Notes From the Field: What GovCon Teams Are Really Solving For

Hello, everyone. Welcome back. If you are a part of our session, if you’re a regular attendee, welcome if this is your first time here. My name is Allison Ritz. I am senior director of product and market intelligence at Visible Thread, and we get together here every other Thursday to talk about Visible Thread goings on, the market, adoption, tech in our process, as well as what we’re seeing every day. So today, we have a special guest with us. Kyle Peterson, our VP of account strategy and customer success is here. Kyle has been doing a lot of work with our customers in the field, attending events, and getting a lot of good feedback and noticing some really interesting patterns on teams that are adopting technology, teams that are managing processes, all of the things that people are encountering in this sort of new world of technology that’s changing every day. So he is here today. We’re gonna talk through some of the things that he’s seeing, some of the ways that teams are six… Are being successful today, as well as some of the methods that they’re using to adopt and expand technology within their organizations. So I’m gonna go ahead and let Kyle introduce himself, and then we will get kicked off. Yeah. Thanks, Nelson, and thanks for having me. Really looking forward to the discussion. So just as Allison mentioned, kinda heading up our account and customer success team, the most important part of our job is understanding customer context, process, pain point, and an opportunity Where can we improve, help optimize using tech, governance, process, etcetera? So I’ve been hitting the road doing a lot of listening, doing a lot of learning, lots of observations to share, and so we’re looking forward to, the discussion. Thanks, Allison. Excellent. So let me jump in. A couple, housekeeping items. If you are not new here, you’ll know that I always give you a little bit of an overview. Again, every other Thursday, forty five minutes ish, We… Thirty to forty five minutes. We… These sessions are free. Replays after. So last time, if you missed our last session, we… It was shift left, so we talked about building compliance and clarity into the first draft. That’s downloadable, as with all of our, previous sessions. Today, of course, Kyle and I are gonna talk about, you know, what teams are really solving for, and then our next session is built for trust. So, really, we’re gonna focus on the Visible Thread platform and talk about how the platform gives teams control over AI, all the governance that’s in place, and all of the sourcing and structure and control that you have over the data that’s being fed in and being used to create your proposals and drafts. So again, connect with us on LinkedIn. You can sign up for the session if you haven’t. I encourage you to do that. We’d love to see you here more often. And then just to draw your attention to the console, so you have a few things there for download. So I wanted to draw attention to a webinar that Kyle has coming on October 27, so talking about the RFP life cycle maturity model. So it’s a framework for how GovCons move from ad hoc disconnected pursuits to real strategic capture. Additionally, there is a maturity model framework if you follow that link, so it’s a self assessment and a lot of really great information there, feedback for you, some areas where you can work on, how you can get started, and Kyle’s gonna talk through that in his session. So I wanted to make sure everybody had access to that. We have another webinar that we are advertising there, so if you’ve been on our events page, all of that information is there. But just wanted to highlight a couple of these. And then some guides, so our AI adoption guides, our buyer’s guide, and then, again, if you go to our guides and reports section on the website, there are some additional things there. I just wanted to highlight those. We’ve got some great feedback on them, and there’s some really good information in there for you if that’s something of interest. So that’s all of my housekeeping. Promise we’ll get started with the conversation. So as we’re going in, please feel free to drop questions in the Q and A panel. We are gonna leave time at the end for questions. Would love to have, you know, as interactive a conversation as we can in this format. Chapter Universal AI Adoption Patterns But I want to go ahead and get started. So, Kyle, I wanna get started with… You know, you’ve spent the last several months. You’re doing a lot of on sites with, you know, proposal capture, BD teams, solutioning teams. What has kind of surprised you the most, or what are the things that you’re seeing that are, you know, universal with the folks that you are talking to? I’ll kind of start high level and drill down from there. I’d say the vast majority of teams that I’m working with now, they have embraced AI tooling in many different forms. I think maybe this time a year ago there was lots of excitement but lots of vetting, rightful vetting of, you know, is this secure? Is this compliant? Can we use this in our processes? So now what I’m seeing is teams have adopted this technology. Now I’m saying adopted, I’m not yet saying a 100% successfully. Yeah. But they… They’ve brought it into their workflows. I think we’re still very much in the things are moving very quickly. Things are changing. So we’re in a little bit of a wait and see pattern, or I’m vetting. I’m trialing multiple solutions at once. Whether I’m looking at external kind of vendor built solutions, a lot of organizations are building their own, solutions as well in house. I mean, it’s amazing what you can do vibe coding. I I build agents now, and I’m not Mark Zuckerberg. So I’d say the the understanding and the motivation that we want to use this, we know it can help is there, solidified, and is universal. Where we’re still seeing discussion, maybe consternation depending on the org and the culture, is I know we need to use it. I know it can help, but at the moment, I either don’t yet have a repeatable, consistent playbook of where I’m bringing in this tech, why, and who’s using it. So there’s still a lot of work to be done from a process standpoint, from a governance standpoint. And basically moving beyond, I have some couple power users, early adopters, who are building their own agents and doing amazing things, but I’m not scaling And growing that… Those aptitudes and skills across the broader team. So I think that’s kind of the current challenge. Yeah. And I’ll pause there. It’s a good question. Any kind of follow ups there, Allison, or I can I can keep going, as you know? No. Yeah. For sure. No. I think that you brought up a really good point that that’s sort of what we’re seeing across the board. It’s what I’m seeing with the folks folks that I’m working with internally and, the connections I have within the industry, that everybody is sort of on board with… Chapter Challenges Transitioning Past Pilots They understand that these tools can be helpful within the process, and this can be a really valuable thing. It’s just which platforms do we use? And you brought up a really excellent point of life beyond the pilot, that so many organizations are not really successfully transitioning past the pilot phase, that they’re piloting platforms. And when you do that, who do you bring in? You know, your most advanced users, the folks that are most enthusiastic, and you get this sort of best case scenario view, and everybody’s really optimistic. And then when we go to expand that, it kinda falls flat. People don’t adopt it the way that you expect. So as that… That seems to be a universal challenge. Is that an open conversation that you feel like organizations are having? Are they aware that they’re gonna face those type of challenges beyond the pilot, or is this something that’s kind of a trial and error thing that’s happening? Yeah. I think trial and error is going to happen. Think I think the issue… A lot of orgs know kinda what the future state… What they want it to be. Like, I want this, Kyle, or any other vendor of choice. It’s getting from, I have the tech. I have my desired end state, but it’s not always a clean straight line From a to z. And so what I think this is a nice segue and and dovetail into is, you know, we just… Allison and I started the discussion talking about, you know, AI, the tech, and the tooling. Let me bring in maybe parts of the workflow use cases that I’m hearing to kinda weave into what are we attacking with tech tooling and people, whether using, you know, using AI or otherwise. Chapter Substantiating Claims and Accuracy So some common points of friction, some common processes that I’m hearing a lot about is I wanna move beyond pure just content generation. You know, generating it in, you know, a management volume, an exact summary, you know, a white paper response. That’s table stakes. It’s more so whether I’m using AI to generate strategic customer facing content or not, how can I determine is it accurate? Is it compliant? And and the best word I’ve been really using a lot lately coming straight from a customer is substantiating claims. Okay. And that’s that’s consistent whether you use AI or not to write a poll, is where is the claim coming from? Is it validated, and how can I prove that me or my business can deliver this product, service, etcetera? And so that’s very common lately, and I think it’s a rightful concern is if I have people using generative AI solutions to write this… And Allison, you’ve had great kind of thoughts, content, and expertise in governance, traceability, and also automating content generation can just accelerate dysfunction, and it can accelerate unsubstantiated claims. So I think the risk is growing, and so I’m hearing a lot of organizations say, Kyle, great content. I have my human vetters, but how can we more quickly and accurately identify unsubstantiated claims, orphan requirements, where say we’re gonna deliver it, but we don’t have any backing to it? That’s been really, really consistent, and I frankly think it’s actually a more interesting use case for tech and for people. And then the other thing I’ll mention, and I’ll pause, is there is a lot of ongoing discussion. I don’t see major consensus yet in terms of kind of where does the human stop, the machine start, or vice versa. Yeah. And I’ve come across some really interesting conversations and processes that I’m happy to share in terms of where each leg of the stool comes into play. So those are very common thoughts, themes, etcetera. And then also, finally, just, you know, getting left. How can we get started earlier? Framing requirements, customer intelligence. I’m hearing a lot more of that, beyond just initially it was, let’s get… Let’s start writing better, faster, sooner. Over. Well, and you you brought up a really great point there because I think that was the first wave. It was very… You know, this is new. This is exciting. We can generate content. And then really what you saw was instead of this huge time savings, we were just pushing that down downstream. So we’re creating the drafts more quickly, but they’re more problematic. So it’s going to take us more time to review. And then, yes, you get into the challenge of scaling that dysfunction and getting that content cycle back into a repository, and we’re building off of flawed source material. So it’s sort of this downhill, like, avalanche that happens. But I think that kind of what I want to focus on is thinking about that gap. So you spoke to the delta that people are seeing, that they understand where they want this to go or what good looks like, but they’re not really sure how to get there. So when you spoke about, yes, we want to create this content, but we really need to understand, I always say, is defensible. Every claim that you make has to be defensible. So when they’re thinking about defensibility, I know we talk a lot, or I definitely talk a lot about combination workflows. That’s really becoming an open topic of conversation, that applying tech to the right jobs and how can we create those guardrails to use AI effectively but also make sure that it doesn’t, of course, end up with false claims but… Or doesn’t just end up in, We have to back check everything that’s happening in AI so we’re not saving ourselves any time anyway. So are you seeing in practice with the customers that you’re working with and the folks you’re working with in the market, that that combination tech approach is working for them or, like, those deterministic pieces of the of the, of the workflow are solving some of those issues from a governance perspective? Chapter Mapping Mature Bid Workflows Yeah. So the the most, I’ll say, sophisticated or mature is probably the better word. The more mature organizations that I’m working with, they’ve built out process maps, generally PowerPoint, Visio, but essentially flowcharts where they’re charting their kind of entire bid proposal to… So, you know, pre award to post award life cycle, and they’re breaking up that life cycle into chunks, you know, pre RFP requirement shaping, competitive intelligence, influencing into proposal development, into estimating, into proposal reviews, you know, color teams, red team, gold team, etcetera. But they they have their process broken out into these buckets. And then maybe what’s most important and aligns with your question and your thinking is they will identify key deliverables or pieces of content within each stage, and they’ll call out which tools, processes, or people they’re using to generate it. So they acknowledge this is a layered approach. It’s not always a one tool to rule them all And say, you know, we are using ProPricer to generate our estimates and our cost roll ups and things like that. Great. When we take a price, when we take basis of estimates, when we take WBSs and kind of build them into that cost volume for the sake of argument, do we have gaps where we can kind of substantiate claims? So for anyone on the line that hasn’t done this, I’d really recommend sit down with a BD, a capture, a proposal, a contracts. That’s probably a good start, maybe subcontracts depending on your business, and map out that workflow. And and it doesn’t need to be perfectly exhaustive, but just core deliverables, core pieces of content, and what tech or tools do we use today to generate them. That visibility is really helpful, and not only can it help you identify gaps, oh, here’s something that we might automate, and it might be appropriate for a purely generative AI use case versus, you know, a pricing, a compliance matrix, SAL amendment reviews. I’m not gonna farm that out to generative AI maybe. I might use some deterministic software that I already have available. So that has been a pretty common through line from organizations that have better tech and tooling adoption because it’s it’s pinpointed, it’s targeted, and it’s following an existing process as opposed to kind of a top down, overly exuberant, great news. Use this, and we can go from, you know, blank page to full up proposal in an hour. Yeah. You could. You can, but we can’t miss kind of the existing workflow. So that’s been a a… I’d say best practice, and it’s certainly a good use of time. And that’s something bring your vendors in to the extent you’re comfortable doing that. So the reason I know all of that is I’ve kind of consulted on where I think visible thread in this case could help. And we’re not gonna we’re not gonna achieve every part of the process. We’re not a silver bullet, but there are absolutely key areas where I think we can play a role or any other kinda external vendor that you’re using. Yeah. I think that that’s an excellent point. It, kind of speaks to the same idea when we talk about, data cleanliness in that side of the house of these are the things that I want to create. Well, what’s the easiest way to understand what I need to do that you need to walk back from, What is my desired output? What do I want, and what do I need to get there? So I think we’re seeing the same thing from a process perspective. So I mean, what you’re speaking to, what you said in the front end and here is ultimately the tooling is… Chapter Prioritizing Process Over Tooling And this is my opinion, if you agree or don’t, but I think the tooling, lot of time, is secondary. The tooling is secondary to process, to governance, to really change management within an organization, how you’re going to adopt these tools. Because what you see a lot of times in folks that are falling flat or where these were really not getting the takeoff or adoption that we need is that we’re sort of bolting on tech to processes that we already have instead of building new processes and new workflows around the, you know, technology that we’re bringing in, so moving those checks and working through what that looks like long term. So when you sit down with the team, is there a specific gap that you’re seeing most often? Because a lot of times, that’s something very unglamorous. It comes down to governance. It comes down to assignment, ownership, and, you know, what those things look like because, obviously, if we’re bringing in a repository or we’re bringing in a data sourcing system or we’re bringing in a prompt library, somebody has to own that. Somebody has to have the say. Somebody has to, you know, create an update and governance structure. So is there something kind of across the board from a gap perspective? Kind of where is it falling first? Chapter Data Sourcing and Taxonomy Issues Yeah. There there is definitely a lot of consternation around data sourcing and taxonomy and Okay. Where… So not only it… Where can I find the data I need to power one or more tools or people? Once I find the data, is it good? Is it recent? Is it relevant? And so there’s still ongoing debate there and still a fair amount of pain. And I think, yeah, in kind of the exuberant rush to adopt, that stuff, it’s not always fun and it’s hard. Yeah. If you’re an organization that’s been around for a couple years and you have more than 20 people, you can have mass amounts of data spread across multiple systems, local files, SharePoint, Teams. I’m I’m hearing more and more kind of, Teams is becoming, interestingly enough, a data repository. I think it’s because it… It’s cleaner. It’s lighter. You can kinda take. I think a lot of teams are rightfully daunted by the data pool. And so… And and, ultimately, that will then determine, like, your approach. So for instance, you know, Kyle, can I point? I’ll say it just a software. It could be visible thread. It could be anything. Can I just point a tool at our entire SharePoint site? You could. I don’t recommend it. But ultimately, I’m hearing that a lot from groups where their data house isn’t highly organized. And so in that kind of desperation or lack of organization, they’re kind of focusing on a go wide motion. Or at least I need help just narrowing down content, which I could then structure and organize more efficiently. For organizations that have, I’ll say, more disciplined and more findable data repositories, pools, etc, once again, we’re seeing greater adoption of new tech and of new processes because people just aren’t having to hunt, sift, and doubt. So that’s something I’d recommend. I mean, some organizations kinda have data teams to do that now. I find a power user, find an experienced proposal writer, capture. That’s an ongoing challenge that kind of tech won’t necessarily help you on. Yeah. But I think it goes back to what content do I need to generate my work, competitive content? What is it? What is good? Then we can focus on getting it into a shareable, accessible location, and then you can bring the tech or the tooling in. And there are all all kinds of tools out there that I think kinda lend itself to that. So, you know, Visible Thread’s ability to create collections is great, repeatable, and shareable. And if you’ve got some existing content kinda sites ready to rock, you’re good to go. Yeah. So I could I could talk about that for, you know, two hours, but let me pause there. No. Absolutely. It’s really… That’s an excellent point. And I think that as we’re thinking about data preparation and structure and access and what that looks like, I think that it touches on, for me, one of the ideas that you hear most often. So why are specific things failing, or why are people not using it? And you brought up well structured data. Well, the long term or the downstream effect of well structured data is defensible content is content that users trust. A lot of times, see fall down when you’re trying to expand out of a pilot because folks are… They’re working through the system, and they feel like the results are poor. They don’t trust the claims that are being made. They don’t trust the AI that’s creating these. They feel like the quality of content isn’t where it needs to be. If So people don’t trust or they have issues with the process, then obviously, you’re gonna get less adoption. So I really want to talk… Think about what folks can do. So first, one question, then we’ll talk a bit about how, you know, how we can action off of this. So my first question is when you sit down with someone… Chapter Identifying Strategic Red Flags Because I have, you know, my idea of what mine are, what are the sort of initial red flags or things that come up for you as this is going to be an issue, or there are definitely some unresolved issues here, there are some things that we haven’t thought through. For me, it is there’s not a clear picture of what they want to achieve and process that it’s gonna take to get there. So it’s an… It’s this idea that you sit down, and it’s like, Well, we would love it to create content and do analysis and manage our systems and help our solutioning team and do… Help our legal team. So it’s all very generalized ideas of how I think this is going to work, and there are… It’s hard to action when these are just generalized sort of big picture ideas. So I think that a lot of folks have an issue. There’s a real delta between what I think I’d like it to do and actionable, achievable kind of results in a thirty, sixty, ninety day plan of what we really can do. So my kind of ask or throwing back to you is, Is there something that you’re seeing that is always kind of a red flag or always something teams should be looking out for? Like, if we’re doing these things today or if this is our situation today, that’s something we need to address. Sure. Chapter Effective Pilot Program Strategies In no particular order, I would… If you’re managing people, you’re managing a team With… You know, organizations need to be stress testing solutions. Do that. That’s not wrong. I do see a lot of teams being overwhelmed, you know, multiple pilots, multiple solutions being trialed or piloted kind of in parallel, and that can lead to poor adoption, confusion. So to the greatest extent possible, organizations should limit some overlap. There’s always gonna be some overlap, but I might have, you know, tiger team a working on this workflow with this piece of tech and piloting that. I’ve got tiger team b piloting this workflow, this piece of tech, just so that you’re not kind of saying, well, I’ve got my own brain. I’ve got Copilot. I’ve got Visible Thread. I’ve got, you know, so and so so on and so on and so on. It’s not helping your team’s morale and adoption of these solutions in whatever form they take. So specific workflows, specific tech, specific teams. The other thing I’d recommend I don’t always see is there should be a control group. Still have a control group that kind of does things it’s a bad word like the old fashioned way and then you can measure Tiger Team A against baseline Tiger Team B against baseline I’m not seeing a lot of that we’re kind of throwing the baby out with the bathwater, one. Two, I would say kind of other red flags or things that give me pause is I’ll expand upon your point there very ambitious goals that don’t withstand kind of second or third order questioning. I was just on the phone the other day. Kyle, I would love to have a solution that could give me overall oversight or insight into my contract health. Okay. Are these active contracts, past contracts? Are they with specific customers for specific products, certain contract types? What does that visibility look like? I mean, do we just need to see where they live? Are we looking for warranty or data rights or IP obligations? Kind of ultimately, what does that really look like? And when we get vague, general, ambitious goals And when you kinda start to question, it falls apart. That’s something to think about. So generally, with this thing, you wanna kinda start small to focus on targeted wins and scale up rather than taking a massive question and then drilling down. Data is a great example. You know, you don’t need a repository of a thousand of your best proposals. If you have it, great. By all means, well done. I’d like to talk to you. Five, 10, 15, 20 is a great starting point. And then maybe finally, I’m hearing a lot of a lot of this, and this is once again, this is a this is an issue if you use zero tech at all outside of Microsoft Word in a CRM, or you’re totally GenAI enabled Is kinda lack of coordination from BD capture proposals into kinda post award. And that… That’s a common statement. That’s not revolutionary. You know, Kyle, you’re a real Warren Buffett. I think it’s following the the paper trail of, you know, BD Capture has great data insights analysis, black hat type stuff, trip reports, things like that. Is it flowing where appropriate and where applicable down to our proposal and writing teams? Once our proposal and writing teams issue just an awesome, tight, competitive, differentiated response, once it leaves and goes to the customer, whether it’s won or lost, where does it end up? Putting it into buckets, a win bucket, a loss bucket, and even an overlap… A lap bucket. I hear this all the time. Kyle, I have some really great proposals that are in the loss bucket, but I don’t wanna not look at them just because they’re losses because maybe it was due to pricing or just, you know, the customer, canceled the the program or lost their funding or, you know, what have you. Is, you know, when teams aren’t acknowledging some some degree of reasonable overlap, when it’s all kinda black or white, or it just kinda disappears into the ether, just never leaves its SharePoint folder ever again. It just… It lives in that submitted folder, and you might have thousands of submitted folders. That’s something to think about, once again tying it back to getting your data lake under control to some degree. I’d say those are three things that I hear that give me pause or at least warrant further discussion. Yeah. No. That’s a… That’s an excellent point. And when you’re thinking about, okay, what those items look like… So if we’re we’re talking about data structure and where we need to reference or, you know, achievable goals and, you know, some of the other things that folks can do in the front end, is there anything… Or what would your recommendation be if they’re… I’m trying to… I’m losing my… I lost my question, but I… Because I’m thinking of too many things at once. So ultimately, if you have these problems… So if you’re an organization and the things that you’re saying right now are really resonating, so if there’s somebody on the call that this is really resonating with them, what are your next steps? So how do we drive action from that, and how do we address those things to move forward in a more successful way? So if you were in place with the customer, these are the things that you’re seeing, what would you do to help them solution that? Sure. So… And there’s gonna be some overlap from what I’ve already said, but just to be kinda thorough, and I’m gonna add in some new stuff too. So right off the bat, do we have a process workflow map available? If so, great. We want that. High level view, let’s pinpoint and surgically target what parts of the process we want to automate or make better. Rarely can we just massively overhaul an existing workflow in one fell swoop. So process and workflow visibility is number one. I’m using the word map. You might use a different word. An organization should find that. Generally, they have them. If you don’t, I recommend putting one together, and it doesn’t need to be detailed down to the last period. General concepts is a great starting point. Okay. One. Two. Once again, when I’m generating pieces of content, what data sources am I using to generate them? Don’t yet worry about where they’re at, And that might… That’s a crazy statement. Kyle, you’re out of your mind. Just what are the pieces of content that we need to generate this or to reference it or to to bump it up against each other? What is it? Then you can get into the where. You know, who’s doing this? Kinda who’s interacting? And then look for kind of, synchronicities. There’s your kind of management buzzword of the day across teams. To what extent is BD Capture collaborating with proposals, collaborating with contracts? Just look for where paper is changing hands. Great area to kind of pinpoint. We can turn that into a potential data source where I might need this. And then finally, really pinpoint within the process where am I validating or verifying claims? What data do I need to make that verification analysis, and then we can get into I’ve been doing some really cool prompting exercises with, you know, here’s a customer RFP statement of work, what have you. Here’s my response that I wrote, and then thirdly, here’s the data source that I use to help me generate my response. Hey, AI. Identify where claims are substantiated, where we see orphaned or unsubstantiated claims, and then make recommendations or improvements. Sorry, Recommend me improvements. Not actually rewriting anything yet. I’m using AI to kinda pinpoint where there might be gaps and where there’s alignment. And then if I wanna use AI to help rewrite it, I can. But to start, it’s a great way to kinda find that Without using the naked eye. And as long as the system provides some traceability, you can go a little bit deeper into the data source. So that helps with the trust and the find element. And maybe last thing while I’m on a roll is using solutions tooling to identify strong content. Once again, maybe not looking at your entire SharePoint. Once again, if you want to, you can. Looking at just specific folders to start is just pulling data at random. And that’s a dangerous word, but I’m gonna lean into it. Just pull a bunch at random and then query kind of a chatbot. Hey. I’m looking for this kind of content for this kind of customer with this kind of tone with this kind of contract type. You know, help me identify which one of these I might wanna look at to investigate further. So I’m not outsourcing the entire search and find… Or rather, it’s doing kind of the the it’s smoking things out for me. Yeah. And then me, the human, me, the expert, can then go a little bit deeper and say, yep. If I’m gonna create a new dot content library, this is where I’m gonna start as opposed to looking at 200 folders and going, shoot. How do I just break this down? So it’s a great question. I’m hoping that’s a helpful way to break it down. And just you don’t need perfection and mass improvement overnight. And then use your tools thoughtfully to advance that instead of looking for one tool to optimize or fix an entire business winning process that might influence a billion dollars a year. That’s a massive undertaking. Be careful. And I think that is… You’re spot on when you talk about using the systems, using the AI as part of the analysis on the setup. So I would say my… In my day to day life, the, you know, analysis and kind of… I call it just sort of, like, getting in the gap, showing me the thing, uncovering my blind spot, showing me the things that I don’t know, making connections that I might not make myself. When you’re working with AI, thinking about data, thinking about sourcing, you can identify, particularly in a workflow, if there are… If you’re finding gaps, ask what type of content would help improve the… This is what I’m looking to strengthen. What information do I need to give you to strengthen that position or to expand on this idea? And let the model tell you, this… If I can see this, that that will be really helpful for me. That will help me build and grow. I work with my… I work with the models a lot like that, and it helps me from an analytic and development perspective. So I think that people sometimes are stepped back from these are all the things that have to happen before I can make this successful, and I don’t think that’s necessarily the case, that it is a phased approach, and it can be collaborative with the models that you’re using. Ask because every model is a little bit different. Different models will have different… They would like prompt structured a little bit differently. They would like information presented a little bit differently, so I think that working with the model and understanding what the optimum kind of working relationship is, treat it more collaboratively, which I think people are not doing as much. It’s demand and answer. Demand and answer. For sure. Kind of one user at a time instead of thinking about things as kind of a collaborative partnership. Yes. And actually, we’re on a roll now, Allison. Before I forget, I have to get this one in. In terms of actionable use case that I really like, I’m excited about, I think if you’re not using AI to do this yet, consider it. And it’s… Chapter AI For Scenario Generation I’m gonna move away from claims entirely. If you’re nervous about, can I substantiate claims because I use Gen AI and some hodgepodge of data to create this? For the time being, use generative AI for use cases that don’t involve making claims. Great example. I have customers, Kyle, you know, I’m seeing more and more oral presentations. Use AI to help you basically generate scenarios. Hey. Chatbot LLM AI. This is the situation. This is kinda my my value proposition. I’m not making any claims. I don’t know how that was created. Here it is. You’re an evaluator. You’re an air force, contracting officer for the sake of argument. You know, based on what we’re seeing here, based on your requirements, based on my response, help me identify questions I need to be prepared to answer. Where might they challenge my numbers? Where might they challenge my past performance? You’re not generating any kind of claims. You’re just getting a second set of eyes that can scan across multiple data sources to just generate scenarios to kind of stress test or war game, things like that. I’ve seen really interesting negotiation prep prompts. Same thing. Here’s the RFP. Here’s my response. You’re, you’re an evaluator. Where are you gonna ding me on price? What are you gonna challenge me on when you’re looking at my my basis of estimate? Yeah. Great use case where you don’t have to worry about generating claims Yeah. Because you’re not making any claim. That’s a great starting point that I’m seeing more and more teams do or at least I’m recommending teams do that. I think it’s a really cool, use case that gets us out of the the prickly where is this coming from? So we don’t wanna shy away from that, but I just wanted to get that in there because I think it is very relevant. Oh, absolutely. I think that talking about different use cases… Because going into this, I think that we were more on the side of, okay, what are the things that we do manually that we want something to do? I wanna create a compliance matrix. I wanna create a first draft. I want to, you know, do a gap analysis. Great. You know, technology can assist in that. That’s fine. But that really… Especially if we’re talking about using an LLM, especially if we’re talking about some of the deep analytics, that’s not really what some of these models are best at. That’s not what the technology is best at. And really digging into approaches, looking at gaps, looking at potential risks, looking at areas where you can strengthen a position, those are really powerful use cases and things that I see less frequently than folks trying to kind of put a square peg in a round hole and create a compliance matrix. For for sure. For sure. And not everything needs AI. So once again, about the process, what… How do I generate it, and is it even AI appropriate? If not, elbow grease or deterministic software. Great. Well, I have one. We’re wrapping up. I have one other question, and then I have a q and a question. Sure. So my last question is, what is the request that you hear that from a technology perspective that maybe people are trying to do and they’re not pleased with it, but just generally, what do you think is there as far as process is concerned, as far as applicability of technology that you feel like we haven’t solved yet that you think that will be kind of the next focus or where we’re going from here? Yeah. That’s a good fraught question. I’m hearing a lot of chatter about, work breakdown structures, estimating basis of estimates, where a lot of orgs want help because it is manual, it’s labor intensive, it is risky. You have to be right. There’s very low tolerance for mispriced, miscosted, misestimated elements. But that’s generally the long pole in the tent. So a lot of orgs rightfully, man, if I could just get started estimating sooner. I’m hearing a lot of that. How can I get this over to estimating sooner before we even talk about how they’re gonna do their own jobs better faster? How can I get things to estimating sooner? And then how can I help make these really beefy labor intensive exhibits, WBSs, BOEs, how can I get them done quickly, accurately, with a little less heartache? We’re not there yet, because generally, you know, each… It’s hard to… Can use historical actuals, obviously, but they have a limit eventually. Every opportunity is different. So there’s a data challenge there versus a determinism, not everything is like for like. Yeah. And you… And also, I’d say just rightful risk aversion from cost and pricing teams Is we want help. We wanna improve this, but we really don’t wanna be wrong. Yeah. Because, whoops, we just are eating margin on this project because we got ahead of our skis. So there’s a lot of need and consternation there. We’re not there yet, but I do think that’s kind of the next frontier. And I’m doing some interesting work and having some interesting conversations on, once again, taking bite sized chunks out of those processes to get us started a little sooner. So that has been a recurring theme probably the last six months or so. Okay. And that makes a lot of sense that if we think about the areas that AI has been particularly helpful over the last several years, I say you think about the outcomes, that what is the level of risk if you were wrong, that low risk areas have been helped quite a bit. There’s a lot that we can do there. But the high risk areas, we’re still not quite there yet because even 99% isn’t accurate enough. So there is not a lot of tolerance for inaccuracy, and I don’t think… Those are the folks that still aren’t getting a lot of help because they have their very specific guidelines. I feel like when you’re using, Gen AI specifically, the more gray area, the more kind of, like, left brain you can be, the better it is. That’s it. That’s it. Like, arithmetic should not be Gen AI ed yet. No. But… So that makes a lot of sense for sure. So I have one question. Sure. And this goes back to something that we talked about previously, and I think this is kind of the universal… That speaks to sort of a universal issue that folks are having, that you mentioned, you know, we’re trialing a lot of different softwares. We’re using different platforms to see what sticks. We’re kind of seeing what this output looks like. So you get into the situation. So the question is, what happens when you’re in what seems like a pilot that never ends, and how do you get out of it? How do you actually make a decision and move forward? Yeah. So it it starts up front. And now that’s… Yeah, Kyle. You know, hindsight is twenty twenty, man. Yeah. But I… It’s still relevant to just… Let’s lay out ideal, then let’s talk about if we’re in it, how do we arrest things? So, ultimately, the best pilots I’ve been a part of where we haven’t needed to pull the rip cord is customers saying, Kyle, these are the exact deliverables outcomes that I’m looking to achieve, and they are targeted. They are specific, and they’re tied to specific people. Okay. That is not a silver bullet, but that’s gonna avoid a lot of that consternation because generally, think about like a search and destroy mission. That’s where these pilots get a little bit haywire. Is I have this vague idea of where I wanna go when I’m not quite sure we’ll figure out as we go. And that’s where these pilots can drag and drag and drag. Is well, are we seeing ROI? Are we doing what we thought we would? Well, we didn’t really have a firm target in the first place. We’re gonna keep trying. So really specific upfront deliverable goals. That’s what you wanna start with. Also, having timelines, having kind of yardsticks in the sand. So this might be a two month pilot. One, do we have a reason for two? Do we need three? Is, Oh, whoops, we’re gonna do a three month pilot, but the middle chunk of the pilot is in August September, right at the end of the government fiscal year and the entire org is heads down. So be thoughtful on timing. When are we gonna do the pilot? I love pilots. Honestly, October. And even end of year’s crazy, so bear with me. But I do kinda like November, December if you can carve out some time so that you can get into 2027 now. How crazy is it that we’re already looking to 2027 without kinda getting to January and thinking, gee, we gotta really start, kicking the tires on this. So thoughtful timing, bounded timing. Hey. By month one, what should we have achieved? Just kinda gives you, like, those buoys that you can you can tie to. Too many orgs just don’t have that. And then finally, clear up expectations up front of our piloters. What’s expected of you? What do we need from you? I’ve seen pilots where, alright. We’re gonna have a user session, and we’re gonna attack a specific use case with specific people. Oh, I I have a I have a proposal effort. Can’t join. Busy. Oh, I’m in the car. So, you know, we need to make sure that users know what’s expected of them And they do show up ready to rock. When it’s kinda seen as join if you can Yeah. People might not always join if they can or can’t. Yeah. So that doesn’t help in terms of this is what I want you… Me… This is what I need from you, and this is the this is the the rough time estimate that this is going to require. That visibility is just as key to manage expectations as the what are we going to pilot. So I could… I mean, I could talk about that once again for an hour, but for the sake of time, pause there. Absolutely. And I think that it speaks to, you know, it’s fascinating that we’re talking about AI and we talk about new technology and we are talking about what are the challenges. They are the same challenges that were there before. They’re the same challenges that have been there forever. It’s… You know, the people and the process are so important in this. And even if the tool is the most amazing thing in the world, if you’re not ready for it, if you don’t have the bandwidth, if you don’t know what success looks like, you are likely going to have a rocky time. So I think that, you know, focusing back on the basics, focusing back on the animal that is human behavior is a big part of the process, so I appreciate that. Absolutely. So, everyone, we’re actually a couple minutes over. Thank you so much for being here with us. And, again, like I said, register for the series every other Thursday. Thank you so much, Kyle, for joining us. And, you know, Kyle talked a lot about… Lot today about what you can do about process. Sign… You can sign up for his webinar coming up here shortly talking about process maturity frameworks, and then, like I said, the worksheet is downloadable in the console as well if you wanna get an assessment of where you stand today and some of the things that you can focus on. Yeah. Allison, actually, on that link, is that in the email? Is that in the slides they can click on? I’m just… I just wanna keep pushing. It’s it’s in the console here. So on their screen, there’s a downloadable link. Chapter Follow Up Email Resources But once we get off of the call, everybody will get a follow-up email, and those links will be in the follow-up email as well. Beautiful. I don’t I don’t have the the snazzy console, so that makes sense. Thank you. Yeah. That’s okay. I didn’t… I don’t even think you need to download your own link because this… No. Probably not. I have it favorited. But… Okay. Alright. Well, thank you everybody again. Thank you so much, Kyle, and hopefully we will see you in two weeks to talk about the VT platform and governance. Everybody have a great rest of your day and a great weekend. Thanks. Bye.