Merging BPA and Agentic AI for User Trust with Conveyor’s Chas Ballew

This is Textron tv. Hey everyone, welcome back here to techron tv. I've got a new company to talk to you about.

They haven't been on our show before, and it's the first time guest. I just met him, his, let me introduce you all to Chad Ballou, Chad, the CEO of a company called Conveyor. Hey Chaz, how are you today?

Hey, Alan. Good morning. Nice to see you.

Good Morning. Welcome. So, I, I guess let's, before we get into conveyor, and there's a lot we're gonna unpack here, Chaz, let's start a little bit with you though.

You're the CEO there. Um, give us a sense of, you know, how you came to be the CEO of this company. What, what's your journey been like?

Yeah, uh, well, again, thanks for having me. Um, my journey, I grew up in the Midwest and always knew, uh, you know, just hacking on computers and very typical story, you know, middle class, born in Detroit, grew up in the Chicago suburbs and grew up, you know, playing computer games, then getting interested in programming, then building computers, and was just kind of in place, you know, for the early internet. And that first wave.

And I know, always knew I wanted to, you know, build a software company and, and make a living in tech to go to college. I really wanted to go to Princeton and run track and cross country there. And, uh, to make that work, I did an ROTC scholarship and signed up and served in the Army for, uh, four years on active duty in Washington dc.

Good for you. Yeah. Well, thank you.

Thank you for your service. Thank you. Yeah.

Um, and it was a, it was a, a very, uh, interesting experience. I, uh, before doing that, I went to law school. So when I was on active duty, I was a lawyer in the Pentagon working at the Army Office of the General Counsel and Jag.

I was like, yeah, exactly right. Yeah, that whole world, we were working on the, um, not on like UCMJ or you know, court marshals or like Jag stuff. We were working on Army, like Pentagon stuff.

Like, you know, what bases go where and how are we doing, um, all kinds of stuff. The particular group that I worked in was anything to do with any land that the Army controls. So all, all, you know, all kinds of stuff from basing and what bases are opening and closing to what can you do on a firing range to what, you know, how do we build new, uh, bases or how do we tear down old stuff?

Like, you know, you could have all kinds of stuff, housing base, typical base stuff, but you could also have weapons, munitions, explosives. The Army has a bunch of stuff that's like beyond normal, uh, real Estate. Yeah, no, it's not your usual land use kind of zoning.

Yeah. You know what I mean? Commercial, we also had it's commercial or residential.

Yeah. Yep. And we also had, the Army also runs, um, the Army Corps of Engineers.

So we had a ton of stuff related to, uh, the Army Corps, including, I mean, everything from Katrina and the Katrina litigation that was a trillion dollar combined litigation, stuff like that to, um, you know, when there's a flood and the Army Corps of Engineers decides or decides not to blow a levee and flood a, you know, flood a field or stuff like that, that all gets turned into, uh, takings claims, constitutional claims, and, you know, claims for, Hey, let's, um, let's get everything domain and stuff like that. Yep, exactly. Yeah.

So we, I mean, I, I could, I could tell you all war stories for hours about all the interesting weird stuff. Um, but that was, uh, for four years and the whole time I was doing this, thinking, okay, I'm gonna have a whole bunch of interesting experiences and background for when I go start my company and work in technology. And so, um, 2013, I got, I, I had gone to undergrad, gone to law school, served four years in the Army, and I got out and immediately started my first company, uh, company called Adaptable, which was a, we started it, uh, as a secure HIPAA compliant cloud hosting platform that runs on AWS and we sold it to digital, uh, health teams and said, look, this is like HIPAA compliance in a box.

And we, we took that do, we did a Y Combinator in 2014 with that and raised a Series A and uh, got that profitable, like nicely profitable. Good for you. Yeah.

And my co-founder is still running that. And in 2021 we said we had this profitable growing business. Um, and then also we saw the opportunity for what was going on with Conveyor and what Conveyor became, which is trust building or, uh, you know, how do you prove to one of your customers that they should trust your security posture?

We saw that at Adaptable in 2021. We said, let's split our bets. Let's take the profitable company and the hosting thing and get it out, you know, into its own vehicle.

And let's take this wild card, which was conveyor this totally, you know, seed stage idea, but really interesting strategically in that what we were seeing was the Adaptable's customer base was small, small companies, startups, and the customer is engineering teams. 'cause they're buying a DevOps platform. And what we started seeing with Conveyor was big companies with big teams and big problems with communication and how they get their story straight.

Like our customers like DocuSign or Workday have thousands of employees. They have hundreds of people on their sales and security teams. How do they all get their story straight and make it as easy as possible for their customers, which could be big banks, fortune five hundreds to trust 'em.

And that is, um, that's still what we're working on with conveyor today. Uh, so that's a bit of what a great story. Yeah.

My background, how we, you know, I'm, I'm a basically the two time, you know, software founder here, and conveyor is the latest iteration. I think the, I mean, that brings us up to AI today and what's going on today, which is just that we started conveyor in 2021 working on a lot of it was like Salesforce automation and automating, getting an NDA signed and automating, getting customers secure documents. Like, oh, you have a SOC two report or a pen test report, or, you know, like collateral that your sales team has been given.

And then you gotta get that to customers, but it also, it's gotta be secure, it's gotta be authenticated. You want sometimes, uh, you know, like timestamps and water watermarks and you know, hey, confidential, don't distribute this type of stuff on the documents. So we started, um, with all of that in 2021.

And then, yeah, just what is happening with deep learning and AI has come in like a freight train and it's getting stronger. You know, the rate of change and the rate of capabilities increase is really astounding. And we, we started in 2022 using, uh, AI to do rag flows and fill out paperwork and say, okay, so you got, you know, you got a security form from your Fortune 500 customer, it's 300 questions long, it's all embedded in this Excel sheet.

We started making software to go and do retrieval, augmented generation, look at a knowledge base or your database and then draft new answers. And getting that right has been really fun, really rewarding, difficult technically, but also pays off in huge ways for our customers. 7 on average.

7% of the time. Like first pass is perfectly accurate. Like, it's like put it to bad accuracy, which is given how security, like, there's a lot of sometimes, uh, bonkers questions or a long tail.

So 95 is actually approaching per like, as as good as you can get. We think about 97 is about 3% of questions are just on average gonna be kind of off the wall. Like you, you won't really be able to like fully predict them.

And so as you get to like 97% automation, you start, you know, that's like as good as you're going to get. That's bad, certainly superhuman. Absolutely.

com days. And, um, we were doing hosted Oracle and PeopleSoft and all these old applications before there was cloud, before you had multi-tenant, all that stuff. And we used to think if you got to 80, 85% it was superhuman.

Mm-hmm. Right? The other 15% is always just craziness, quite frankly.

Right? It, it, it's total one-offs. Just total one-offs.

Right. And so, you know, anything over 85 97 is, is ridiculous. I, it just ridiculous.

So let me ask you one quick question though. So did you run track at Princeton? I did, yeah.

Yep. Good For you Man. And 5K and yeah, Very cool.

What a great story. What a great story and what a great lead into conveyor and, and and what you're doing now. As you mentioned ai, I mean the whole business process automation and low code, no code and so many other things are being disrupted here, but whether, you know, sometimes it's better to be lucky than smart, but whether it was luck or or being smart, you find your place, you find yourselves in a place where agentic AI actually is gonna help you because it, it's kind of what you're doing.

Before there was ai, before there was agents, now you can really accelerate that through the use of AI agents. Correct? Yeah.

So we, in, we were talking, you know, warming up, we were talking a little about, about this. com today and go to our website and look at our use cases, you'll see that we're marketed as automating security reviews and security questionnaires. And we're really targeted towards that.

And all the time we get asked, will you do RFPs too, or RFIs or, you know, how are you gonna handle that? And that's a big part of what we're doing this year is expanding, um, those, you know, use cases. But fundamentally what it really is is automation.

It's workflow automation and it's a specific kind of workflow. The specific kind is that when your customers need answers to sometimes technical, sometimes legal, sometimes compliance questions, and they're putting it on their paper and they're saying, Hey, I need your answers to this. I need you to basically ver you know, attest to the answers to these.

Sometimes that gets pulled in explicitly like into an MSA where they say, okay, here's the security appendix and you promise that, you know, it's a violation of our contract. Or sometimes they're just sent in procurement and there's no explicit attestation that hey, this is part of the contract. But oftentimes these things have, um, the, the stakes matter, like the answers need to be correct in a way where with a lot of internal tools, AI doesn't need to be a hundred percent correct.

'cause you're gonna have a human who looks at it and is like, okay, well thanks, and then decides what to respond to the customer. But the kind of software that we're building the goal, you, me, you mentioned like, you know, what are we doing is to fully automate all of the work that your customer needs to buy you for these, you know, clear you as a vendor for these kinds of processes. And a big part of that is when they have a form for you to fill out filling out the form and getting really high performance on that.

But a much bigger part of that work is all of what our customers call, like the stupid work around filling out the form. So it's what happens before you fill out the form, which is, you know, the sales team sends over five different documents and only one of 'em is a security questionnaire. And they said that there were two things for you.

You know, like stuff like that. It's like, wait a minute now you gotta go back to the sales person and figure out what's going on here. That kind of stuff.

So you haven't even gotten to the point where you're gonna use AI to blast through one of these forms. You're just doing the stupid work around coordinating or AI comes back and it's like, well, we don't have answers. Well, who do you talk to?

Well, what kind of questions are they? Oh, it's product questions. Now we gotta track down somebody in product management, right?

And like, get the answers or figure out have things changed all of that. Do you see what I mean? Like, that's all the work around, even if you have a, a magic rag tool or a really great, you know, form filler like we do, there's still so much other work around that form filling, getting ready for it, figuring out what happens and how you, you know, how you complete the whole thing.

And that's what makes Agen AI so exciting. And that's also what makes, because it 'cause what we can do and what's working is you take all of that stupid work and you really break it down. You really break it down into excruciating detail.

You really, you blow it apart into literally like a thousand different pieces and they're, you know, they're organized by each of these, each of these parts of stupid work can sometimes involve a bunch of different decisions, but you can really break down the decisions and then you can build code and AI working together like scaffolding and calls to models that together can process this cognitive architecture or this decision making flow. But using AI to go through and do it and breaking it down gets you really good results versus one trying to one shot everything in a big prompt and say, you know, hey, like here's a bunch of context. Get the answer right or, um, make a decision.

So anyways, that's kind of getting into the weeds around how to get good results out of this, but the big news is it works, it really works, which is incredible. This is software that's never been possible to build. You can never build software before that makes good decisions out for you.

I mean, look is wild. I've known, I I person cha I've personally known at least three different entrepreneurs who over the last, let's say 20 years have tried to, you know, using BPA kind of stuff, have form fillers, if you will in different ones. I've seen them in real estate, in medical, uh, in, uh, one of my friends actually created the first mortgage application that you didn't have humans involved in.

Right? That was pretty cool. Um, but, you know, the whole ai agen ai potential to just really disrupt this whole thing is, is crazy.

But you said something early on that I wanna come back to as we, you know, round out our time here. And that is, all of this stuff is the means to the end, but the end is building trust between you and your customer. And I don't mean conveyor and conveyors customer though, that's part of it.

But you know, the, the people who are using conveyor, it's their companies building trust with their customers because at the end of the day, that kind of makes the world go round and round. Yeah. It's essential.

And we see it with, we see you security reviews and building that kind of trust, but we also see our software being used for RFPs and RFIs and other types of trust. Like we started, I I, I mentioned this briefly before, but we, um, we started seeing manufacturing companies and agriculture companies. We have a freight forwarder, a logistics, like a shipping, uh, a trucking company basically using our software.

And they're getting basically these, again, these forms they had to fill out. Sometimes they're called RFPs, RFIs, sometimes they're called quality management questionnaires. com is still very much, Hey, this is designed for security compliance automation.

But what we're, what we're seeing is that the underlying technology to represent the data involved in an organization with multiple products, a bunch of different, you know, usually dozens of people on the sales team, a dozen people on the security team, several different people on product and engineering, uh, workday's our customer, they have 12 different product lines across three different subsidiaries. So being able to take all of that complexity and wrangle it really well is the foundation for getting good results with ai. And that's what our, that's what people are seeing when they're trying this out for quality management or RFPs or RFIs they're seeing is the hard parts are processing bad file formats, right?

And import and export out of Word and PDF and stuff. Or like you were saying, um, form filling, like being able to go headlessly and go to fill, you know, get a form and uh, you know, ultimately fill it out with no human involved and make sure it's accurate and that, you know, all of that stuff. Um, those are all, you know, those are all tools, but like you were saying for our, our customers, what their, what their ideal state would be anytime any customer has anything that they need in any format, if they need docs, if they need a form filled out, if they need whatever, they can go to this system called conveyor, right?

But their customer doesn't need to know that. com or you know, Workday or the Workday, uh, customer, um, you know, compliance portal or whatever they want to call it. But they can go to the system and their customers either through, if they need like documents, they can get documents.

But if they need anything with ai, it's gonna be secure. Like they're gonna authenticate to the ai, the AI will be able to talk, pull up their Salesforce record, just like talking with a human. We'll be able to say, Hey, gimme any form, any questions, anything you need filled out and I'll be able to get it back to you in a few minutes here.

Like complete, complete automation of this kind of, yeah. Like trust, trust building, not just around security. You know, today a lot of our marketing, like I said, is focused on security, but what it really is, is anything that the business is gonna be making these kinds of representations too.

So it in it includes helping automate RFP and RFI stuff. You know, like I said, you know, companies that have promises to their customers that, you know, they ultimately get ended, it ends up being like compliance paperwork, right? Your customer sends you something and they say, Hey, you have to attest whether you comply.

But like for, um, for some of these manufacturing companies, they're not making software and the questions aren't about digital security. They're making precision manufacturing machines or, uh, they're making potatoes or they're making logistics like freight, you know, shipping. And they are still, what they wanna be able to do is achieve that dream where they have a system, everything, all the knowledge is in the system, and then customers just get instantly anytime, whenever they need it.

Perfect. Trust building. Perfect.

You know, and that it's like not just a form filling, but anything they need, any answers, oh, one off questions, like here, here you go, on demand instantly at scale. That's the dream for them. And so ai, all the AI stuff and the various products, and we have a, uh, uh, agentic customer trust intake agent today, we're launching an RFP agent, uh, next week.

Uh, you know, like all of that future is still just trying to serve the dream of perfect instant, you know, high quality trust building, you know, un tap anytime. Love it. Ja.

We're way over time. Yeah. I gotta pull the plug on this one, but hey, worries.

Thanks for coming on here and, and getting us smart on conveyor. It's a gr look, this is a great story. They'll make a movie of it one day, but, um, you know, keep up the great work.

We'd love to hear you ki as agentic ai. 'cause look, that's still in itself fairly new, right? As this continues to develop though, we'd love to have you back on and talk about how it's kind of taken you to the next level here, or allowing your customers to build trust with their customers.

Yeah. And as I, you know, we didn't even talk about this, but as you have not just one agent, but multiple agents, how do you oversee them? How do you manage them?

How do you deal with quality? How do you deal with Orchestration Errors and exception handling, you know, all of this And yeah. The ton of, a ton of really interesting stuff to go into.

It's interesting times for sure, my friend. Yeah. All right.

All right. Chad Belu. C Chad Belu, CEO of conveyor here on Techstrong tv.

We're gonna take a break. We'll be back with more.