Episode Summary
In this episode of Leap to Scale, Justin Davis and Greg Ross-Munro are joined by Doug Smith, Director of AI Operations at Element 451, for a wide-ranging discussion about the emerging AI operating model for service businesses. Drawing on experience spanning finance, private equity, consulting, education technology, and AI operations, Doug explains why companies are sitting on enormous amounts of untapped operational data and how modern AI tools are making that data useful for the first time. The conversation explores “data exhaust,” hidden business signals, and how organizations can begin building practical automation systems without needing massive engineering teams or enterprise budgets.
The episode also dives into the rapid evolution of AI tooling, including OpenClaw, AI agents, workflow automation, and prototype-driven product development. Doug shares stories about using AI voice agents, automating operational tasks, and rapidly validating software ideas before turning them into production-ready products. Along the way, the group discusses hallucinations, security risks, the changing role of human expertise, and why speed may become the defining competitive advantage in the AI era. It’s a practical conversation for leaders trying to understand how to operationalize AI inside real businesses without getting lost in hype or abstraction.
Episode notes
- AI has dramatically reduced the cost of prototyping, experimentation, and rework.
- Showing someone a working prototype can produce better feedback than spending hours discussing an abstract idea.
- AI models are powerful, but hallucinations and overconfidence still make human review important.
- Most businesses have far more useful data than they realize, often spread across CRMs, accounting tools, conversations, and operational systems.
- Creating consistent identifiers across business systems can make fragmented data significantly more useful.
- Existing customer data can help businesses identify lookalike prospects and new growth opportunities.
- AI agents can rapidly validate product ideas before a company invests in building production-grade software.
- Businesses should prioritize automation that affects revenue or eliminates significant recurring time and toil.
- Sales research, CRM updates, meeting notes, account preparation, and customer call analysis are strong early automation candidates.
- LLMs can interview business owners and employees to uncover inefficiencies, undocumented processes, and tribal knowledge.
- Future operating dashboards may measure successful human-agent interactions, AI recommendation adoption, and time returned to employees.
- As automated outreach creates more digital noise, trust, reputation, relationships, and human interaction may become even more valuable.
Episode Transcript
Introduction
Greg Ross-Munro:
Doug, thank you for coming and giving us your time.
Justin and I started this show primarily for our clients, who are mostly in professional services. They’re consultants, trainers, engineers, and similar firms.
They’re coming to us because they’re trying to productize their services, which has always been what we’ve done. But AI’s ability to work with fuzzy information has really accelerated that opportunity.
Based on your background, we want to talk about a few things.
What signals tell you a business is ready for automation? What level of maturity do they need?
Where is data hiding inside businesses? A lot of businesses don’t think they have much data, but normally they have tons of it.
What should they do first? How should they think about humans in the loop? And what does the future look like?
Before we start, is there anything you particularly want to plug? Element451?
Doug Smith:
We can always say Element451 is solving all the AI problems of higher education, which we are.
It’s pretty cool what we do.
Doug Smith’s Background
Greg Ross-Munro:
Today we’re talking with our friend Doug Smith.
Doug and I graduated together from the same Master of Science in Entrepreneurship and Applied Technologies program.
Doug has always stood out to me as somebody who thinks one layer deeper than everyone else, especially me.
That’s true whether we’re talking about politics, crypto, or technology. But never in that fluffy growth-hacker kind of way.
Like Justin and me, Doug is someone who never seems to accept that the way something works today must be the way it continues to work.
Officially, he’s the Director of AI Operations at Element451, where he focuses on putting AI into real operational systems.
What makes Doug especially interesting for our audience is the path he took to get here.
He started in finance and strategy, moved into operations improvement, then data science and AI consulting, and now AI operations.
He’s also very involved in the Tampa Bay technology ecosystem. He shows up, participates, and sends me more papers and AI tools than I can possibly keep up with.
He’s not a programmer at his core, but he’s a hacker in the classic sense. He’s someone who wants to tinker with systems and understand how to make them work differently.
Doug, thank you for coming on.
Doug Smith:
I appreciate it. I always enjoy chatting with you all whenever I get the opportunity.
Greg Ross-Munro:
Usually we do it over beers or while we’re yelling about something. This time we have microphones in front of us.
Tell us your origin story. How did you get to where you are today?
Doug Smith:
I started in finance. My academic background was economics. I did a lot of trading while I was in school and afterward.
Eventually I ended up in a really interesting position with a venture capital firm, looking at deals and new ideas.
I loved the newness of that world. There was this sense that the world could become a better place because people were working on really thorny problems that could save time, lives, resources, and so on.
Then I worked with a private equity group pursuing some of those ideas in education.
You start by looking at the numbers and understanding how these companies make money. Then you move to the next layer. How do they stay open? How do they actually solve the problems students have?
You start digging into those systematic challenges, including things like Maslow’s hierarchy of needs.
How do you figure that out?
Data.
So I moved into data and started understanding how it could be applied. I had an economics background, but it was one of those things I didn’t fully use until several years after school.
You start looking at regressions and building regression models. You identify signals associated with why a student might not be successful.
Then you say, “Great. We’ve figured out who may not be successful. What can we do to remediate that?”
Now you’re working in systems, processes, change management, people, and technology.
That’s really where my path went.
I’ve always loved super ambiguous problems that people say have never been solved.
Someone has tried it before, but I tend to think there’s probably a way to solve it if we approach it differently.
I’m a big proponent of looking at how other industries solved an equivalent problem and asking whether that approach can be applied to a more traditional business.
Reclaim and demonstrate. Don’t necessarily do hardcore R&D.
Someone has probably figured out your problem somewhere else, in another use case.
That’s one of the beautiful things about academia. There is a huge amount of research that may already tell you how to approach your problem.
Now you amplify that with LLMs, and it’s addictive.
As Greg says often, if you step away from AI for a week, it feels like an entire year has gone by.
It’s fantastic to see.
I’m simultaneously extremely excited about it and aware that it is going to disrupt a lot of people’s lives.
People can now accomplish things that were previously thought impossible, or that would have required hundreds of thousands or millions of dollars and enormous teams.
Now they’re doing those things themselves with agents.
Does that mean we don’t hire those people anymore? Do we stop solving certain problems the old way? Or do we just keep doing more?
I don’t know.
But that’s how I got here.
Turning Domain Expertise Into Systems
Greg Ross-Munro:
How did you get from private equity and finance into education?
Doug Smith:
I went into one of the portfolio companies and stayed there, then moved into consulting.
I became the person who could look at things differently and help highly competent people who had enormous domain expertise but didn’t know how to translate that expertise into technology and data.
They knew how to solve the problem manually, but not necessarily how to solve it at scale.
Maybe they could brute-force it by pulling spreadsheets together and spending countless hours figuring it out.
Then you take that Access database or collection of spreadsheets, put it into SQL, automate it with scheduled jobs, and produce the report automatically.
Instead of five managers spending their entire day building it, the system does it.
Your Business Probably Has More Data Than You Think
Greg Ross-Munro:
That brings up something we encounter frequently.
Companies know there is probably hidden value in their data, but a lot of them don’t think they actually have much data.
In reality, there are operational signals everywhere.
Justin Davis:
Kind of data exhaust.
Doug Smith:
Exactly.
There are little artifacts of data sprinkled around. Sometimes you have to massage them to make everything come together.
But almost every company has this.
You have clients. Maybe it’s a handful, maybe it’s thousands.
Are there signals that indicate why someone might leave you?
If you have a product, maybe the product emits data.
If you don’t have a product, maybe something else acts as a key signal.
Your accounts receivable can be a signal. Someone stops paying you.
If it’s a restaurant, maybe the same people used to come in every week and then suddenly stopped.
Why did they stop?
What’s the root cause?
Often you begin simply by having those conversations, then asking, “Could I standardize the way I collect this information?”
Now you’re creating a layer of higher-quality data.
That’s one of the biggest challenges every organization has, from giant corporations down to mom-and-pop businesses.
Think about how you want to collect data so you can understand how, what, and why something happened.
You also want some kind of unique key.
If it has to be something rudimentary like an email address, phone number, or physical address, that can work.
Ideally, create something more deliberate and use the same unique identifier across multiple systems.
Greg, Justin, and Doug are customers. Give each one a unique key.
Put that key in QuickBooks, your inventory system, your product, and whatever else you use.
Now you can associate the work.
If Doug follows you on Facebook, maybe that’s connected too.
You can begin to discover that certain customers are influencers or that certain behaviors correlate with other outcomes.
There are a lot of hacky things you can do.
The biggest challenge I often see, though, is that organizations have a very strong belief that their data is good.
Greg Ross-Munro:
People actually believe that?
Doug Smith:
Absolutely.
They’ll tell you they have great data architecture and great data streams.
Then you walk in and they can’t answer basic questions.
How many people are using this particular product or feature?
“Well, we can figure it out in Mixpanel.”
Then they spend five hours creating a one-off query.
That’s difficult if you’re an account manager trying to understand how someone uses the product. Account managers shouldn’t need to be experts in querying data.
The interesting opportunity now is that if you can gather the data and combine it with task knowledge, you can start translating the expertise of founders, directors, and managers into something repeatable.
Greg, I imagine you have a very different approach to handling an angry client than some of your staff.
How do you codify that and translate it?
Greg Ross-Munro:
I try to make them laugh. Once they’re chuckling, they get less angry with you.
Doug Smith:
Exactly. There is a process there, whether you’ve documented it or not.
Data Exhaust Is Becoming Valuable
Justin Davis:
What’s interesting is that a few things can be true at the same time.
Some organizations have a naive belief that they have lots of great, well-organized data.
But there are also companies that don’t realize they have data at all.
They don’t think about the fact that their invoicing system contains a huge amount of information.
Previously, using that data was difficult.
You had to know how to access it. You needed SQL, integrations, APIs, and technical expertise.
Then you had to know how to analyze it. You needed somebody who understood regressions, k-means clustering, and all of that.
Now you have tools that can help with those things.
So the exhaust that used to be mostly useless because it was difficult to do anything with has suddenly become an extraordinarily valuable asset that almost every company is sitting on.
I think that creates a wave of data-governance work.
We effectively turned a bunch of useless garbage into gold, but somebody still has to refine it.
That feels like a coming layer of services and products.
Greg Ross-Munro:
Doug used to do that job.
Doug Smith:
Yes, and it’s much easier to do now.
Here’s a useful example.
Who are your clients?
You know their names. You probably have their job titles and companies.
You could go to something like Clay and say, “Find me lookalikes in this geographic area for new business.”
Here’s my ideal customer profile. Here’s my customer list.
Find people who look like them.
Now you find LinkedIn profiles and email addresses. You can very quickly scale up your prospecting.
Another interesting example is trades businesses.
A lot of trades have overlapping client bases.
The person who needs lawn care might also need hardscaping.
Greg Ross-Munro:
This is why Justin and I keep talking about lawn care.
Justin Davis:
All the time.
Doug Smith:
Exactly.
Someone comes in to take care of your lawn, but they may also know how to paint, do light carpentry, fix a fence, do hardscaping, or clean exterior surfaces.
That’s a great upsell.
Land and expand.
And they already have the customer’s address.
They can use mapping or imagery to understand the property without even driving there first.
They may be able to estimate additional work before arriving.
A Temporary AI Advantage
Justin Davis:
There’s a dislocation in the market right now.
We’re recording this in spring 2026, and there is an insane amount of power available for nearly free that most people either don’t know about or aren’t using.
If you lean into it, and if you’re listening to this podcast you probably are, you can build a sizable advantage while that dislocation exists.
Eventually the market will catch up, but that takes time.
Greg Ross-Munro:
And by then, you’re already ahead.
Justin Davis:
Exactly.
Doug Smith:
You’ll be a survivor.
The ability to stay nimble is one of the core advantages here.
What Is OpenClaw?
Doug Smith:
People have gotten very excited about OpenClaw, and I understand why.
Greg Ross-Munro:
We should probably explain what OpenClaw is.
Doug Smith:
OpenClaw is an open-source framework you can install on a computer or virtual machine.
You connect it to an LLM, such as ChatGPT, Claude, or Gemini, using an API key.
Then you can have conversations with it and ask it to build or perform things for you.
That may include web apps, workflows, task processes, and other technology.
A lot of the “What if I could do this?” ideas you have can become conversations with OpenClaw, and it will try to execute them.
Will it always use the most efficient approach? Probably not.
Will it be hacky? Yes.
Could it create security nightmares? Yes.
Would I recommend installing it directly on your personal computer? Questionable.
Greg Ross-Munro:
I’m going to say no unless you really know what you’re doing.
Doug Smith:
Don’t install it directly on your personal or corporate computer right now.
AI Agents as Rapid Prototyping Tools
Doug Smith:
Without dismissing the security risks, which are very real, the unlock has been remarkable.
I have friends in commercial development.
They need to understand where deals might exist, which parcels are available, and what is happening in planning meetings.
These are smart people who know how to do the work, but doing it manually is expensive and time-consuming.
With an agent, they can ask it to build a prototype.
Maybe it listens to planning-board meetings across counties in a target state.
Maybe it maps potential properties along an interstate using GIS tools.
That work would previously have been incredibly expensive.
Now they validate the concept first.
Then they start asking, “Could I turn this into a product and sell it?”
That’s where you have to harden it.
Now you go through the traditional SaaS process.
It needs to be secure. It needs to scale. It needs to be repeatable and reliable.
That’s where a company like Sourcetoad can help turn the prototype into a robust solution.
But you’ve reduced some of the market risk because you’ve already demonstrated that the thing solves a real problem.
That’s what I love about tools like OpenClaw, Replit, and Lovable.
You can validate a concept extremely quickly.
AI Agents in Everyday Life
Doug Smith:
OpenClaw even solved a Valentine’s Day problem for me.
There are hundreds of thousands of skills for these tools.
One connected to restaurant reservation systems. Instead of repeatedly checking for an opening, I told the system, “Keep checking. If a reservation becomes available, book it immediately.”
It worked.
That solved a huge pain in the ass.
I’m also extremely lazy and didn’t want to call 11 door installers.
So I had a voice agent make outbound calls and explain the time window when I wanted someone to come.
People showed up.
Three of them had an interesting experience because the agent decided to speak to them in their native languages.
One thought I spoke Turkish. Two thought I spoke Russian.
Then they met me and were very confused.
Greg Ross-Munro:
A guy comes to your house to install a door, realizes you don’t speak Russian, and now you have to explain OpenClaw and voice agents to a Russian-speaking door installer.
Doug Smith:
Exactly.
Justin Davis:
It would be even weirder if the voice agent were trained on your actual voice.
Then the person would say, “I’ve heard you speak Russian. Why are you telling me you don’t?”
Doug Smith:
I intentionally didn’t do a full voice clone. That starts to feel too dystopian.
What Should a Business Automate First?
Greg Ross-Munro:
If you’re a service business and you have some data, or you’re beginning to collect it, where do you start?
What’s the first automation or system you should build?
How do you identify the low-hanging fruit?
Doug Smith:
Start with dollars and cents.
What’s going to bring in new dollars? And what’s going to make sure you get paid?
Understand how a customer ended up with you.
Maybe you have a marketing employee or agency.
Have you ever asked them to pull the data together?
Take that information and pass it into an LLM, ideally through a secured business account.
Start asking where customers are actually coming from.
Then you understand where you’re spending money and time in the front office.
If you have sales representatives, you can begin analyzing the work they’re doing.
A friend of mine uses an LLM-based solution to review calls and transcripts from his team.
What questions are customers asking?
What themes are appearing?
What’s actually happening in these interactions?
Previously, doing that at scale was nearly impossible for a human.
Nobody is going to listen to 10 to 40 hours of calls every week for every employee unless you hire people specifically to do that.
An LLM can.
Now you can understand why a client is angry.
You can summarize the reason.
You can identify patterns.
Summarization of customer interactions is extremely useful.
Revenue Versus Cost Reduction
Greg Ross-Munro:
So you’re suggesting starting toward the top of the funnel rather than choosing an internal business process to automate.
I’ve historically thought the opposite.
I’ve tended to ask, “What’s burning the most time that these people could spend doing something else?”
Doug Smith:
Time and toil absolutely matter.
But as you trace the revenue process, you often discover where the time and toil are.
Take a traditional sales role.
Where is the time spent?
Prospect research. Finding information. Documenting everything afterward.
One of the biggest time sucks is documenting the outcome.
Can you automatically populate Salesforce, HubSpot, or Notion based on a conversation?
That’s time an employee no longer has to spend manually entering information.
Justin Davis:
It’s like writing meeting notes.
Doug Smith:
Exactly.
Now you start creating a data pipeline.
Here’s the conversation.
Here’s how this person became a client.
Here’s what’s important to them.
Now you have a better picture.
If you have account managers, they can understand where the client came from, their aspirations, and how they use your service.
As you continue through the process, you’ll see the really massive time sinks.
Account preparation for an account manager is one.
Account preparation for a salesperson is another.
Drafting quarterly business reviews or monthly business updates can consume enormous amounts of founder and executive time.
Can you automate some of that?
You know it’s coming every month.
But here’s my warning.
Don’t automate something that takes one or two hours per month.
Automate the thing that takes one or two hours per day, or something consuming five to ten hours per day across the team.
That’s where you start seeing real impact.
Think about the classic matrix of high impact, low effort.
Let an LLM Interview You About Your Business
Doug Smith:
If you’re asking, “Where should I start?” one practical approach is to go into ChatGPT, Claude, or another LLM and tell it to interview you about your business.
Ask it to act like a business coach or expert and grill you.
Greg Ross-Munro:
What would you actually prompt it with?
Doug Smith:
Give it context.
“I’m John Doe. I run a hardscaping business in Florida. I have X employees. We make X amount of revenue. I want to grow. Here’s what my business does today.”
Then ask:
“Help me understand what I should know and what I should be thinking about if I want to grow this business.”
Ask about personnel management, human capital, investment opportunities, data you should be tracking, and each part of the organization.
That’s one of the magical things about LLMs.
You have access to an enormous amount of encapsulated knowledge and ideas.
But you need to tell the model who you are and what context you’re operating in so the advice becomes actionable.
Context is the foundation of almost everything we’ve discussed.
The better the data you put into the LLM, the better the answer tends to become.
If you’re building an AI agent, export your key reports and give those to the agent as context.
Now it knows what fields you have and what kind of data exists.
It can help perform a data audit.
Where are you missing information?
Where is information malformed?
Maybe your data isn’t actionable because of formatting.
An agent can potentially normalize phone numbers, emails, or other fields for you.
Founder Mode as an AI Coach
Doug Smith:
One example I built was based on Paul Graham’s essay “Founder Mode.”
I took the essay, put it into a custom GPT, and told it to act like Founder Mode.
Talk to me about what I’m doing.
Give me an action plan for one thing I should do every day.
Tell me what I should ask employees, vendors, and people around me.
It becomes a coach that keeps pushing you forward.
That’s one clear way to use an LLM for scale.
Greg Ross-Munro:
So everyone goes to Paul Graham’s blog, which still looks like it’s from 1999.
Justin Davis:
Which is almost a credibility indicator on the web now.
Greg Ross-Munro:
It is kind of cool.
Go find the Founder Mode essay, put it into an LLM, and ask it to interview you using those ideas.
And once you’ve done the interview, save the output somewhere.
Save it as a text or Markdown file.
Those answers may be important enough that you want to use them with a different model later.
Doug Smith:
It’s all about context.
I’ll share the GPT with you.
Greg Ross-Munro:
Send it over. We’ll put it in the show notes.
Capturing Tribal Knowledge With AI
Justin Davis:
You made me think about another use for the interview concept.
Go to your internal employees and interview them about their jobs.
Record it.
Ask them:
“What do you do?”
“What’s the worst part of your day?”
“What are the tricks you’ve learned?”
“What do you know that other people don’t?”
You could capture tribal knowledge through interviews, then have LLMs parse those conversations and generate process ideas.
That feels like a sneaky, high-fidelity way to generate a lot of ideas very quickly.
You could probably do it in an afternoon over lunch.
Doug Smith:
Yep.
The CEO Dashboard of the Future
Greg Ross-Munro:
Let’s get more future-focused.
If you were CEO of a service company, what would the ideal operations dashboard show?
What will it track in the future that most businesses aren’t tracking today?
Imagine the dominant hardscaping company two years from now is crushing everybody else.
What are they tracking on the CEO dashboard that competitors aren’t?
Doug Smith:
A major one will probably be human-to-agent interaction.
Did the interaction result in a successful resolution and positive outcome?
You may also track the percentage of AI-agent recommendations that the organization adopts.
In theory, the agent is collecting enough information that it can surface novel recommendations.
So you may compare human-developed solutions with AI-generated solutions.
Another thing you’ll probably track is ROI in terms of human unlock.
How much time did your staff avoid spending on nonproductive work?
How much time did AI give back to them?
Greg Ross-Munro:
So you’re talking about feedback loops.
Doug Smith:
Exactly.
I think CEOs and managers are going to become orchestrators of agents and AI workflows.
You’ll still interact with humans, especially in service industries.
Unless you have drone swarms flying around cutting down trees.
Greg Ross-Munro:
I’m on it.
Doug Smith:
I think there’s going to be enormous demand for productive uses of drones.
Greg Ross-Munro:
Someone still has to convince another human being to buy my drone fleet instead of somebody else’s.
Unfortunately, that still requires people.
AI Could Make Business More Human
Doug Smith:
Text messages, emails, LinkedIn messages, and other channels are already getting drowned in noise.
I think the nature of sales is actually going to become more human.
The tide went one way. I think it’s going to come back.
Justin Davis:
Almost out of necessity.
The automated systems may choke themselves out by creating so much noise that we go back to having beers with people because the digital channels become too crowded.
Doug Smith:
Exactly.
And if agents increasingly do research for you about who you should work with, social proof becomes even more important.
Google reviews become more important.
Better Business Bureau information becomes more important.
Those are artifacts and signals that agents can evaluate.
I’m very excited about the future.
I’m particularly excited for people in service and trades businesses.
If English is your second language, you may suddenly have access to a high-quality AI receptionist who can answer the phone professionally for you.
Then you show up and do the work.
That’s an accessibility improvement.
Or maybe you’re extremely introverted and hate doing certain interactions.
You can have agents make some of those calls on your behalf.
They can also do the work you simply don’t want to do.
“Go interface with this vendor.”
“Make sure I get paid.”
“Pester them until the invoice gets resolved.”
Doug’s Recommended AI Tools
Greg Ross-Munro:
Doug, thank you so much for spending time with us.
Every time we talk, I walk away with new ideas.
And then there are the ten new AI tools you send me every week that I can’t keep up with.
Doug Smith:
That’s my job.
Justin Davis:
Before we finish, give us your top three tools people should look at.
Doug Smith:
If you work in an office and deal with a lot of data, and you haven’t tried Claude for Business or Enterprise with Cowork, you’re missing something.
Cowork is incredible for people who need a data analyst but don’t necessarily have the technical chops themselves.
You do need connected systems and useful data.
You need a CRM that actually contains information.
But it can do great work crunching data, conducting research, and helping people analyze information.
I also think ChatGPT is going to come back around in a major way.
The broader integration of agentic capabilities into these platforms is going to be significant.
OpenClaw is another logical thing to explore and tinker with.
It’s closer to the world people have imagined where you ask something to solve a problem and it actually goes and does the work.
I have a friend who spent many nights trying to program his home-automation system manually.
Then he solved the problem in about 15 minutes with an agent.
That’s the type of shift we’re talking about.
For voice agents, check out tools in the Vapi category as well.
The larger idea is that you can increasingly have something like a personal assistant doing work for you.
Uber effectively allowed ordinary people to have a personal driver on demand.
AI is beginning to give ordinary people access to personal assistants, programmers, analysts, and other capabilities that used to be expensive.
And right now, many of those capabilities are heavily subsidized, so they’re dramatically cheaper than they’ve ever been.
A Great Time to Be Curious
Doug Smith:
If you have a young kid, there are even tools like Gemini’s storybook functionality.
You can describe the story you want and it can create the imagery and read it aloud.
Greg Ross-Munro:
I built a version of that two years ago.
I spent a lot of time building the whole thing.
I thought I was a genius.
Doug Smith:
That’s one of the biggest challenges now.
The idea in your head has probably been executed by somebody else in some capacity.
Sometimes they’ve done it better.
But there are also very specific ideas that only matter to a tiny group of people.
I’m on the board of the Florida Holocaust Museum, and I’ve worked on GPT experiences around historical figures and pioneers connected with St. Petersburg.
You can create something that lets people have a conversation based on the writings and known history of a person.
I did something similar for one of my great-great-grandfathers.
Who cares about that?
Maybe five members of my family.
But now those five people can interact with something that probabilistically represents what that person might have said.
That’s the fascinating part.
Greg Ross-Munro:
Last question.
Is there a book you’re going to write?
Because I’d read it.
When is the Doug Smith book coming out?
Doug Smith:
I’ve been thinking about that.
I should probably just force my agent to help me knock it out.
At the end of the day, if you’re somebody who’s curious and likes to learn, I don’t think there’s ever been a better time to be alive.
The things I’ve learned through this technology are incredible.
I’ve even learned things I never thought I’d care about, like the proper way to mop a floor and which chemicals to use.
How else would I have gotten those answers so easily and at scale?
And when you mop, you don’t want to mop yourself into a corner.
Greg Ross-Munro:
And don’t install OpenClaw directly on your everyday computer.
Doug Smith:
Correct. Don’t install it casually unless you understand what you’re doing. Use an isolated environment.
Justin Davis:
I, for one, welcome our new overlords.
Greg Ross-Munro:
I’m going on the record that I’m still waiting for the Doug Smith book.
Doug, thank you for being gracious with your time. I really appreciate you.
Every time we talk, I come away with new ideas.
Thank you for the conversation.
Doug Smith:
I appreciate you all. Thank you.
Justin Davis:
Thanks, Doug.