Then Is Now: How a Black Technologist Is Using AI to Reclaim What History Tried to Erase

Want to see your PDFs think like a database and your chatbots answer with context, not guesses? We sit down with technologist and entrepreneur Max Riggsbee Jr., co-founder of Gadget Software, to unpack how compute-ready documents—what he calls semantic twins—turn unstructured content into structured, queryable knowledge that both humans and machines can trust. Max explains why simply chunking PDFs into a chatbot loses meaning, and how deep descriptors, QA pairs, and entity maps let you navigate ideas, not just pages.

We go inside directed AI, where you choose the exact slices of content a model can touch, then generate summaries, outlines, or tables grounded in that selection. Max shares results from work with Signal65, Dell, and Broadcom showing fewer hallucinations, faster token throughput, and better energy use when AI is fed structured, contextual data. From there, we get practical about agentic workflows: the validator checks you need before any output ships downstream, and why human-in-the-loop review still matters. Max’s “Georgia” test—person, state, country, or font—reveals how ambiguity explodes without metadata. He also breaks down a real failure in a political cartoon pipeline where an agent inferred a story from SEO slugs instead of reading the article, and how guardrails caught it.

Our conversation widens to legacy and Black history. Grant traces family records from enslavement to land ownership, underscoring how caricatures like Jim Crow distort truth when they calcify into the story we’re told. Maxwell introduces Then Is Now, the audio diary project he runs with his 90-year-old father, using authentic voice recordings and AI to frame the surrounding historical context. We talk about scanning non-digital originals like the Pentagon Papers as a stepping stone to microfiche, county archives, and the overlooked documents that can restore names, places, and property to the record. On the technical edge, Max shows how rich textual descriptions can stand in for heavy images, enabling vision models to re-render diagrams on demand, making insights lighter and more scalable.

If you work with unstructured data, lead AI projects, or care about preserving the story with accuracy, this conversation gives you a roadmap: structure your sources, validate your agents, and keep humans in charge of meaning. Subscribe, share this episode with a friend who needs better answers from their data, and leave a review to help others find the show.

Thanks for tuning in to this episode of Follow The Brand! We hope you enjoyed learning about the latest trends and strategies in Personal Branding, Business and Career Development, Financial Empowerment, Technology Innovation, and Executive Presence. To keep up with the latest insights and updates, visit 5starbdm.com
.

And don’t miss Grant McGaugh’s new book, First Light — a powerful guide to igniting your purpose and building a BRAVE brand that stands out in a changing world. - https://5starbdm.com/brave-masterclass/

See you next time on Follow The Brand!

Transcript

Meet Max Rigsby And His Focus

SPEAKER_00 0:02

Hello, everybody. Welcome to the Fire Man Podcast. This is your host, Grant McGall, and I'm going to welcome everybody to this particular episode. I am so happy to have my guest here, Mr. Max Rigsby. He has a story background. We were sharing notes. This is Black History Month. And these are the type of things that we want to talk about. And I want to bring something out right now. I don't see a lot of black technologists. It's almost like the unicorn, but they're out there. They're out there. And when you come across them, you'll find out how fantastic they truly are. And I also want to speak to my young, my young professionals, especially in the black community, because they say, well, you know, Black History Month, yeah, but they're always talking about something that happened 100 years ago, 150 years ago, you know, you know, 70 years ago. Well, guess what? Max is alive right now. And he's been doing some very, very phenomenal things. And him and his his father has been doing some phenomenal things of just getting the research done. But he comes from a very long line of individuals that has culminated in his presence. We're going to first talk about some of the things that he's working on, get a good understanding of his story, and then we'll go deeper in into our discussion because it is Black History Monk. So, Max, you'd like to introduce yourself.

SPEAKER_01 1:27

Grant, it is a pleasure to be here. Um, I'm super excited. Uh, you can see I'm wearing a little blakey today, right? I love jazz, right? Um, and I certainly I certainly love the the great heroes of it for sure. Definitely excited about the discussion we're about to have.

SPEAKER_00 1:43

Well, let's jump in about some of the things you're doing right now. You share it with me. You demonstrated to me some of these compute-ready documents, what you call like semantic twins. I don't think a lot of people understand. Now you're in a certain area of uh artificial intelligence and and technology and business technology, business intelligence, but you are all the way up the stack. I want you to help us understand, if you would could, the simplest before and after story that you can use to explain the leap to it to an enterprise buyer around what you do.

What Are Semantic Twins

SPEAKER_01 2:19

Sure, absolutely. Thank, thank, thank, thank you for the question. So I am the co-founder of Gadget Software. We are based in Hackensack, New Jersey. Um, we've been around uh 10, 12 years, um, and our principal focus over all that time has been making what's referred to as unstructured information, think of PDF documents, think of mail, um, actually make that kind of content operate the way information operates in a spreadsheet. Okay. And to accomplish that, we use a lot of AI to read in content and to organize it into what you described as semantic twins or compute ready documents. So, what are those? Um today, and this is possibly true with many of the members looking at this audience, if they want to use, say, uh a chatbot, they may take a PDF file or several PDF files, uh, chop them up in a variety of different ways and then feed them on. Um, the there are challenges with doing that. Um, and and some of it is context is lost, right? There's a lot of context that can be lost there. Context that we, as readers of content, always have available. I mean, think about having um, say, a scholarly book and it didn't have, say, a table of contents, or maybe it didn't have page numbers, or maybe it didn't have the right sort of uh pieces to help you understand where do I find exactly what I'm looking for? What we do is we take that content and we organize it into components, and each of those components is deeply, deeply, deeply described and then organized into a single document that we call a semantic twin. So think of it as a document that's literally built for machines with all the instructive information, all the nuance, questions and answers, keywords, all types of semantic data, but it is specific to help chatbots, but not just chatbots. We we our data can actually be read into business intelligence tools. We do it all the time.

SPEAKER_00 4:22

You can visualize and navigate it and you can use it into any talking about is so important because I think like when you're building out a um you know a pipeline and you think about well, is this about ingestion of data? Is it topic segmentation? Is it metadata synthesis? Is it you know an evaluation, QA? I mean, what reveals exactly, you know, the molt that you have right now, your operational focus? Because people I can still see, like, well, what is you you're taking a PDF or something like that, and then you're you're you're turning into machine language so that becomes easier to ingest. Help us understand that more.

From PDFs To Compute-Ready Documents

Directed AI And BI Integration

Validated Results With Industry Partners

SPEAKER_01 5:02

Yeah, not just not just easier to ingest, but also easier for us to navigate as well. So imagine for a moment that you had um a thousand documents, right? You have a thousand documents. Um, and you want to understand like what are the common intersections in those thousand documents? Are there people in these documents? Is information classified in a variety of different ways? Are there organizations in here, right? Um is can any of this information be summarized? We do all of that work. We do all of that work so that when you and I do it all the time, you can literally take a corpus of material that's been turned into compute ready documents and read them into Power BI and build a complete visualization and navigation that will allow you to instantly see oh man, look at look at the common words that that these documents all share, or the common concepts. Or wait, look at all the different places that they're mentioning. Now all of a sudden it's quite visual. Um, we also have a mechanism through which we can integrate a kind of artificial intelligence inside of that so that when I find the pieces that I'm interested in, I can direct it at AI to say, you know what, give me a summary or an outline, but just on the pieces I've selected, right? We call that directed AI. That's very different from when you're using a chat bot and you're asking it a question, and then it goes out and it finds the pieces and it pulls them back. In directed AI, you get to be in control of precisely what it's going to use. Um, and we're seeing situations now where companies want to use these things in tandem. I want to be able to see what's in my large content set. I want to be able to navigate it, imagine being in a corporation, but then sometimes I want to use a conversational AI to organize that information, say for output, maybe to organize it into a report or organize it into a table, you know, or you know, who knows, do do a variety of different things. And you can really only do this if you come up with a common data structure, which is what we do. And the good news, the really good news, what we're excited about is the structure of compute ready documents and semantic twins. We've actually had this validated. Um, we we we basically got involved in a um in a uh partnership with Signal65 Dell and Broadcom, where we actually took our technology into an AI lab in Colorado and worked with them. There are white papers on our website about it, there are white papers on theirs, talking about how this is a massive differentiator in making AI not only work more effectively, hallucinate less, you know, better, better energy footprint, lower, you know, faster token throughput, all those kinds of attributes that we think about. Our technology actually makes them all more efficient and at the same time allows you to have a better understanding of what your large corpus of data is all about.

Designing Validator Agents And Checks

SPEAKER_00 8:10

Let's take that a step further. Because that now you people start to say, well, what is it? Then you've actually why. What is the benefit that you get out of it? And so we want to get into this question here. So if you could design what I call the ideal validator agent, you know, for agentic workflows. That's kind of downstream for where you're working with. So what are, in your opinion, what are the three checks that it must run before anything ships downstream, right? Right. So I know there's a lot of assumptions, there's a lot of reads, but you know, connect this directly to how AI works. Okay.

Context Failures And The “Georgia” Example

SPEAKER_01 8:55

So it actually is really funny you say this because I just had a very interesting AgenTic workflow pipeline breakdown recently. Not in my, not in my technology specifically, but in some tests that I run. So let's let's let's talk about that. So one of the challenges in um any form of an AgenTech and autonomous flow is you have to have certainty that the inputs and outputs are correct and that the handoffs are accurate. Way we do this in our technology, right? We use a lot of AI, is when we are generating the art, what we call the artifacts or the metadata, right, uh uh around um for for uh compute ready documents, we actually have validators. So for example, if I've got a piece of context, right, and I'm generating QA pairs, then I have a validator that says, well, hold on a second, I'm gonna take a look at that context and I'm gonna look at these outputs, and I'm gonna score whether it's relevant, I'm gonna score whether it's you know accurate, right? And depending on what that scoring is, right, we can make a decision whether we think we should keep it or do it again. Okay, so the notion of having a validator, particularly when you're either ingesting or more importantly, outputting information is absolutely critical. So um, in terms of your agentic workflow, you have to have a way of saying, Hey, do I have the right piece of data? Well, some of the ways you do that is to grab some of the surrounding metadata, right? Does it have the right attributes, right? You know, let me grab a summary or let me grab a collection of keywords or let me grab some classifications to make sure, for example, I'm talking about one of my one of my favorite examples. If you want to drive an AI crazy, you can ask it questions about the word Georgia. Okay. And it doesn't really know what to do, right? Because Georgia can be the name of a person, the name of a state, or the name of a country. And so without context around Georgia, it actually doesn't know what you're talking about. Right. But if I gave it Georgia, and then I gave it some descriptors around it that suggested it's a place. I'm gonna no longer talk about a person, right? I know it's not a person named Georgia. And then if I have some additional information that suggests that it's say near Russia, then I know it's a country. But that's all contextual relevance. That's what we provide around, sort of around that piece. And so from an agentic perspective, you want to be able to understand the descriptors around the data you're looking at. And those descriptors provide you the clues to say I've retrieved the right piece, right? And now I'm gonna do whatever work you've asked me to do on that information. Hopefully that makes that makes sense.

Agentic Pipelines And Human Guardrails

SPEAKER_00 11:47

Context is is very uh important. I always talk about a context-aware environment. We're trying to get there. How the internet's going to change, it becomes a context-aware environment, meaning so it does understand what it is that you're talking about. But human language is not that simple. You just talk English. Lots of words that that sound the same, maybe even spell the same, but mean something completely different. So you you you you're asking the machine to uh to to decipher that and and come up with uh the right answers. And that's how you get into what we call drift and and the the AI or you call it hallucinations. When you start having this, and I I I'm warning people about this when they get into addictive systems, you're taking something, you're used to a deterministic world, meaning one plus one equals two in a binary system. It's never gonna equal more than that. It's always so it's very, very stable when it comes to that. Where if you have a probabilistic system, AI is probabilistic. That means it's probabilities. So, like one plus one, I've used this analogy and I and I use it again. I am one, Max is one, there's one and one, but together we make three. It's groups, so it's different, so it's a different probability of what I might say and what he might answer next time. So it's not an exact science, is what I'm getting to. So, in an Egyptian system, if you're dependent on it to have uh a similar outcome or or uh result, you might, you know, from handoff to handoff, you get more and more drift from where you're at. All of a sudden, you were talking about a woman in Georgia or named Georgia. Now you're talking about a woman named Georgia that's in Georgia, in Russia. You know, I mean you can you can really get off track, is what I'm saying. I don't know if we completely have solved that yet. Maybe with um these different formulations from the LLM, I would say, you know, and then we start getting into these image AI or spatial AI, maybe that'll help more. I'm not sure. What's your thought on that?

Political Cartoon Workflow Breakdown

Correcting Assumptions And SEO Drift

SPEAKER_01 13:54

It's a good question, right? I mean, a lot of this is about contextual relevance. Actually, as you were saying that, I forgot to mention there's a fourth Georgia. That's typography. So, so, you know, um, you know, and I love that personal example because, you know, if I were to throw the word Georgia out, you could come back with an answer, but without the context, you really don't know what I'm talking about. In terms of um what all this means in terms of the world of agentic and so forth, we need a lot of checks and balances. Um, we definitely need human in the loop. Um, and we we need to build guardrails around these systems. I had a situation last week, so you know, and my goal here is not to bring in too much politics, but I'm gonna I'm gonna bring up this one particular issue because one of the things I do is on a daily basis, outside of my platform, I test um I test AI systems every single day, and I do it through a mechanism of political cartooning. Now, on the one hand, I mean, it may look to people, oh, you know, I'm just creating these political cartoons, but actually that's not really the whole reason. The reason why I create political cartoons is it requires AI to read information and then organize output in a manner that's either dealing with sarcasm or it's dealing with some type of point that needs to be made or some type of issue that needs to be raised. I work a tandem pipeline on this. I have two agents. One agent serves as a researcher, the other agent serves as an illustrator. I don't do it in one, I do it in two. And the reason is because I need to test does the researcher understand the nuance of the political cartoon it needs to create? And can it hand off a prompt that gets the illustrator to envision that? I run these every day. If you, you know, if you go to my LinkedIn page, it's just filled with these political cartoons I do all the time, right? Last week I ran into a very interesting problem. Um, I was dealing with the Medgar Evers problem with the National Park Service. So last week they decided, hey, we're gonna pull out all the pamphlets and you know, we're not gonna talk about racism anymore, and we're really not gonna talk about how Medgar Ever designed, and we're really not gonna talk about you know what happened to him. We're gonna change all that, right? We're gonna change the history. Um, and so I, okay, we're gonna we're gonna build a political cartoon, but the cartoon was wildly wrong. It was just wildly, wildly, wildly wrong. And and and and and what it was, it was somewhat suggesting that there was an apology um, you know, coming from coming from 1600 Pennsylvania Avenue and and basically saying don't call me a racist. And I was like, well, this isn't even in the story. Like, I don't understand where this came from. And so I ended up in a conversation with uh the researcher, right? So the researcher is handing off information to the illustrator, and the illustrator drew this wrong picture. And what it told me was it actually never read the article. What it told me was it made an assumption and it made an assumption based on values in the slug in the URL. Hold on. So it made a collection of assumptions about nine words that are really designed to be picked up in SEO, and it used that to create an entire story, and then, hey, we're gonna then create the prompt and we're gonna tell the illustrator to draw this picture. And the picture's wrong. Now, now on the surface, it doesn't sound like much, right? And you know, I looked at it and went, no, we're not gonna do that. And I asked it, you know, some questions, and it's like, no, I actually didn't, I didn't read it. Hold on, I'm gonna read it now. And then it read it and went, oh, guess what? My bad, my bad, I got it wrong. Here, you know, all right, you got it wrong. It's the cartoon writer's like, yeah, no, the cartoon's wrong too, you know. So create a new one, and then it created the correct one. But think about this for a second. This little innocuous workflow that I run, right, is all agentic. I don't actually tell it how to draw a cartoon and I don't tell it how to interpret the information either, right? I have a very simple prompt that tells it it's gonna create a political cartoon based on 19th century cartooning. That's kind of it, right? And then my job as the publisher and the editor of the output is to review it to make sure, yeah, yeah, yeah, this is good, we're gonna go publish it. But think about this in a tandem workflow. So, what happens if that first that first uh agent in the pipeline misinterprets medical information? Absolutely, right? And then it's handing that off to another agent to you see, you right. So you see, you know, something as simple as a cartoon, it's suddenly not that simple. It's suddenly not that simple, right?

SPEAKER_00 18:57

Here we're at, Max. Now I gotta bring because you know it we gotta speak to this.

SPEAKER_01 19:02

Yeah, yeah.

Family History And Reclaiming Narrative

SPEAKER_00 19:03

Especially during Black History Month. Now you do you're doing a your family's done a fantastic project. We want to talk about that. I have done my own project doing some research around, hey, where did the McGall name come from? Number one. Brought me all the way back to seven grandfathers, all the way from the European uh McGall that came all the way from Scotland, came to uh Virginia, fought in the Revolutionary War, they won the war, and then he was gifted land out in uh Virginia, and then he acquired the services of said, we call him Joe, you know, became my great-great-grandfather, very true story. But here's the deal: as you go through Joe's experience from Virginia to Tennessee, and then finally to Ray County, Missouri, where he then has a son named Green Magal, who's my grandfather's grandfather, who became the first person to actually be enlisted on the U.S. Census in 1870, who was born into slavery but ended up being a landowner in Oklahoma. I bring this up when you have context around history and you know the true narratives around it, and we start talking about things like we talked about earlier, Jim Crow. Jim Crow is not a correct narrative of the African American experience in the United States, it is a caricature that was created by a whole nother population of people for the purpose of oppression.

SPEAKER_01 20:35

That's right.

SPEAKER_00 20:35

And then when that becomes reality, it becomes cascaded over hundred, a hundred, two hundred years. People think that, well, that was the way it was, that was real. Like there's nothing real about it. That was someone's opinion that got then printed into history. What are you thinking about that?

Then Is Now: Audio Diaries With AI

Legacy, Endurance, And Telling Our Stories

SPEAKER_01 20:53

Well, I think about that all the time. Um, so I have a uh project that I run with my father. He's 90, um, called um then isnow.substack.com. That's where it is. It's public. Um, and and part of the reason it's public is because um he and I decided that we felt it was very important. Somebody's got to tell the history of what was it like growing up, you know, during during the legal definition of what was called Jim Crow. Um, but in order to understand him, we had to go way back in history. And we had to get like you, right? You know, I can actually put my my great great great grandparents, right, on plantations in North Carolina, right? You know, and I have, you know, a great, great, great, one of those greats, right? Um, you know, grandfather that actually testified at the trial of a governor who was being impeached in North Carolina. And he talks about um how he was attacked literally by the Klan. I mean, it's it's it's public record. The man the man talks about this in public record, and he talks about it in public. And so, you know, when when when we begin to understand these things, what's become very clear to me, whether it's my father or the people that were around him or the people around them, right, from a community point of view, you know, they were striving for something. They were always striving for something, right? However, they were doing it in the communities that they were living in. Um, and that creates sort of this interesting counter challenge for us, right? Because we're not striving to get somewhere. We're actually striving not to be pulled somewhere right now. And that is why my father wanted this diary that Nintri that he does called Then Is Now, because he wants people to understand the path that we're being towards is actually the path I was walking away from. I'm seeing, I'm I'm I'm hearing phrases, I've heard these before, I'm seeing activities, I've seen these before. I've grown, I grew up with this and grew out of this. Right. And and I see that we're we're you know, we're slip sliding back into it, right? And and it's it does get interesting to me um when I start to work with him on these audio diaries, and we use AI to embellish aspects of his audio diary. The audio portion is always true, it's authentic. It is him. He's telling his story, right? That's the authentic part of it. We use AI to help frame the broader um historical narrative around him, right? So that's his story, but let's tell you the story of what was happening around his story, right? Fill in, fill in the blanks. Here's what's gotten interesting, Grant. My father and I we have two different, we have two different sort of views on what this project is about. For me, it's always about his audio diary. Like, oh, that happened. That's not what it is for him. What it is for him is we have AI podcasters, we use Notebook LM for this, and he's always interested in what they are saying because he wants to know, well, what was happening around me? Yeah, right, the history I wasn't really paying attention to, right? I didn't really see because I was too busy living my life. And he's basically telling me now, like I'm beginning to understand things about what was happening to me that I didn't see at the time, which is fascinating. So you're you're 100% right. You know, this this notion of especially in Black History Month, but you know, maybe maybe we should always just be Black History Year. Maybe, maybe we shouldn't even make it a month. It's just now it's the 2026 edition. Um, and and we should just we all collectively need to understand as best we can our personal stories, our family stories. And then we want to understand that story in the context of the broader history because it is very grounding. Very, very grounding, and it also it's a legacy story. You know, you and I talked about this before, right? You know, you're sharing with your children, right what's behind you. Yeah, right. These are the people that, you know, the I mean, look at this, right? I mean, look at what folks really fought for, right? And look at what they endured. Endured is an important term here, right? They didn't just roll over, they endured it, right? And that's why I'm here, and that's why you're here. And we all need to respect that, but more important, we need to understand it.

Bias In AI And Max’s Career Path

SPEAKER_00 25:37

Yeah, and I think right now, as we are on our own media, right? Follow the brand podcast, that is, I think is a siren call to all of us. We have to tell our story. You cannot depend on another group of people to tell your story, they're gonna tell whatever story they want. You have to have your story. Back hundreds yeah, I called Joe McGall, he couldn't tell his story. It was illegal to educate black people at the time. Illegal. Think about that. Illegal to educate anyone. Uh um, and then and then to grow out of that. We are the American story when it comes from an underdog perspective. What are we doing? We are definitely fighting for freedom. We're fighting for freedom every single day for the right to to exist, to be who we are. We are the original Americans, if you really look at it. We didn't, you know, we didn't come over to the 1920s on big boats from Europe. No, no, it was a little different journey. A little bit of a different journey, yeah. A little bit of a different journey. But we're here, and because we have things like uh our AI tools, we can do deep research very quickly. There are so many micro archives that show where your family was, or you can get me in my case, I had a very uh easy to trace last name, so it wasn't that hard for me to get a lot of information, and you start thinking about well, who acquired what? And you can start seeing these acquisitions over time, even through slavery and whatnot, even to the boats that came over from Africa, they had records. So then you can do some research around that type of stuff. I don't want to get too far deep into that, but I want to bring it back to you because your story is powerful in the fact that I just stated earlier, you are African American and a technologist, high level. Most people, when they think of technologists today, they get a mental image of a potentially a younger white guy who's kind of geeky living, you know, inside of a data uh center for the most part. However, that image is not correct. So getting back to our identical box series, and if I tell AI to draw me a uh technologist, most likely it's not gonna draw Max Rigsby because he makes it very, very prominent that he is who he is. How did you get into this space?

The Hard Problem Of Unstructured Data

Testing On Scanned History Like Pentagon Papers

SPEAKER_01 27:58

That that AI bias is real. So I have been uh wow, so I've been in the AI, not the AI space, I've been in the data space for very, very long time. Um, I've worked in financial services, I've worked in Silicon Valley, I've been uh a chief technology officer, I've been a general manager, I've done consulting, uh, and I've been uh I've been involved in early startups, um, many several actually. Um and and it's mostly been around data, um, whether how data is stored or how data is retrieved, that's really been the area, that's really been my area of focus. And because of that, the one thing that I actually learned is that everything goes to ground. No matter what you're working with, it all goes to data. If you have no data, nothing works. I mean, look at the large language models. The large language models don't work without data. They don't work without data, right? That's that's that's what they eat, sleep, and breathe, right? And so for me, as a black technologist, in particular in the area of of data technology, that's both networking and storage, um, and now actually involved in schematic structures, right? Um, that's where I've built my expertise across a number of different industries. One other thing I'll say is I've been working in AI before, well I won't call it generative. I've been working in natural language processing before the whole generative thing even took off, right? And so for me, I've I've been doing this for a long, long time. And and it it's the kind of work that a lot of people don't think about. They don't think about unstructured data. They don't think about documents, they don't think about these things. Um, what I will tell you is it's extremely difficult work. It is very, very hard. Um, there really are no two documents that look alike. Um, they have strange behaviors. Um, they come in different timelines. Um, and so what I do sometimes just to test our systems. I recently, you know, you know, my the engineers will say, you know, go find us like hard stuff, like go find hard stuff, right? And sort of like, all right. So I recently went and got the Pentagon papers. Like, now the Pentagon papers are all scanned. There's no there's no PDF as we think about it. That stuff didn't exist, right? How does the platform do? How's it how does our system work? Can we really get the history out of this? Can we understand it? Um, can we can we extract information in such a way and describe it that I can give it back to another AI and it can draw it? Like this is literally the kind of stuff that we do and that I do all day long. Um, you know, either with my team of analysts or I do it with my engineers. Um and the beautiful part, and I'll I'll I'll tell you something that's really before you go forward.

SPEAKER_00 30:52

Yeah, I want you to, because I I like what you just said that you said about scanning the Pentagon papers. Yeah. How about scanning the microfetch of these county records that had all these transactions of African people? That's what I'm working towards.

Toward Microfiche And County Records

SPEAKER_01 31:11

So, so so um it's a strange dynamic when you when you work with very complex unstructured data systems because you weirdly work backwards. So you start with less complex, that's maybe um XML structure. So one thing you should know, I probably have you're probably this is crazy, but you're probably looking at uh a person in a company that has possibly one of the largest collections of federal regulatory information in the country. Um, and I have 25 years of it, and the last five years I've converted into these compute ready documents. We make some of this stuff available on our website. You can literally navigate through complex regulatory data in Power BI. Right, in Power BI. So we start with something like that, then we start working backwards. Okay, let's start to work with PDFs, okay, modern PDFs. What happens to a PDF that gets closer to the birth of the PDF, which is say in the early 90s? What happens if we get beyond that and it's all scanned in, right? Because what we have to begin to understand are the attributes in the files. It's crazy. I know people don't no one ever think about this, right? You just sort of flipping through stuff. But when we come at it from a machine perspective, there's a lot of stuff going on in there that's like, I wasn't expecting that. Here's one, and everybody's done this, and nobody really understood it, but you've done this, you've gotten a PDF and you've um you've got a copy, yeah, and then you paste it, and it's like, what is that? Right? Like a bunch of junk. Okay. So, so, so, so what's what's happened often in that is that there may be what's called a glyph. There's there's some type of of um uh you know uh letter structure, right? But actually in the PDF, it didn't know what it was. It didn't know that, oh, this version of an I actually maps to this version of an I. So I don't know how to connect the dots between the thing that's on the on the on the screen, if you will, and what character it actually is. And so I'm just gonna give you junk back. I'm just gonna give you junk back, right? I go through all of that stuff, right? We figure all that kind of stuff out, right? And then how do we make it coherent so that when you're working with not just 10 documents, but thousands, and oh, by the way, of they're all mixed. So to get back to your microfiche story, we have to be very clear that we can actually ingest something that does not have a digital origin, which is why I went for the Pentagon papers, right? Doesn't have a digital origin. We discovered some things, we learned a few things, like in ingesting that. It's like, okay, these kinds of strange artifacts can appear. We need to know how to adjust for them, but then you're on the move towards things like microfiche. You're on to move to to even less um refined content, right? Yeah, because the idea is to be able to get this inside of machines effectively, right? So that it can be used and merged, right?

SPEAKER_00 34:31

Let me ask you this because one thing that intrigued me, and it goes back to my my grandfather's grandfather, his name was Green McGall. And the reason why that is intriguing, because we have an image of him, they took a camera photo of him when he uh achieved homestead status in Oklahoma. That was part of the process that takes a picture of him. So we have an actual image of what he looked like, and he was smiling. Yeah, he was happy, right? So this had to be around the 1890s. This is before right before Oklahoma became a state. So around the 1890s, at that time, he was about 50 years old. Uh, but we have an image. So imagery, when I was talking about spatial AI and things like that, there's probably, and I I think photography became uh of use in what the 1860s, 70s, or something like that.

Images, Descriptions, And Vision-Language

SPEAKER_01 35:37

Yeah, and then by the by the by the end of the 1800s, we have the Kodak camera, right? The first idea of the notion of something that's a little more instant where you didn't have to set everything up. So yeah, he's right in that zone, right? Yeah, yeah. Right in that zone. Um, so images are very interesting to me, um, not just in terms of the the output, but language models have very interesting characteristics. And when you work with vision models and language models together, you can do some really, really interesting things. So one of the things that we we we we are working with right now, um we ingest, um, let's just say that we're ingesting some type of document. Right now we're we're working with flight flight manuals. And flight manuals have a lot of imagery um and a lot of diagrams. Here's the thing I don't actually want to carry the weight of an image. Um, you know, images, if you look at documents and you look at the size of documents, right? Sometimes I can tell I can I I can tell by the size of a document that it has a substantial amount of imagery in it. Because PNG files, these these things are heavy. Text is very light, but images are not. Um we don't want to carry all that weight. And so what we do is we get into uh prompt descriptive aspects of images, right? We we we want to read that content, we want to keep it in context, and we want to get into very verbose descriptions of the imagery. Why? This is one of the things I'm testing right now. So let's say that you are reading in a piece of context, right? And let's say it's out of a manual, and in the manual, it is as if there's a picture of a plane, and maybe the picture of the plane is telling you, like, you know, uh how the wind, how air travels. Well, I don't want to carry that image, but I don't need to carry that image. If I have an effective description, I can actually hand it off to a model that draws images, and he just creates it for me.

SPEAKER_00 37:44

Yeah, you can get around and what he's talking about there because anybody's known that like you have video, he gets into the gigabytes of information, pictures well, get into the tens of megabytes where text and audio is very light. Audio is very, very light when it comes to bandwidth and how much it takes. We gotta wrap up, but I want you to answer this in a way that I think you'll you'll like. I want you to put act as if you are now talking to your grandchild, who's now maybe maybe he's 15 to 20 years old, and you you have to explain to him your life, not just your life, but the life and times in which you lived. How you maybe were younger during the civil rights. I'm not sure that your age, but I remember how old I was going through civil rights. What was that like? I remember McGovern. I remember running for president. I I I remember a lot of different things from the 60s, 70s all the way up. How would you describe your life and to your point and what was around you up till now?

Advice To The Next Generation

SPEAKER_01 38:57

I'm actually driven by one fundamental idea, curiosity. Um, curiosity drives nearly everything that I do in my life. I am curious about how people think, I'm curious about how things work. I'm curious about if you do A and then connect it with Q, like what is literally gonna happen here, right? Um and and it's that sense of curiosity that leads you to your own form of thinking, right? Because you're not necessarily following a path that's been created, you're following a path of discovery, right? And and curiosity always also gives you the ability to listen differently, right? You know, when people say things, like even when I'm listening to you and you're talking about, you know, your heritage, I'm actually really curious, right? And so my curiosity, you know, if we have more time, would lead me to ask you a whole lot of questions, right, about your heritage. How's that connected to you? And through that curiosity, I'm often thinking, how's that connect to me? Not necessarily that we're relatives, although that could be cool, right? But also, like, where's the connecting thread? And then what am I going to take from this? Because that is the other thing that I've learned that I would share with anybody. The one thing that you can share and never lose is knowledge. So do it. Do it all the time and also take it from other people because they don't lose it either. Right. And so that's the one thing, or two things, maybe, I guess, that I would absolutely be sharing with a grandchild. Always remain curious, always take knowledge, but more importantly, give it, give it out because you're you're not diminished on either side of that equation.

How To Reach Max And Closing Call

SPEAKER_00 40:46

Man, it's absolutely beautiful. This has been a great discussion. You've got to let people know how to contact you because you have come up. Here we are talking about Black History Month. Matter of fact, in Oma, Nebraska, we've got our inventors um display up in the Black Museum. You've invented something. You like I said, this is not something that happened in you know in 1870 and blah blah. This is 2026. We have Max here, Max Rigsby. He is an entrepreneur, he is an inventor, he's a digital technologist. You got to tell us how to contact you and how how we can take advantage of what you've built.

SPEAKER_01 41:24

Perfect. So on LinkedIn, um, I'm Max Rigsby. I'm on LinkedIn. Uh actually, my dad's not there, so it's just me. We share the same name. So I get away with that one, right? Um, and actually, that's that's a very easy way to reach me. Um, people can reach me literally at my business, max at gadgetsoftware.com. Um, you can reach me there. I'm a I'm a co-founder. I I I I'm one of the few people, even on LinkedIn, I literally answer when people write me. I just I do. I just I don't know. I grew up in a time where you just do that. You just you know, you acknowledge human beings, right? Um, and certainly if they'd like to visit my father's uh page, you know, then isnow.substack.com. I certainly encourage it. Um there's a lot of history there, right? Um, but wrapped in a modern vibe. Yes, wrapped in a modern vibe. So um, you know, that's it. Um I'm open, love to hear from people. Uh Grant, I can't thank you enough for spending the time with me this afternoon. This has just been you know a wonderful conversation. We're gonna have others offline as well as online, right?

SPEAKER_00 42:32

And it's always great, as you said, and I I did connect with Max um over uh LinkedIn, but also through another friend of ours who is part of this. He's also a black technologist, and uh, and Darrell, he does some phenomenal things. We'll have him on, he'll probably have me on his show as well. But we need to continue this conversation, and we have to tell our story, and we have to have it in the way that we we collect this story and so that it becomes available. There's no reason information and data is so readily available these days that we it's just a matter of how you want to use it and put it together and have your story about what was happening. That's why I created follow the brand because it affects everyone, and that everyone, I believe, has a story and a story that's impactful for other people.

SPEAKER_01 43:24

So, one last thing, Grant, as as we close out, that I just want to say to the audience um the most important thing with this AI shift, and this is make no make no mistake, this is a dramatic shift. I mean, it is it is unprecedented in so many ways. But if you're gonna do it, you have to be a practitioner. You have to do it, you have to put your hands on this technology, find a project, find something to do, find a reason to use it, find a reason to engage it because it is transformative in how you work, um, and is our opportunity. Right to leapprop. We really can. We really can.

SPEAKER_00 44:03

Yeah, no, a hundred percent on that. You've heard it directly from Max. Get involved. Get your feet wet. There's always something you can do. Start utilizing these platforms for the greater good. And that includes social media and all those things. It's just those are just platforms. How you utilize the platform is how you get the outputs that we have. So thank you again for being a guest on the Follow Brand. Thank you, thank, thank. And you can see all the episodes of Follow Brand at Five Star BDM. That is the number five. That is Star S T A R, D for Brand, D for Development, Informasters.com. I want to thank you again for being on the show. Thank you. You're welcome.