When a bill is moving, AI can't keep up: Miranda Hoover, founder and CEO of 3C-House, explains that ChatGPT and Claude rely on one or two official websites for live legislation, and those can lag the room by hours or even days. Paris Childress talks with Miranda about verifying AI answers, why AI won't replace lobbyists yet, a real-time legislative data feed for AI delivered through MCP, how 3C-House uses Otter.ai, Claude, and HeyGen, a central AI brain for a company, proactive agents like OpenAI's Dots, and why every company should start recording its meetings.
When a bill is moving, AI can't keep up. Miranda Hoover, founder and CEO of 3C-House, explains that for live legislation, ChatGPT and Claude rely on one or two official websites, and the people who update them fall behind when legislation moves fast. The lobbyists in the room stay ahead.
In this interview, Paris Childress, founder of Hop AI and co-founder of GEOforge, talks with Miranda about source reliability, real-time data for AI, and how a government affairs firm uses AI every day.
For a live bill, AI can only draw on the official legislative websites, a federal, state, or parliament site, and the staff who update them fall behind on fast-moving days. Miranda recalls a client whose ChatGPT answer placed a bill where it had been 35 minutes earlier. Bill tracking software sources the same sites, so it shares the lag. Her rule for clients: trust it, but verify it.
"Now Chat and Claude are relying on one or two websites." (Miranda Hoover)
Not yet. Lobbyists sit in the room where bills move and decisions are made in real time, and AI can't yet follow who is speaking or the nuance of each chamber. Nevada adds its own twist: its Legislature meets for 120 days every other year, starting in February of odd years, so a lot happens in a short window.
Paris and Miranda sketch a business: AI recorders in legislative sessions feeding a real-time database that AI labs license, the way X feeds Grok. Proprietary, real-time data is what AI engines need when query fan-out runs out of good public sources. Delivered through an MCP connector instead of a user interface, it could be the next generation of bill tracking software.
3C-House uses Otter.ai to record and summarize meetings, Claude across the team to draft emails into drafts, run weekly automations, and keep project trackers, Claude to build in-house projects, and HeyGen for AI avatars. Paris adds his vision of a central AI brain for a company, fed by every meeting, chat, email, and document, and shares how GEOforge has been built with Claude Code.
Proactive assistants like Meta Muse and OpenAI's Dots, which launched on September 29, will start more conversations themselves. That makes prompt tracking and attribution harder, even as AI grows as a marketing channel. Paris and Miranda also agree that AI avatars should be disclosed.
Subscribe to The GEO Show wherever you get your podcasts, and watch the full episode on YouTube: https://www.youtube.com/watch?v=WVZd5miX9LE
Paris Childress [00:01]: Hi, everybody. Welcome back to another episode of "The GEO Show," brought to you by GEOforge, full self-driving for AI visibility. Today my guest is Miranda Hoover. Miranda is the CEO of 3C-House, a really exciting firm doing lobby- lobbying, PR, and a whole bunch of other cool stuff that she'll tell us about. Uh, so Miranda, welcome to the show.
Miranda Hoover [00:27]: Thank you so much for having me, Paris
Paris Childress [00:29]: Why don't you start us off by telling us who you are and what you do?
Miranda Hoover [00:35]: Thank you so much. Yeah, so I'm Miranda Hoover, the founder and CEO of 3C-House, as you mentioned. Our firm specializes in all things government affairs, public relations, business development, and helping businesses and individuals to grow, um, their business and idea, be the best advocate that they can possibly be by utilizing new technologies, by getting introduced into new industries, um, which could be anything that, uh, is currently happening or an idea that they have and they would like us to help them, uh, navigate some of those regulatory hurdles that, as we all know, tend to come up, um, and how it is that we can continue to push forward in this ever-changing, um, technology and business world
Paris Childress [01:22]: Great. Thank you for that intro, Miranda. Uh, tell me a little bit more about the types of companies and industries that you focus on most
Miranda Hoover [01:33]: So we tend to focus on a myriad of, of industries. We are really focused in the energy space, the, uh… And when I say business space, really any type of business. That could be a small business with one individual person, or that could be a multimillion-dollar type company. So energy space, which also includes startups, also includes very large energy companies. We specialize in transportation, mining, healthcare, um, construction, just to name a, a few
Paris Childress [02:05]: Great. And is part of your remit and scope with, with your clients, does it include helping them improve their AI visibility and the way that AI surfaces them and recommends them?
Miranda Hoover [02:21]: It is, yes. One of the biggest things that we have been working on with clients, um, for many years is all about SEO, and I'm sure your audience is well aware that that means search engine optimization. And, you know, since the 90s, since the, the life of the internet that was created, we've all been really captivated, um, by the SEO as Google had come out, and Bing, and Firefox. And how is it that your webpage is ranking amidst all of the rest of them? So when somebody goes to, to say Google, Miranda Hoover, you know, where is it that I'm popping up in the SEO? Well, we now have this incredible thing called artificial intelligence, and with AI, we are now having to analyze not SEO per se all the time, but now in a GEO setting. So instead of just googling my name or my company or, you know, a question you have, people are now putting it into AI to say, "What is this company?" And… Or even something as simple as, "Make me a grocery list based on my dietary restrictions, based on this amount of people that I need to feed." And so when we're creating these questions, we're writing the prompts, um, as in the AI world we like to call them.
Miranda Hoover [03:44]: is the information being sourced? And so now we're getting into a situation where, um, we're having to make sure our answers and the information we're getting back, are they reliable? Are they secure? Are they legal answers? Are they the right answers? And where is it being sourced from? So it's like opening up, you know, a million different webpages all at the same time, sourcing information that sometimes is contradictory. And so you really have to ensure that especially when you're in business, regardless of what type of business you're in, you are verifying. You can trust it, but make sure that you're verifying it. And so that's something we've been helping businesses to navigate as we do a lot of digital marketing, we build websites. But again, on the full public relations side, it's an all-encompassing service. And so we know for our clients, without having that proper, you know, behind, uh, behind the webpage work and services that we do, the public view of the website really at this moment doesn't
Paris Childress [04:51]: Mm-hmm.
Miranda Hoover [04:51]: We have to make sure it's safe, secure, reliable, and legal.
Paris Childress [04:54]: Yeah, absolutely. I think so much of public opinion is now formed before a user comes to a brand's website, because now a typical journey starts with chat, like a ChatGPT, or even on Google now, AI Overviews and AI Mode is now taking over traditional- the traditional search engine result page. And you-- The, that buyer's journey is kind of collapsing into chat, and by the time someone visits your website, you can assume that they have already learned a whole lot about you from AI and have formulated some opinions. And then your website is there to validate that, yes, you do exist, and you're professional, and you have a nice-looking site.
Paris Childress [05:37]: And that last bit of trust is kind of withheld until the person visits and, and validates that by visiting the website. Um, but I, I think it's just fascinating now how much power AI has in shaping, generating demand, shaping, uh, public opinion about a brand. And what, what I'd like to know, 'cause you mentioned the accuracy thing, and this is a huge issue. A lot of times, AI, if it doesn't know something about a brand, rather than say, "I don't know," it will make something up, and it will hallucinate. It used to be worse. I think it's getting better.
Paris Childress [06:15]: Do you encounter many of these hallucinations when you're spot-checking different prompts? And how do you deal with that, and how do you go about trying to fix that?
Miranda Hoover [06:25]: As, you know, someone who also works intimately in politics and in the government world, it's every single day. And, you know, unfortunately right now in, in this world where it's very hard to have something trusted and verified, you know, myself as an expert in the field, when I put something into Claude, when I put something into ChatGPT, I'm reading it more for an overview or more for bullet points of helping me. I'm not having it draft 100, you know, percent of a briefing, a memo, an email, because 99% of the time there's gonna be something in there, even if it's one simple word, that is not going to be accurate. And when we talk about, you know, in the government sphere, when we talk about a legislative bill, whether it be federally, state, a local ordinance, something looking to pass, issues with sourcing. And
Paris Childress [07:22]: Mm-hmm.
Miranda Hoover [07:23]: sourcing, I mean the federal website, a state legislative website, a parliament website. There's only one, maybe two of those. And so instead of a ChatGPT relying on a thousand different sources to then pull it all in, try to decipher which one is, quote, "the most accurate, most up to date," now Chat and Claude are relying on one or two websites. And many times on the backside of those websites when we're looking for updates on a legislative bill for the time that an update happened, what the update is, who created the update, we're relying on individuals who haven't had the time to make those updates.
Paris Childress [08:08]: Mm-hmm.
Miranda Hoover [08:09]: so it's happened already where we've seen-- we've, you know, put into Claude or ChatGPT or, you know, a client calls us and says, "I put this into chat, and it's telling me this bill is actually in this spot,"
Paris Childress [08:21]: Mm-hmm.
Miranda Hoover [08:22]: fact, that was thirty-five minutes ago, and we're already well ahead, um, of where that bill was. So again, when we, when we talk about industry, it's you can trust it, but make sure you're verifying it. At, at this moment, you know, it would not be safe to take 100% of every single word, every single project that AI is creating for you and being able to fully trust it.
Paris Childress [08:45]: Mm-hmm.
Miranda Hoover [08:45]: you have to make sure that you are rereading it yourself and that you have a little bit of an understanding of what you've asked AI to do for you so that you can edit it, um, if and when necessary
Paris Childress [08:57]: That example that you gave with a bill that is out of date after 30 some- 30 minutes or so, that's really, really interesting because you're relying on government officials to update in real time, and that, that's just not gonna happen. But then if people, people are still searching and asking ChatGPT questions in real time, so how, how do you resolve that? Do you have direct access to the webmasters of these government sites, and can you email them and say, "It's really important that you update"? I mean, maybe there was a hearing and something changed, and now th- that has to go right into the website.
Paris Childress [09:34]: I- am I, uh, going in the right direction? So I assume that there's some sort of a decision, maybe a judge, uh, or a court hearing of some sort where a bill, um, is resolved or changes and that impacts a lot of people's lives. It might impact h- how they would vote, and then it needs to get updated in the public record, starting with a g- some government website somewhere, and that doesn't happen for 30 minutes, or maybe it doesn't happen for a few days. I don't know.
Paris Childress [10:03]: Um, how do you make that happen if, um… Do, do you have direct access to people who have the content management system?
Miranda Hoover [10:14]: Well, you know, it's funny that you ask this question as, you know, everybody seems so afraid that AI is gonna put, you know, all of these people out of work, and I can confidently say,
Paris Childress [10:26]: Mm-hmm.
Miranda Hoover [10:26]: in the government affairs lobbying industry, we are not going to be losing our jobs anytime soon because of AI,
Paris Childress [10:33]: Mm-hmm.
Miranda Hoover [10:33]: of exactly what you mentioned. We get to sit specifically in the room where the bills are moving, where decisions are being made in real time, and it's only because on the back end, the secretaries, um, the individuals who are making those updates, they get very behind because of how quickly legislation moves. And so, you know,
Paris Childress [10:58]: Mm-hmm.
Miranda Hoover [10:58]: ever see a, a standpoint where AI or the computer's ever going to be ahead of us.
Paris Childress [11:03]: Yeah
Miranda Hoover [11:04]: in a couple decades if we get to a point where they are, you know, manually transcribing every single decision that's happening, you know, in a senate chambers, you know, in, in parliament, you know, all of these different avenues, and being able to directly go, "Okay, I know who's speaking, I know where they're speaking, I know their vernacular, I know their tone, and I can then decipher from a, an AI technology perspective exactly what's happening with this bill." AI's just not quite smart enough yet to be able to do that, and you know, now that I say that, AI's probably gonna start piecing things together just, you know, from my computer and my, and my phone listening
Paris Childress [11:44]: Mm-hmm.
Miranda Hoover [11:45]: But they don't know who, you know, the speaker is. They don't know who the majority leader is, and, and they don't understand the nuances that are happening in every country, every state, every city, every county. So until AI gets smart enough to be able to physically be like a person
Paris Childress [12:03]: in the room?
Miranda Hoover [12:05]: those
Paris Childress [12:05]: Uh-huh
Miranda Hoover [12:06]: us as lobbyists, I think, are, are pretty safe in, in our
Paris Childress [12:09]: Yeah
Miranda Hoover [12:10]: I don't see a, a time where we're gonna have those real time updates within even probably a couple hours. Um, within a legislative session, you know, every, every legislative session across the country, they have deadlines. And so say there's an, an April 1st deadline, um, and it's midnight. There's going to be things going on until 11:58 PM,
Paris Childress [12:34]: Mm-hmm.
Miranda Hoover [12:35]: and throughout that day, though, the secretaries, those who are including information, um, into the system, they're gonna be really behind 'cause there's just so
Paris Childress [12:44]: Mm-hmm.
Miranda Hoover [12:45]: moving. So it may not be that we even have an accurate update for days after this
Paris Childress [12:51]: So there--
Miranda Hoover [12:52]: unless you were in
Paris Childress [12:52]: So there are human-- The secretaries are human note takers that are trying to keep up with what's going on in real time in these legislative sessions. And are they-- Do they-- Can't they just turn on an AI recorder on their laptop and have it start listening, or is it, is it not that easy?
Miranda Hoover [13:10]: it's gonna be, I think it's gonna be really interesting to see if state legislatures start, um, passing legislation that allows for
Paris Childress [13:20]: Mm-hmm.
Miranda Hoover [13:21]: At this moment, they do have transcriptions, and that allows them to help with their committee, um, secretary minutes, that allows them to be, um, to take floor minutes.
Paris Childress [13:31]: Mm-hmm.
Miranda Hoover [13:31]: But they have to then take the transcription, and they're still typing it, editing it based on what it's giving them.
Paris Childress [13:38]: Yeah.
Miranda Hoover [13:39]: know, as, as you know, AI still cannot decipher emotions very well. It can't pull those nuances of so and so senator said so and so, and in that one sentence, it maybe killed a bill. It maybe resurrected a
Paris Childress [13:56]: Mm-hmm.
Miranda Hoover [13:57]: Um, so until we get to that point, it is still vitally important that these secretaries and individuals who are inputting this information are making sure it is accurate, um, and that the minutes are being transcribed appropriately
Paris Childress [14:11]: and then all that eventually will get published as soon as possible on that government website, and you're saying that that's the only real source that AI trusts. Because if you, you can go in and take your own, you can bring in your own AI note taker, I mean, you can probably just record from your phone and immediately publish something on your website or maybe on other websites that this is what happened tonight, this bill was passed, this bill, this bill died, there was this debate between these people. But is AI going to trust enough your update, even if, if it comes out 48 hours before the government, the official publications come out? Or is that an effort that's worth doing, or have you tried or thought about it?
Miranda Hoover [14:56]: So I'm actually gonna be very curious to see. So Nevada has a very interesting legislative session calendar that no other state in the entire country has. So Nevada only goes into a 120-day legislative session every other year, and it happens in odd years from February to June. So we won't be in our next session until February of 2027. And because of this res- you know, huge impact of AI in the last, you know, 12 to 18 months, wait to be back in our Nevada legislature and see who from a lobbyist perspective is using AI and how they're using it. Um, you know, there's some newer technology as well that's come out that I'm sure you've seen. It's like a little box that people are carrying around. Um, it looks like a little hard drive that they're utilizing, I think it's called the Pocket, um, that they're utilizing to, uh, take notes and have things recorded. Um, personally, in my company, we utilize Otter AI. Um, we love Otter.
Miranda Hoover [16:01]: Um, and you know, we can just add it to simply any meeting, and it takes not only fantastic transcription, but it also gives us fantastic summaries, call to action, next steps, all of the things that both myself and anyone else on the call, um, you know, need to follow up
Paris Childress [16:18]: Mm-hmm.
Miranda Hoover [16:19]: Um, so I'm gonna be curious to see how that goes in legislative session, what we do. Um, and the second point to that is, you know, when we talk about the sourcing, there are a ton of different softwares out there called bill tracking softwares. The problem with them is you end up spending thousands and thousands of dollars to utilize their platform to simply put into a PDF or a report the way that your company and your clients are looking for a report. The only place that these bill tracking softwares can source are the state or the federal websites. There's no other place for them to source. So even you paying thousands of dollars for these products, you are still waiting on the lag that's happening between the state or federal website and what the bill tracking software is able to
Paris Childress [17:14]: Mm-hmm.
Miranda Hoover [17:14]: up. So, you know, there's pros and cons in using all of them. I would say at this day and age, I'm, I'm gonna be surprised to hear if, if lobbyists these days are still using a lot of bill tracking
Paris Childress [17:25]: Yeah. That sounds like an industry that's on its, on its way out. I mean,
Miranda Hoover [17:30]: That one might
Paris Childress [17:31]: it, it actually sounds like a great business opportunity to, uh, to be real-- to basically to offer a real-time solution via AI if you're… If you have access to those sessions, which you do, and they allow you to bring in recording devices, AI devices. Who knows? It could be all sorts of things. It could be your Meta Ray-Ban glasses. It could be, uh, that little hockey puck that OpenAI is about to release at some point, the Jony Ive product. And then you mentioned there's already some stuff out there that you can-- it can listen, it can digest, it can process it, and then you can walk out of the room, and it will already have a great summary prepared, and it's ready to publish it. It's ready to push it out there.
Paris Childress [18:15]: And I, I guess that, I mean, AI should-- The AI crawlers that are always crawling and trying to update their knowledge, their training data sets, um, they would also probably agree and, and want to tap into those data sources if, if someone would stand up, uh, such a data source. I, I think this sounds like a great business opportunity for you, Miranda, actually. Um,
Miranda Hoover [18:39]: It might be.
Paris Childress [18:41]: 'Cause I've been, I've been covering a lot of stories on this show about how AI-- Most of what AI does when it conducts-- when it tries to deliver your answer is it conducts a real-time search. It does this thing called a query fan-out. It takes your prompt and turns it into a couple of dozen search terms. It goes to Google, and it goes to other search engines, does these searches, goes very deep into the search results, pulls in, in a, in a m- in a matter of seconds, it just pulls in an enormous amount of information and then synthesizes that all into an answer.
Paris Childress [19:15]: And what I have been seeing and reporting on through, I think, very credible sources is that AI, like OpenAI, they don't wanna be fully reliant on search engines entirely. They wanna be able to tap into proprietary data sources that may not be published on the web, but they may exist, and they may provide direct access to OpenAI and others like Perplexity. And if there was a, if there was a database, let's just say you were able to have a, a little AI recorder in every legislative session in every county, every city across the US that was listening, recording, ingesting in real time and publishing to a database that OpenAI had direct access to, I mean,
Miranda Hoover [20:00]: that would be, that would be amazing. Y- look, if, uh, if Meta or Anthropic would like to reach out to me and beta test this in the Nevada Legislature, I would be all about it. Just
Paris Childress [20:12]: Yeah
Miranda Hoover [20:13]: me. And you know, one thing that I think is really fascinating, you know, regardless of people's political opinions, is was-- I had created a timeline a few weeks ago about just an overview of, of AI and kind of where we started, we're at today from the last just year. And when we think about the evolution of Twitter X Grok, when X came out and said, "We are going to be flagging information that we believe is inaccurate, not reliable," when you look at the transgression of that timeline to the transgression of Grok being expanded and, and being launched, Grok is directly being fed through what is happening on X.
Miranda Hoover [21:09]: so when we see this reliability kind of checkpoint, to me, it's the first real-time media source that said, "If you are-- do not have reliable information, it's going to get flagged, and it's going to get removed," because Grok is being directly fed through
Paris Childress [21:28]: Yeah
Miranda Hoover [21:29]: So it made me start thinking, you know, what other sources are out there from a media perspective? And it could be something, you know, that's off of a search engine, but something that's reliable that could be continuously fed into a chat, into a Claude, you know, something like that.
Paris Childress [21:48]: Mm-hmm.
Miranda Hoover [21:49]: it's really fascinating what, you know, Elon Musk has done, where he said, "We have this media outlet. We know a lot of news is being sourced here. There's a million plus people on this platform. How do we make it where it's more reliable, more trustworthy, but also have that be what is feeding Grok as, as our AI
Paris Childress [22:08]: Yeah. That's a great example. That's really the best example of, uh, an AI-- uh, one of the major AI frontier companies, which is, uh, SpaceX AI, I think they're now called, leveraging, leveraging proprietary real-time data from their social media platform, X. And I think what we're talking about here is to do a mini-minified version of that, be able to provide real, real-time legislative updates, um, to all the other, to all the other models because Grok does not allow…
Paris Childress [22:44]: Well, I don't, I don't know. Do the-- I've never actually seen any X, um, citations in any ChatGPT answers or in any Google AI overviews or Claude. I think that Grok keeps that away from all the AI crawlers, so they don't have access to it. I'm pretty sure that's true. So this is what-- This business idea that we're hatching right now would be kind of like a, uh, an X and Grok relationship, but for all the other models, specifically for legislative updates.
Paris Childress [23:19]: That could be, that could be a real hit. Um, I'm just thinking now, I guess to subscribe to something like that, I'm a little bit stuck on the actual business model, though. Who is the, who is the customer here that pays? Because people will get those updates through ChatGPT and Claude. Um, well, actually, I, I know who's the customer, of course. It's OpenAI and it's Claude, An-Anthropic. They would pay a fortune
Miranda Hoover [23:48]: may simply be, it, it may simply be the new evolution of a, quote, "bill tracking software."
Paris Childress [23:54]: Yeah
Miranda Hoover [23:54]: The bill tracking software would then be in real time legislative updates that you would still pay a subscription
Paris Childress [24:01]: Ja.
Miranda Hoover [24:01]: um, to have that. But, you
Paris Childress [24:03]: Mhm.
Miranda Hoover [24:04]: the Nevada legislature or California legislature, any legislature say, "Sure, it's fine for us to have this thing in every one of our rooms transcribing and then talking to an AI database," um, would have to have approval, but also they wouldn't be able to say only one company is allowed to be in here, 'cause that would be unethical. So there would have to be some type of middle ground that numerous companies could come in and do that legally
Paris Childress [24:35]: Yeah. And maybe that's a, that's a perhaps a big assumption, but what's your opinion? Would these, these legislative bodies start to allow AI, um, recorders? I guess you can always record anyway. People can r- always record from their phones or from their laptops, and all those devices have been in these sessions for a long, long time. Is it a leap to assume that they would allow these, um, these AI devices that would not only record and transcribe, but then would also
Paris Childress [25:08]: pr- process and, and, um, synthesize and then, uh, be able to deliver that in, in real time to the models? Do, do you think they would say yes to something like that?
Miranda Hoover [25:22]: I think there's a difference between someone like myself, lobbyist or public member, versus asking the legislature to approve device like that to be in the
Paris Childress [25:33]: Mm-hmm.
Miranda Hoover [25:34]: Because our legislatures, they're very people, citizen first, could go in and u- and just set my Otter AI on my phone or my laptop, and not only would they probably never know, but there's not really anything they could do about that because they themselves are already recording the entire meeting.
Paris Childress [25:55]: Mm-hmm.
Miranda Hoover [25:56]: the audio may not be the best, and also you're not gonna be seeing who's speaking, right? Um, versus in a legislative hearing, they have all the video, all the
Paris Childress [26:07]: Yeah. Mm-hmm.
Miranda Hoover [26:09]: production team. So it'd be a little different. I would be surprised if any time in the next few years our Nevada legislature says, "Oh, you know, okay, Meta, okay, uh, Claude, come on in with your own essentially production team, with your own camera, your own transcription capability and, and go
Paris Childress [26:29]: Mm-hmm.
Miranda Hoover [26:29]: it." Um, I, I just don't see that necessarily happening. But, you know, don't let the legislators, uh, you know, fool you. They are all using AI.
Paris Childress [26:40]: Mm-hmm.
Miranda Hoover [26:41]: So, um, whether they are using something to transcribe their own meetings so they can get better real-time summaries and updates, um, you know, I don't know, but they're using AI too
Paris Childress [26:54]: I'm sure that at the very least, they're all using AI to summarize these huge bills that, you know, hundreds of pages that they have to vote on. I mean, to, to read every page of a bill has gotta be pretty, pretty difficult. So A-AI is a perfect use case for summarizing that. It is a really fascinating use case because it also reflects what's happening in the broader software industry in SaaS, which is that a lot of people talk about the SaaSpocalypse, how AI is actually now replacing a lot of specifically point solutions software that maybe you used to pay 50, $100 a month for a subscription, but AI can do it now.
Paris Childress [27:38]: I think those, those are really under threat. But even the leading software SaaS companies like Salesforce and HubSpot, they're actually providing a headless solution via MCP, via an MCP connector directly to AI so that users can access their, their tools and their data through Claude and through ChatGPT. And I think that the same would be true here, that basically you'd be-- I mean, you, you could, you could be the bill software 2.0 that actually never really sells directly any sort of an interface to people, but that you're really just packaging that data and then providing an MCP connector to, to the, to the models. And then everybody in, in that industry would then, um, connect via that MCP connector. They would basically use your software through Claude and ChatGPT the way that they're now using so much software through MCP connectors right now. Do, do you use any MCP connectors or does this ring a bell? Uh, ha-have you hooked up any, any software via MCP to, let's say, Claude or something?
Miranda Hoover [28:53]: Yeah. So, um, we've been dabbling in this as well. Um, and it's been interesting. Uh, you know, we as-- at 3C, um, we have some in-house projects we're working on, um, one of which I can't speak to yet 'cause it's not public. But we have been sourcing from the Nevada legislative website, Congress, US Senate, um, to plug in as MCT for sourcing into different projects that we are now building, um, and utilizing those sourcing, uh, items to model, you know, kind of where we see gaps, um, successes, what's reliable, what's in real time, and then from our end being able to add more to, um, to that reliability, to that timestamp.
Paris Childress [29:42]: Mm-hmm.
Miranda Hoover [29:42]: So, um, we've got some projects in the works that we're looking to beta test during next year. Um, but MCT is definitely, I think, gonna be a big topic for, for the twenty twenty-seven, you know, economic development, uh, technology rooms.
Paris Childress [29:57]: Really interesting. And more broadly speaking, Miranda, how are you and your team at 3C using AI just to tr- be more productive, or have you had any, any major breakthroughs? Uh, are there things that you're doing now that you-- it, it just would've been way too time-consuming before? Are you doing new stuff? Um, just what's, what's been your experience and your journey on the, on AI transformation within your agency?
Miranda Hoover [30:26]: I love it. So, um, as mentioned, we use Otter.ai, and we use it for almost every single meeting. Um, it's really great for our team, and, you know, something you and I talked about offline already is I love the ability, and, and I always have been this way, of being able to be 100% present during a meeting and not having to be scribbling notes on the side and saying, "Hold on one second," you know, "Can you repeat that?" Or, "I'm taking these notes." And so Otter.ai has been revolutionary for us. Um, we have so many meetings, you know, every single week. We are talking with governors offices, legislators, regulators, clients, um, potential clients, lobbyists, colleagues, um, you know, ev- and everyone in between. And being able to remove the necessity of me t- taking notes, then going back in and either typing them into a computer or cleaning them up, and then creating the summary, creating the call to action, what we all need to do, it has revolutionized the way that we are working. I think we
Paris Childress [31:36]: Mm-hmm.
Miranda Hoover [31:36]: so much time just simply on that one ginormous piece. Um, and we tested a few different, you know, note takers, but, you know, we, we love the one that we chose. Um, number two, we have also used, you know, lots of different AIs. We've tried Copilot, we've tried ChatGPT, we, we use Grok. Um, but we settled on Anthropic with Claude, and our team has a Claude subscription that we use. And, I mean, Claude has also just revolutionized how much work we're getting done every day.
Miranda Hoover [32:11]: from, you know, kind of simple tasks on my end from an administrative perspective. Um, I have Claude, uh, in my, in my email account,
Paris Childress [32:22]: Mm-hmm.
Miranda Hoover [32:23]: while I have specific allowance, it, it is not allowed to send emails. Um, it has to put them into my drafts before they're
Paris Childress [32:31]: That's the same thing I do, yeah
Miranda Hoover [32:32]: the send button. But I also have automations for every single week. You know, certain reminders for my team with the same, you know, Google Drive link that everyone's working on, project trackers that are updating in real time. Um, and being able to pull and ensure that things are not falling through the gaps, that our clients are always being taken care of, we're always getting back to legislators and regulatory agencies. Um, and so a-again, I mean, we save hours and hours and hours by simply not having to manually do, um, that work.
Paris Childress [33:09]: Mm-hmm.
Miranda Hoover [33:09]: then on the third, third side, um, we are using Claude to, um, build out a couple of the projects that, um, again, I can't speak to quite yet, but we're utilizing these programs to build, um, build things where in years past and, you know, essentially since the creation of, of the internet, people have had to code and people have had to understand how to, um, you know, what is DOS and, and how to use, you know, WordPress coding, HTML.
Miranda Hoover [33:38]: are getting away from that by simply building prompts in a program
Paris Childress [33:42]: Mm-hmm.
Miranda Hoover [33:42]: able to have AI code on behalf of us. so we've been dabbling a little bit in other programs l- such as HeyGen, um, for agent building. I think that's something that is going to become, um, more used, uh, in the near future where HeyGen, you know, gets to come onto a video call with me, learn my vernacular, learn my tone, learn, you know, how I use my hands sometimes and, and how I speak, but it's also gonna learn and understand all of my expertise and all of my knowledge. And then it creates an AI exact replica
Paris Childress [34:24]: yeah
Miranda Hoover [34:26]: An avatar. And it's, it's fascinating. I think that's something that hasn't quite fully been tapped into yet from a, a user perspective, and I think companies definitely should be looking into that one more, uh, to, you know, to make sure that they're not hiring, you know, a whole bunch of new employees simply to be, like, on-call experts for them. They can now build these AI avatars who have all of this loaded, you know, expertise, knowledge, information, um, and it's still coming, you know, quote-unquote, "from me."
Paris Childress [34:58]: Yeah. Wow, that's a whole lot. Uh, let me try to summarize th-those main points that you delivered. So f-first of all, Otterly call tracking. I think anyone listening, uh, I, I… Most people, I do believe now have note takers, but, uh, I know Otterly is great. We use Fireflies and have stuck with that for a long time. I think what our next step is with the call, with call tracking, AI note takers, is I wanna centralize all those transcripts into some sort of a central AI brain so that it can actually deliver real strategic insights at a higher level. So I, I would be able to ask… And by the way, it would not only be the transcripts of all, of every meeting, but it would also include all the Slack chat content across all Slack channels, probably all e-email communication as well. So em- our employees would have to accept a degree of lo- a lesser degree of privacy there because all the emails we get ingested in all of the chat. We, we actually don't use Slack, but we use, we use Google Workspace chat internally. But we use Slack with, with most of the cl- the clients. But if all that came into one central repository like a brain, then we could--
Paris Childress [36:15]: it could proactively say, let's say, uh, I don't know, something may have happened in a meeting that you didn't attend, and it might not have bubbled up to you, but this AI would proactively reach out and say, "Hey, Miranda, we have a client now that's at risk because I detected a very subtle statement in a meeting from last night that your, that your colleague so and so was on, and they said that they're starting to experience some financial trouble and, and maybe they might have to cut costs or something like that, and you should be aware of this." And I think there would be so many opportunities if you were able to centralize, starting with your call, with your note taker data, but then including your ch-- all of your chats and all of your emails. And of course, every document that gets produced in, whether you're in the Microsoft ecosystem or, or G Drive or whatever it is. Of course, every, every document, spreadsheet, presentation, all has to go in there too. But then you would have such an amazing body of context that AI could, could really become It could become the strat- the strategy lead. It could become your chief strategy officer of, of your business. Proactively tell you, uh, what you should really be prioritizing this week or today. Where are there unknown hidden risks in your business that you weren't able to spot on the surface? And a whole of other… lots of other things. It would, it would in-- dictate that employees would have to buy in and kind of agree that, all right, all of our stuff now is gonna go to AI for this type of processing, and there's, there's kind of radical transparency related to that, but I think it's a good thing. Um, this, this first idea came from, uh, well, I listened to a podcast with Jack Dorsey, the co-founder of Twitter, who we were talking about earlier, and this is, this is his vision now for the modern org chart of a company, which is that a central AI, um, let's call it a brain, for lack of a better term. I-- There's probably better ways to put it. But if, if that AI system has absolutely every bit of context that's happening in the company, it then can actually make better decisions than, than almost any manager in the company. And you should, you should allow it to at least suggest the right decisions and, you know, create the agenda, so to speak, for the leaders and the decision-makers in the company. So that was part one of what you said, Otterly and note-takers, but I think that's where organizations are headed.
Paris Childress [38:42]: Um, and then the Claude stuff is, is also amazing. We have Claude licenses for all of our team as well. Uh, massive productivity gains and scheduling tasks, like you mentioned, that's super useful. Um, and the coding, uh, our, our own product, GEOForge, is mostly-- We, we did have a developer build the core product, but then after that, all the, all the feature development and evolution since then, and that was about April. So it's been a while, the last six months or so. It's all been kind of vibe coded, really, with Claude Code. And our head of product is not even a trained developer himself, but he's been doing amazing stuff.
Paris Childress [39:23]: But that just democratizes coding, 'cause you used to have to pay a lot of money and wait a lot and just deal with developers who were not always so nice to deal with. Uh, but now, um, anyone can really build if you have an idea for anything. And,
Miranda Hoover [39:40]: Mm-hmm. And I think that's probably the beauty of, of AI and technology is can do it.
Paris Childress [39:49]: Mm-hmm.
Miranda Hoover [39:50]: And so, you know, I, I know your listeners are probably a whole lot more tech savvy than some of the industries that I, I work in, but, you know, for anybody who hasn't yet tried ChatGPT or Claude, go try the free version. Just go try it and, and use it, and, you know, I'm sure for your team as well, Claude for, for us knows each of us individually very well by this point. It knows, uh, my tone. It knows how I like to speak. Um, you know, Claude, when I would, uh, use it for some business stuff at the beginning, would come back to me and s- and flat out say, "I know you're looking for an answer that does not beat around the bush, because you don't like BS and you want answers direct to the point." And I went, "You know me well
Paris Childress [40:40]: Mm-hmm. Bingo. Yeah.
Miranda Hoover [40:41]: Great."
Paris Childress [40:42]: Yeah. And it's, it's only gonna get better and better the longer you work with it, the more context, history, and memory gets accumulated, and that's, that's true, I think, across the board. And we're moving into now a world where I think it's transitioning into these assistants like Meta Muse. Actually, Grok Bot, I believe, started it, and I, I love Grok Bot. Um, I have a whole team of Grok Bots. And then next came Meta Muse, which I think is… I, I'm in Europe, and I-- we don't have access, unfortunately, but I've heard it's been a runaway success, early success. And then yesterday or the day before, OpenAI quickly followed up with something called Dots, which is the same kind of thing. It's a, it's a personal assistant that has all the context about your, your work life y- and your personal life, whatever you wanna give it, and it's, it's there for you, chatting with you throughout the day.
Paris Childress [41:38]: And unlike the classic LLM experience where you go to, you go to the chat with your question, with your problem, or with your task, but you're the one that initiates it. The agents are now the ones that are proactively saying, "Oh, Miranda, a reminder about this," or, uh, "You asked me to watch out for this pr- the price of this thing. Once it drops to a certain point, it ping you. It just did. Now is the time to go buy it." Or all sorts, I mean, millions of examples where AI now becomes a proactive assistant who's nudging you about the right things at, at the right times throughout the day, keeping you focused, and knows your priorities. And I think that's the next wave, and I think that it will change even how we do AI. I mean, we're, we're, we're thinking about AI search as a marketing channel, but imagine when there aren't really p- Prompts anymore. I mean, there still will be prompts where you go in with a task or with a question, but maybe a lot of these conversations will be initiated by the AI agent and then the, the, the human is reacting. So, um, it's, it's not gonna be so clear about, you know, tracking prompts, tracking visibility, as my brand mentioned or not. I think it's gonna hap- things-- The impact of that will happen more downstream in terms of the actual events like purchases or lead generation.
Paris Childress [43:03]: And, and, and we have to solve for this attribution problem. How do you know how someone really came to your website? Because they didn't click on anything from a chat experience, or if they were in an agent and they sent their agent to the website to go do something on, on your behalf, well, how does that show up? Actually, this… That's not gonna be like a referral visit that you would normally track in Google Analytics. So it's getting messy, it's getting fascinating, um, but I do think more and more the users and visitors to websites will be these agents acting on behalf of, of, uh, humans, and it's gonna, it's gonna make the tracking and attribution a lot harder.
Paris Childress [43:45]: Um, and then as a marketing channel, it's only gonna be more important but also harder to track at the same time. So the idea about the data repository for all the legislative updates in real time, uh, it's, it's amazing because I think OpenAI, Anthropic, SpaceX would all pay a fortune to license that data so that they could serve it up to all the sta- all the, the users
Miranda Hoover [44:13]: Absolutely, 'cause then we get into kind of the next question of who, who's indexing data? And, you know, unlike encyclopedia where have all the information right there, you know, Google has, for many of us, acted for the last, you know, two decades as our digital encyclopedia. But now it's what is reliable, what is actually being, you know, indexed appropriately, what policy positions are legible, are they current- Mm are they citable? Um, I mean, it is an ever-changing, again, regulatory process and, and economic, um, you know, environment that is changing. And so I'm gonna be really curious to see, you know, some of these agents, you know, such as, again, you know, HeyGen that has created. Are they going to get so good at being able to take the way someone looks, the way someone speaks, their, their mood, their, you know, their energy, and transpose all of that to then utilize some of those sources for indexing, and then how we have information, um, you know, being shared. And, you know, I, I think to your, to your point, it's gonna be an interesting, I think, even just next 12 months. We've watched AI, completely transform in the last 12 months, and businesses, you know, looking to be a lot more productive just by simply taking tasks off of, you know, an, an assistant or, um, even the CEO's day-to-day requirements
Paris Childress [45:47]: Yeah, I-- it's gonna be fascinating to see how the political landscape, and in particular, how many politicians will be willing to create HeyGen avatars. I think right now it'll be a s- it's a small minority today. But more and more, the avatars will get better, and probably the avatars will need to be… The first thing that an avatar would need to say in a video would be, uh, "I'm Paris' avatar, and Paris has fully endorsed or ha-has, uh, has, is fully endorsing this message that I'm about to deliver to you." And it's kinda like the political ads that's, that where you hear that audio at the end, uh, "I'm so and so, and I, I endorse this message," but this would be the avatar speaking upfront. "I'm not-- I, I am an avatar for Paris. I know I look very lifelike, and I just wanna let you know that Paris did, uh, endorse everything that I'm about to say here today." And, um, but imagine this, the, the degree of scalability then, 'cause then you could just reach. You wouldn't need to sit down and record every- everything that you wanted to get out into the world. You could, you could push that through HeyGen. I've used HeyGen, um, actually to deliver a, uh, a very long cybersecurity marketing course, and all I did was just work on the tran-- the, the scripting of it, and then just fed all the scripts to HeyGen, and it, it did a great job. And that was about a year ago, and I think HeyGen, I haven't used it since. I, I'm sure it's a lot better.
Paris Childress [47:12]: But it's only gonna get better and better, and it will become in- undetectable pretty soon, that you won't really even know. So you'll, you'll have to demand that, that people disclose, "All right. I, I am Paris," or, "I am Paris' avatar," and that needs to be disclosed right up front. Um, but there's huge marketing opportunities with that because video, people do prefer face-faces, talking heads, videos, um, primarily. That's what's most engaging. Yeah. Wow, so many great, great ideas and so much in- inspiration today, Miranda. It was great to, to brainstorm with you.
Paris Childress [47:53]: And, uh, the, the world that you're in is really so ripe for AI transformation, I think
Miranda Hoover [48:04]: Agreed. And again, it's gonna be ever-changing. I'm really excited to see what comes out over these next 12 months and, you know, are legislators gonna try to over-regulate? Are they gonna kind of let it operate the way it's operating? People in, you know, the public relations, government affairs worlds, um, how are they gonna be using AI best of, you know, their jobs and, and for their
Paris Childress [48:29]: Mm-hmm. Yep. Really fascinating times ahead, and we're right, we're right in the thick of it right now. Well, Miranda, thank you so much for being with me. Is there anything I didn't ask you that you wish I would've asked, or is there any- anything else that, uh, you th- you think the audience could benefit from knowing?
Miranda Hoover [48:50]: Um, I think you asked all the great questions, and, um, I hope that your audience took something away from it uh, that they'll go hopefully implement and, you know, maybe play around with a new AI software that they weren't aware of. Um, but you know, really, I, I loved, you know, as you mentioned, all of our brainstorming and these new businesses that we've now created. I think we've got some, some cool things for us to work on and, you know, for your audience to really be thinking about and considering how they can better incorporate AI either into their personal lives or into their businesses, because it's not going away anytime
Paris Childress [49:27]: Absolutely. And if I take away anything from this discussion, the one piece of advice that I would give to listeners is if you don't-- if you're not recording your meetings yet, start recording. Try to start recording everything because that capturing the voice, uh, the, the words that are spoken, it's so much greater in terms of context than even all the written material that gets produced or all of these spreadsheets and all the, the, the data, all of the presentations. But the, the words spoken in every meeting with prospects, clients, and internally, that is really where this-- the most proprietary, valuable, um, essence of the company really resides, I think. So start recording everything, I think is the, the basic, uh, the basic advice here, because then you will be in a position to really empower AI, and then you can start to transform your business
Miranda Hoover [50:28]: 100%.
Paris Childress [50:29]: All right. Well, thank you so much for being with me, Miranda, and I will check back with you in about six months to see how that new business idea is going, uh, if you're, if you're selling that license data yet to the, to the big AI labs. So good, good luck with that.
Miranda Hoover [50:46]: TBD. I really appreciate you having me on, Paris. This was so much fun, um, and I look forward to talking with you again soon. Thank you
