes, and fiscal responsibility strengthens. CSI is working to reframe early family visibility as critical infrastructure, strengthening family systems. Thank you for joining us for this episode of the CSI Intel Brief, providing the Facts to Act. We look forward to continuing this conversation and engaging with leaders who are ready to rethink how family visibility functions upstream in their systems.