AI Systems, Sales Automation,Reporting, & Tech Stack Strategy for Growing Businesses

Drew Donaldson (00:00)
And then, we have this other piece that we need. And so we have this, but that doesn’t talk to these other two pieces or only talks to one. now like, how do we connect that one to the other one? And so you build these, these, you know, Jenga towers of software that a strong breeze can just kind of knock over because the connections between them are so important and yet they’re not always either correctly maintained or correctly implemented.

Dylan Silver (01:55)
Hey folks, welcome back to the show. Today’s guest, Drew Donaldson is the founder of Automata Intel, a company that builds custom automation and intelligence systems for businesses. He started in marketing and transitioned into tech after seeing gaps in how companies use software. Today, he helps growth stage businesses, design systems that improve operations, increase efficiency, and fit how their teams actually work. He focuses on building tools that scale business without improving.

or removing, I should say, the human element. His belief is simple, businesses that adapt and build better systems will win and those that don’t will fall behind. Drew, thanks for taking the time today.

Drew Donaldson (02:34)
Pleasure to be here.

Dylan Silver (02:35)
Now, when we talk about the automation space, everyone is seemingly focused on getting into automation. But I think a lot of people are already falling behind. Where do you see businesses failing already?

Drew Donaldson (02:54)
So I think a lot of guys out there that are building or getting into the kind of automation space, starting automation agencies, they’re building a business in search of a problem as opposed to solving an actual problem in the market, the core problem in the marketplace. And if we take it back to basics, the core problem is that as the years have gone on, marketing stacks and tech stacks in general have gotten more and more complex as, know, hey, we have this piece of software, but it doesn’t do this. So we got to bolt that piece on.

And then, we have this other piece that we need. And so we have this, but that doesn’t talk to these other two pieces or only talks to one. now like, how do we connect that one to the other one? And so you build these, these, you know, Jenga towers of software that a strong breeze can just kind of knock over because the connections between them are so important and yet they’re not always either correctly maintained or correctly implemented.

So what we do is we don’t just come in and say, yeah, we’re going to put an agent in your inbox and clean it out.

Because it’s like, that might not be the actual core problem that you’re facing. The core problem is, well, how is this, was the system actually intelligently designed or was it bolted on? Because if it’s just a Franken tech stack, like, right. It’s just stuff that’s been thrown together. That’s not going to solve a modern business use case. And so first part is like, how do we, how do we design a system that actually works, that makes all these pieces talk to each other and what pieces actually need to talk to each other versus

Dylan Silver (04:06)
Yeah.

Drew Donaldson (04:21)
what connections are useless and we should actually disconnect them because they can actually cause more damage in terms of the workflow. ⁓ The last thing you want is to just like connect all your systems together and be like, great, I’m done. Some of those connections shouldn’t be there because the inputs from one system may adversely affect the inputs from another or counteract the inputs from another. So that first system design layer is the first big part. The second is identifying the gaps because I think a lot of people are now realizing that

know, custom software development is no longer the million dollar effort that it used to be, right? Advancements in AI coding have shortened the development time and AI in general has allowed some of these workflows to just be way more efficient if you build these small custom tools that can kind of fill the gaps between these larger software platforms. And so that’s the next kind of thing we attack is like, okay, you have these gaps. Are there off the shelf tools that are going to

fit this gap perfectly, like a missing puzzle piece, or do we need to build something that’s custom to square that, to make sure that these two systems talk to each other or that you can make sure that it’s an environment where you don’t have stuff constantly kind of falling out of the loop.

Dylan Silver (06:22)
What’s a common inefficiency that you see daily?

Drew Donaldson (06:28)
So there’s really two big ones that pop up. The first is reporting. So many people, and this is actually how I got into this space, was you have all of these data points spread out over numerous systems. And if you want to get a report, unless you have an enterprise tool like a Domo built in and have a data analyst or a data engineer managing that, you are going to five, six, seven, 10 different platforms to pull that data in.

And then you’re relying on someone to analyze all that data and then put together a report or a dashboard. And even doing that, a lot of times it’s static. And so it’s not updating in real time. The analysis is faulty because you know, if you’re doing this all manually, well, if you start pulling stuff on, you know, Monday, the first, and you’d only finished pulling data on the 14th, that first data pull is already two weeks old. So that’s one of the big efficiencies is just not having all of your data pipelines connected into a reporting.

dashboard that you can actually count on so that you can make decisions. I think a lot of people look at the only solution of this is buying big expensive enterprise software and hiring a data engineer. And it’s like, well, no. the reality is, is you just need something that can take API inputs from all these different tools and then summarize them into one dashboard. And so we actually just completed a project for a client that just did that. It’s a very, you know, it doesn’t sound very big and fancy, but it, it allows them a level of transparency into their operations that they did not have before.

so that’s the, the first thing, the second piece that I think is one of these, like try and try, fail, try again, fail again, is just like the basic automations around sales and enablement. So there’s so many things in the sales process that can and should be automated. And there’s certain things that can’t and shouldn’t be automated. Proposal creation can absolutely be automated. Right.

Like if you have a discovery process and you’re recording that call and you have notes for what the pain point is, and you already have a deck that’s 90 % of the way complete in terms of like what you offer and how you solve problems, that can be automated. Like that’s a simple API connection to a tool like gamma, or you could use nano banana for all intents and purposes and create a really great proposal that is available instantaneously. Right after that meeting ends, you can have that proposal done.

And yet people don’t implement it because they think, that’s, that’s gotta be complicated. How do I load in all my information yet? They don’t think about all the time they waste. The other thing is like follow-up emails, the number of businesses that leave money on the table because they don’t follow up with a client. And I’m guilty of this just as much as everyone else is. So I’m not pointing fingers. Massive, massive. mean, we ran an experiment. We, we hired an SDR last year and to get her kind of familiar with the system. said, Hey,

Just go into our CRM. It’s a bunch of deadwood in there. See what you can do. Within a week, she had booked five calls with people I hadn’t talked to in years. And so it’s like the simple act of just sending an email and saying like, hey, haven’t talked to you in a while. We should catch up. A lot of times that’s all you need to bring new money back in, or I should say kind of old money or lost money back into the pipeline. So like you can automate that.

You don’t have to manually remember 500 different conversations. All you have to do is they say, Hey, not right now. Check in with me in three months, drop them into a three month automation and it will just take care of it for you. So that’s, those are two of the biggest ones that we say.

Dylan Silver (10:31)
The previous complexity to get some of these custom solutions built cannot be understated, right? And now with AI assisting people who may not even have any type of tech exposure to create their own automations potentially is in a relatively condensed timeframe. I mean, you could maybe teach yourself how to do it in a week or two or less, right? Is now enabling more people to look into automations at what?

point do you think if someone let’s say we’re a real estate show right if someone is a new realtor a new investor at what point do think it makes sense for them to start thinking about automations

Drew Donaldson (11:13)
I think it makes sense as soon as you have clearly defined processes, right? If you are still changing how you’re doing business on a weekly basis, it’s not going to make a lot of sense for you because you don’t have a core workflow that can be automated, right? So the sooner you can get to a point where this is the way we do business, this is the way we manage deals, these are the steps in the process, then the sooner you can automate. The one thing I will warn people about is that

automation, if you just go and watch like YouTube courses on like N8N or like Zapier or whatever, like it will look very easy. And a lot of automations are very easy, right? If you just want a simple workflow that like updates, you know, every time your HubSpot deal moves from, you know, lead to interested or discovery call booked or like whatever, and you want that to be triggered by a Slack message, like, yeah.

That’s super easy. that you don’t need me to go and build that automation. Just watch a YouTube tutorial. You’ll figure it out. The problem is, is that they have, people have a little bit of success with that. And then they try to tackle really big problems and then things break and they don’t know how to fit it because data structures are simple when things are small and they’re very complex as they grow. And so one of the things that our company brings to the market is like, I’m going to, if your automation is simple,

Dylan Silver (12:17)
Yeah.

Drew Donaldson (12:31)
If I can explain it in 15 minutes, I’m just going to show you how to build it. If it’s something that’s actually requires some knowledge of data engineering and coding and connecting APIs together and all of that, I’m going to say, you this is, you can try to do this, but it’s probably going to take you six months to do something we can do in two weeks. So do you want to learn all this stuff or is your time better spent working on your deal flow? Because chances are it’s better off trying to make money rather than trying to learn a brand new skill.

that I’ve spent 20 years building.

Dylan Silver (13:03)
One of the challenging parts of this is there’s also a compliance element to it, right? I mean, so you can go and you can build an automation that behind a closed wall works, but if you’re then opening it up to the public, now there’s a compliance arm. You potentially need like terms of service and privacy policy and a client facing website for the business that you’re interested in scaling. And so that can also be tricky for people.

Drew Donaldson (13:32)
Yeah, I mean, I’m not a lawyer, right? So like when it comes to compliance in terms of service, like I can tell you, hey, there’s some websites that’ll generate a great one for you. can go to Claude and like ask it to generate one. But like even myself, I go to a lawyer to have that stuff written or I’ll send him like, hey, this is what I, and even contracts, like I will draft the first version, send it to him and he will go and redline half of it and be like, you don’t want to say any of this. And I’ll be like, okay, let’s fix that.

So like, I agree, there’s a lot of things to kind of consider in this. And there’s also things to know in terms of like data privacy that you can and can’t do and things you do have to disclose. ⁓ You know, in the cold email world, one of, because we use a lot of cold email to generate the leads we work. And one of my favorite things is ⁓ what I call like, it’s kind of hard to explain, but essentially it’s like this copy pasta.

of like people going and writing, like, if you ever get a cold email, send them this exact email. And it’s this email that makes absolutely no sense. And it’s like pursuant to GDPR, you have to send me all this data. And it’s coming from a guy in Arizona. And I’m like, we don’t have GDPR in this country. I don’t have to send you anything. Or they do it with CCPA too and all that. But when you do like for CCPA, for example,

If you check the boxes that make you required to report that data, you need to understand how that data was collected. You need to be able to provide the details. And for real estate, businesses can get very big very quickly, not in terms of like headcount, but just in terms of monetary value and revenue. And you can, if you’re operating in California, that can become a problem, right? That can become a big reporting issue for you.

Dylan Silver (15:00)
Yeah.

Drew Donaldson (15:16)
And so one of the things I’ve seen a lot of people do is just pull out of California if they don’t need to be in it. they’re making, you know, and this is across businesses, if they’re making enough money in other States and they don’t see the need to be in California just for the data compliance reason, they just don’t target it. Now, I think that’s leaving a lot of money on the table. mean, there’s a lot of money in California, but it’s something you need to consider. Like, do you want the headache of having to deal with CCPA requests or don’t you? And I think this is one of the things that

you know, the government in this country really should prioritize as much as like they prioritize kind of setting this national framework for AI is setting a national framework for data privacy. Because right now it’s this kind of thing that if a state passes a law, it’s almost a default law for the rest of the country to follow in a lot of cases, which then becomes a nightmare when you’re trying to market nationally.

Dylan Silver (16:03)
Yeah.

People, especially in real estate, are talking about keeping it human, right? And I think there’s a counterargument to this actually, which is maybe you need to keep certain things human so that you can hear like inflection tone of voice, this type of thing. But I think most emails even, particularly, people are reading a subject line and then skimming through it. It may have some pertinent information in there.

What do you think really should stay human versus what should be automated?

Drew Donaldson (17:22)
So I think one of the areas that AI is never going to be able to fully penetrate, at least not until we reach a point of singularity, which is years in the future, I would suppose, ⁓ is the relationship aspect. Because you can see how some of the sycopency that like ChatGPT 4.0 brought to the marketplace and all of the bad things that happened because the AI would just agree with everything and be like, yeah, you are being watched by the FBI and driving people crazy.

That’s not a real relationship. And I think for the most part, people can see through it. Now there are some people that are dating an AI and like that’s a whole new thing. And what does that mean for the world? ⁓ But I think the core business is that if I am a buyer of something and the only touch point I have with your company is an AI and I know it’s an AI, there’s a level of trust that’s lost there, right?

because I’m saying, why won’t you come on, do a call and meet with me? Why won’t you sit across the table for me? Now that could be because of my age, right? I think one of the things we’ve seen is my core audience tends to be like Gen X and Baby Boomers. That’s kind of been my bullwork for a long time. If you go one generation forward into the millennial and Gen Z, they don’t want to do sales calls. They want to buy everything automatically.

they probably aren’t going to care about being sold by an AI versus a human. So I think these things are changing generationally rather than, well, AI can touch this, it can’t touch this. I think it’s just over a long enough period of time. And we’ve actually already seen this in the marketplace where an AI buyer and an AI seller have crafted a together, right? That’s already happened. So that’s, think,

the, like, I think in the future, when we have AI just managing all of our inboxes and only flagging stuff that requires some kind of input from us, I think you’re going to start seeing people just say like, Hey, I need someone to fix my roof. Go call all of these roofing companies and get some guys out to do the quote and then get all the quotes together and analyze them and tell me which one I should go with. And like that whole process will be done between two agents, a buyer and a seller agent.

And then the only human part will be the guys actually on your roof replacing it. at least for the time being humanoid AI is, I think we’re going to be probably 10 years or more before we really see that take off.

Dylan Silver (19:47)
Bonus question here about voice. I’ve seen voice evolve over the last year or so and it’s gotten to a point where I feel like it’s very good in some instances and still not great in other instances. Do you see voice improving to a point where people can do some of these implementations like the one you mentioned, know, call roofers, et cetera, and get me some quotes? Are we close to that at this point?

Drew Donaldson (20:10)
So in terms of like going and getting quotes from companies, I think we’re already there. I mean, I think it’s just a matter of understanding how the technology works and having an agent orchestration platform that can run those type of processes, ⁓ which is still hard for most people. There’s no turnkey. Like, know, the open clock came out and then all the spawns of open clock came out back in February and March. And we’re still seeing new ones pop up today. But in order to run those platforms, you need to have a significant level of technical expertise. is not the kind of thing that like

Your average Joe is just going to walk into and know how to do. So I think the, the big driver of the adoption of that is just going to be having an application that is as easy to install as any other application on your desktop and as easy to use as your email client, your word processor, what have you. When we get to that point where everybody can use it, then I think that kind of outreach to sellers is going to become the standard because of why would you waste time?

getting quotes and talking to a salesperson when you can have an AI ask all the questions and the AI doesn’t have to feel pressured into a sale because if the AI can’t buy anything, you know, and then it becomes on the seller side, like, well, why am I having a human do these if it’s talking to an AI? Like I should just hire my own AI to sell to the AI. So I think it’s going to be one of those things that like, as that becomes more in vogue, that’s where you’re going to start seeing the change.

The voice AI in terms of the sellers right now, I think where it really makes sense to use is when you have constant calls that are all about the same kind of thing. So like think of a dentist office that’s really busy that, you know, maybe they’re small, they can’t afford to have someone sit at the front desk. Implementing an AI answering service that just gets all the details of the person and books the appointment on the calendar makes total sense. If you’re running some kind of like medical clinic,

where people are asking for records requests or want to talk to a doctor, AI is probably a little too inhuman for that kind of thing. A restaurant, great example. Why, I don’t need to talk to a human to, you know, that’s gonna put me on hold for five minutes while she seats the party of five and gives them, you know, menus. Let me just talk to an AI and say, I want a reservation Tuesday at 7 p.m. So that’s where it’s

Dylan Silver (22:10)
Yeah. Yeah, I think.

The restaurant

is a really good example because if, I mean, everyone has this experience, you’re sitting down and you’re waiting to be served type of thing, and people could be busy, you know, literally juggling plates while they’re trying to get other people seated. And so if you have that speed to service, that makes a world of difference. I mean, someone could just get up and leave, or if they’re waiting to be sat at a table, that would be an issue, right?

We are coming up on time here, Drew. Any new projects that you’re working on and then as well, what’s the best way for folks to reach out to you or your team?

Drew Donaldson (23:01)
So I’ll answer them in reverse. So if you want to talk to me, I’m on, LinkedIn is great. You can just find me, Drew R. Donaldson. You’ll see this face and that’s how you’ll know it’s the right one to click on. I’ve had a beard for like 20 years. So like, don’t have to worry about this being like my winter look. So I look pretty much the same, just slightly younger. In terms of if direct outreach, [email protected] is the best way to reach me through email.

In terms of projects, we actually do have some internal projects that we’re really excited about. One is a new kind of imagining of how email clients should work on desktop. And so we’re developing that first for PC, then Linux, because Linux is catching fire right now. And then we may do Mac, but I’m a little bit more of a loyalist to PC and Linux. we have that coming out, that software is called Letters by Ernest.

There’s another product we have called Sendus, which is really a organic marketing platform, which is a really unique take of combining SEO, social media, and newsletters all into one platform, all run by a team of agents that are specialized in all of the specific roles that would fill an organic marketing team. And that is something that we published a first version of a headless version at the beginning of the year.

It’s, it works great. We’re now building the headed version. about two months into that process. So we’re looking for like a summer launch of that product. So depending on when this comes out, that may be out or not. But if you’re interested in being a beta tester for that software and, you know, giving it a spin, I’d encourage you to reach out and, you know, I’d be happy to talk about either of those projects. We have a couple other top secret things we’re working on the background, but those, those are a little bit more rough around the edges. We’re, we’re still figuring out exactly what they are.

So, but those are the two that are the closest in our pipeline to being published and excited about.

Dylan Silver (24:55)
Drew, thank you so much for joining us today. Thanks for your time.

Drew Donaldson (24:58)
Yeah, thanks for having me, man. Great conversation.