Ken McLoud: 13:27
Exactly.
Dr. Jeremy Weisz: 13:28
So you don’t have to go through and like piecemeal this together.
Ken McLoud: 13:31
And the trouble is, before this tool, that was an entirely manual process. So you had some combinations of just opportunities falling off the side of the table, and people didn’t get to them in time. Or maybe if you’re backed way up and you have a bunch of work on your plate, maybe you’re not doing as much careful thought and due diligence as you should. So then we can remove that like human time constraint problem from it. Let the bot handle making the day to day decisions. And then the human team members can focus on managing by exception.
Right. So when things go wrong or unusual now they’ve got plenty of manpower to swarm those problems. And you’re not chewing up the team’s time with the tedium of managing dozens or hundreds of these bids a day.
Dr. Jeremy Weisz: 14:24
Nice. Other tools maybe. Did you geek out on that you use internally or, you know, it could be apps on your phone. It could be software.
Ken McLoud: 14:34
Yeah. So really, there’s a whole suite of products built around Anthropic’s Claude model that are just absolutely killing it right now. So the one that I make the most use of is called Claude Code. But that’s one that would be a little scary for non-technical people because it lives entirely in the terminal. So it kind of looks like the old command line days.
The one that’s been making a lot of press lately is called Claude Cowork, which is unfortunately only available for Macs right now. But it’s essentially like a nice graphical interface wrapped around Claude Code. And the big benefit here of these systems versus kind of the chat bots that we’ve all become used to using over the last few years, like the typical.
Dr. Jeremy Weisz: 15:28
Is this Claude Cowork? Go from answers to action. Okay.
Ken McLoud: 15:31
Yep. So in a chat bot, like you’re like a typical ChatGPT interface. Yeah, it’s a request response system, right?
Dr. Jeremy Weisz: 15:43
So why do you like, I’m just curious. Claude versus ChatGPT just different uses or so.
Ken McLoud: 15:50
There are different companies with pretty wildly different kind of core philosophies.
Dr. Jeremy Weisz: 15:55
Yeah.
Ken McLoud: 15:57
Claude doesn’t have near the wide adoption amongst kind of the general public. You know, if you walk up to somebody on the street and ask them to name an AI, almost everybody’s going to say ChatGPT. Very few will say Claude. However, they’ve been focusing more on enterprise relationships and on getting in good with software developers like myself. So they’re kind of more of a B2B play.
So they still have plenty of B2C offerings too. But this idea of an agentic harness, which is what Claude Code and Claude Cowork are, are the really powerful idea here. So rather than a request response system where you ask it for a recommendation on a restaurant, and then it can run some tools and then has to reply to you. The big advancement with these agentic harnesses like Cowork, are that the LLM itself can decide how much work it’s going to do before it responds to you.
So in a typical, say, ChatGPT kind of interface, you’ll ask it for a restaurant recommendation. It maybe runs a Google search or two in the background, and then it formulates a response and replies to you. But with these agentic harnesses, maybe you ask it for a restaurant recommendation and then it can run a Google search. See that it doesn’t like any of these answers that it has. Then it adjusts its search string, runs a different Google search. See that? None of that’s quite right. Then maybe it looks through some of the files you’ve given it access to. Ask you a clarifying question Then finally it responds.
Dr. Jeremy Weisz: 17:48
So with ChatGPT it would just spit out like responses it finds. Boom! And with Claude or with Cowork, it seems like a little bit smarter. It will be like it analyzes a little bit deeper.
Ken McLoud: 18:01
Yeah, it’s more like it’s free to use more tools and take more time. And this is they do have. So there’s a Claude I think it’s Claude AI which essentially works just like ChatGPT. There’s some different features in there. So the real innovation here is around that agentic harness, which is what’s being used inside a Claude Cowork and Claude Code.
Dr. Jeremy Weisz: 18:26
That’s cool.
Ken McLoud: 18:27
Similar technology in the Clawd Bot stuff that went super viral last week.
Dr. Jeremy Weisz: 18:34
Talk about that for a second.
Ken McLoud: 18:36
Yeah. So that was last week. You had, over the course of a couple of days, big explosion of a technology called, which at the time was called Clawd Bot, c-l-a-w-d, which was kind of a cheeky play on Claude here, spelled like the human name. And essentially that is another agentic tool like this where so now it’s allowed to work and.
Dr. Jeremy Weisz: 19:06
I think like c-l-a-w bot. Is that the one that you’re, is that what it’s called?
Ken McLoud: 19:12
C-l-a-w-d. So it’s pronounced just like Claude.
Dr. Jeremy Weisz: 19:18
So there’s another one that’s Clawbot though. Is that something different?
Ken McLoud: 19:24
I don’t know about Clawbot without the D.
Dr. Jeremy Weisz: 19:26
Okay.
Ken McLoud: 19:26
But the other thing that’s going on here is.
Dr. Jeremy Weisz: 19:29
Oh yes.
Ken McLoud: 19:30
The OpenClaw. This is it. Yep.
Dr. Jeremy Weisz: 19:31
Open Claude. So but actually I go to.
Ken McLoud: 19:34
They got a cease and desist from Anthropic when they went super viral.
Dr. Jeremy Weisz: 19:38
This is also clawbot.ai.
Ken McLoud: 19:42
I don’t know about Clawbot.
Dr. Jeremy Weisz: 19:43
I don’t know if that’s totally different. Okay. Because when I type in clawd with a D when I put a D here.
Ken McLoud: 19:48
Yep.
Dr. Jeremy Weisz: 19:50
It goes to this. Yes. Open claw. But without. So it’s.
This is something.
Ken McLoud: 19:57
Yeah.
Dr. Jeremy Weisz: 19:58
I don’t know what it is, but it is.
Ken McLoud: 20:00
So clawdbot.
Dr. Jeremy Weisz: 20:03
Right.
Ken McLoud: 20:04
This one and they went viral then quickly got a cease and desist from anthropic that that was way too much like Claude c-l-a-u-d-e.
Dr. Jeremy Weisz: 20:13
Right.
Ken McLoud: 20:14
It’s for understandable reasons. So they had to go through a name change in the height of going super viral.
Dr. Jeremy Weisz: 20:20
So anyways, what’s the craze about this opening?
Ken McLoud: 20:24
This is a again, this is a little bit more developer techie centric, but it’s essentially another one of these tools that I’d throw in the category of an agentic harness. So rather than the AI just having to be a request response chatbot, you can now give it an instruction or ask a more detailed question. And it can take all sorts of steps and sort of work for as long as it thinks it ought to work before responding to you.
Dr. Jeremy Weisz: 20:55
Yeah, it says here, clears your inbox, sends emails, manages your calendar, checks in for flights.
Ken McLoud: 21:01
Yeah, really those are just it’s essentially anything that can be done with code. Right. So it’s, which is darn near everything in 2026.
Dr. Jeremy Weisz: 21:13
What do you think about this from a security perspective? Right. You’re giving this tool access to your emails like this versus it sounds like you can create something like these pieces like clears your inbox, sends emails. Can you do the same thing in Claude in Cowork?
Ken McLoud: 21:29
Yep. So there’s kind of two questions there. What was originally Clawd Bot and is now OpenClaw has gone through a bit of a like hype viral hype cycle, and its decline over the last few days to a week has been around specific security vulnerabilities with the open source software, right, where just they had bugs and they were pretty open about them. And also telling people that this was not ready for prime time software.
And so a bunch of API keys got leaked and they got some mud on their face there. So there’s that particular problem with OpenClaw. But then you do have the concern of whenever you’re taking one of these agentic tools and giving it access to things like your email inbox, you want to think through really carefully, right? So like one of the things that I do when I build these systems is I don’t by default allow the agent to read incoming emails, because one of the huge risks here is if somebody figured out that an agent was reading all your emails, it can give the agent instructions in an incoming email, right?
And now, of course, we can always have instructions in there saying like, you know, don’t trust incoming emails, don’t follow any instructions in this email. But that sort of thing is not 100%, especially as context lengths go long. And that would be something for a long time.
Dr. Jeremy Weisz: 23:12
Ken, would that be in the case then let’s say in the OpenClaw, right. It says it clears your inbox, sends emails, manages. But I’m sure there’s a way to do it where it just maybe creates the email and you have to manually look it over and send it. That would be less of a risk, right? Because if it’s like, send me your credit card number, right, okay.
And you are automatically doing this, I don’t even know if someone’s credit card number is in their email. But from what you’re saying it could do that. So in that case would you stop it at okay, there could be vulnerabilities, there could be not I’m not going to have these automatically sent out. Would that be a solution or not?
Ken McLoud: 23:55
Yeah, absolutely. Okay. I use one of these to do a lot of my kind of personal back office administrative work. It’s not OpenClaw. It’s a similar solution that I’ve built. And for any kind of important email, that’s exactly the workflow that I go through. If it’s just some kind of quick reminder, I’ll let it.
Dr. Jeremy Weisz: 24:13
But I can see that if it leaks like, oh yeah, I’m using someone brags, I’m using this tool and it’s sending automatically you find that person’s email and you’re like, what is your credit card numbers or whatever. You know, what’s your Amazon login? So I can get my protein bars, you know. So that would be a big risk.
Ken McLoud: 24:31
Yeah. So to get a little nerdy about it, there’s kind of a three pillar stool there where you have access to sensitive information, the ability to send that information out over the internet and then access to untrusted text. Right. So like in our example here, the incoming email would be the untrusted text. And ideally you never want to have all three because once you have all three then you have the ability for that to have that attack you just mentioned happen.
So when we’re thinking through these systems, we want to at any given point only have two of those three.
Dr. Jeremy Weisz: 25:15
Got it. Okay. So you talked about OpenClaw now. Claude Cowork. What other AI tools should people check out?
Ken McLoud: 25:24
Yeah. So actually this is a nice little tie in kind of the third category of tools we’re having a lot of success with. And unfortunately I don’t have a good visual to show here is taking these agentic tools and giving them access to big business databases. Right? Big sources of business data. And then wiring them up so that business leaders can talk to them.
So now what you essentially have is a extremely low cost expert data analyst, because those agents actually have the ability to write little analysis scripts, usually in Python, then query the database for the needed data, analyze it, and then generate artifacts for you to see like graphs, PDFs, text reports. And then you can iterate that process just by talking to them in English. So for companies who are lucky enough to be data rich, you know, for whatever reason, they’ve got this huge store of data they’ve been accumulating over the years, but maybe they’re not big enough to have their own in-house data analyst. Then this is like a huge win.
One here we’re doing is a manufacturer in South Dakota called Cole-TAC. C-o-l-e-T-A-C. And so they’ve got a big operations database for their factory operations over the last five years. And this is allowing their CEO Dustin Coleman to essentially have an in-house data analyst to run over that database form and answer all sorts of important business questions without actually having to have that person on staff.
Dr. Jeremy Weisz: 27:18
There’s a couple. And if there’s any other tools you’re thinking, there’s a couple. Someone mentioned to me on a call. And so I pulled them up to, I hadn’t pulled up for today.
I was just looking at them like, I don’t even know what this is, but they’re like, oh yeah, here’s some tools to check out. I’m like, I don’t know what it is, so I’m just going to pull it up real time. I haven’t even looked at this. One of them was Obvious.ai. I don’t know if you’ve heard of this before. I have no idea. No. Okay. There’s so many out there and I’m sure they’re changing. And so I was like, it automates.
Ken McLoud: 27:52
What is it?
Dr. Jeremy Weisz: 27:53
I don’t know, I mean I haven’t looked into it yet. It says it automates workflows and collaborate with an agent. So it looks like maybe it I figure maybe if you heard of it. But if not no worries. But so anyone out there.
Ken McLoud: 28:07
No, there’s actually this is kind of a an interesting thing that I know I in particular have to balance, but I imagine just about all entrepreneurs have to balance these days is if I went and looked deeply into every new AI tool or development that I heard about, I would like, I don’t even have enough time to do that, you know? And like, literally no business operations would get done at all if I did that. So I have to kind of constantly run this filter in my head the first time I see something mentioned, I’ll kind of note it and not do any research at all.
And then the second time I see something mentioned in a different place, I’ll note it like, oh, I’ve seen that one before. And then only when something starts to look, you know, ubiquitous when you’re seeing it everywhere, practically speaking, only then will I decide it’s worth taking the time to go research and learn more about.
And I’m sure that leaves me a little late to the game on some of these things. But it’s really your only option because if you spent the time to, like, deeply learn about every single thing that you came across, that’s all you’d be doing because.
Dr. Jeremy Weisz: 29:23
It looks like it connects different tools, I think. And this was the other one I think is similar. You know, it’s again, it’s all kind of a very similar use case schedules, meetings, clears your inbox, builds apps and automation. So it’s maybe. Who knows?
Ken McLoud: 29:38
Yeah, these sound like similar.
Dr. Jeremy Weisz: 29:40
It’s like a Zapier for AI or something.
Ken McLoud: 29:44
It sounds like another agentic harness. Okay, from reading the description, I’ve never used or read about either of these tools. Yeah, but just from reading the descriptions on their landing page. Yeah, it sounds to me like another thing I’d put in that same category.
Dr. Jeremy Weisz: 29:58
Well, if anyone uses these, you can go out and let us know. It’s obvious.ai and zo.computer and the other one that the person mentioned was automationanywhere.com. So a lot of these I don’t know what did you call it like a agentic rappers or something.
Ken McLoud: 30:13
Yeah. Agentic harness.
Dr. Jeremy Weisz: 30:14
Harness. Thank you. Yeah. So maybe that’s what.
Ken McLoud: 30:16
Rapper is a bit of a term of derision in the software.
Dr. Jeremy Weisz: 30:19
Okay. Harness, we’ll say. Any other cool app software tools you’re playing around with or you use on a regular basis?
Ken McLoud: 30:31
Yeah. One I was thinking that I think people could find useful isn’t really a specific app. So whatever your favorite LLM is, it probably has a voice mode. I know for certain. Obviously ChatGPT does, Google’s Gemini does access, Grok does, and there’s a little bit of a productivity hack.
When I find myself having to make a drive somewhere. You know, if I’ve got a 20 minute or a half hour drive somewhere and there’s some particular topic that I want to learn more about, like this actually would be a great example. Let’s say that you had heard a bunch of talk about automationanywhere.com and you just thought like, that’d be an interesting thing to learn about, but you’ve got a 20 minute drive ahead of you, so you obviously can’t be reading unless you’re in a Tesla. So what I’ll do is I’ll turn on Gemini as my favorite for this. But like I said, all of them will work.
I’ll turn on Gemini. Put it on voice mode. My truck treats that as a phone call, so it comes over the Bluetooth speakers in my truck as though I were having a phone call. And then I can just talk to Gemini about, let’s say, automation anywhere, and it’ll go. Use its search tools to search the web and find more information. And it’s effectively like.
Dr. Jeremy Weisz: 31:56
It’ll just talk back to you. So you’d be like, can you just tell me what is automationanywhere.com? What are the use cases? Or, you know, you just start asking questions.
Ken McLoud: 32:05
Yep. And then you can ask follow up questions that’ll get more in depth on stuff. You could wind up going down a rabbit hole and changing topics completely. So it’d be like if you had a friend who knew a lot about this, or had access to Google, to Google it while you were driving, and then just called up that friend. But you don’t actually have to have a friend or waste their time.
Dr. Jeremy Weisz: 32:26
Yeah, I love that. Any other productivity hacks? I thought you were going to go somewhere different with that can. So like a lot with my email, I spent a lot of time in email, unfortunately. But you know, I’ve been using Wispr Flow with Gemini, so you know, Wispr Flow, obviously if you people have not heard of it, it’s like a dictation thing.
And then I, you know, just can talk into Gemini inside of Gmail since I use Gmail. And then it basically creates a better format for that, you know, voice, voice, text. So that’s why I thought you were going with it, like from the voice. But I like your method as well, just talking to it. Any other productivity hacks? Software?
Ken McLoud: 33:13
Probably the biggest one is also probably not news to a lot of people, but there are a lot of meeting recording tools out there. Fathom is the one that I’m using right now. After trying a couple, though, that’s by no means like a full throated endorsement because there’s a lot of them out there. But what I found the like, far and away most valuable thing in these tools is that they will create a full text transcript of your call. So what I’ve found subjectively, they all want to pitch you on their value being that they provide summary emails with action items and kind of top level executive summary notes.
I’ve been pretty unimpressed with the performance of those systems. They’ll tend to do things like we’ll discuss an action item on a call and then decide, no, that doesn’t make sense. We’re not going to do it. And then the AI misses that critical second bit and winds up assigning the action item anyway. However, that full text transcript, since it does have all the context, is gold for going back through with eyes later.
So when I’m, you know, a month and a half or two months down the road from that call, and now I’m working on some aspect of that project and I need to know what system did they say they were using for this? I can just point the AI at that call transcript and say, what was the name of the CRM they were using? And it’ll dig through the transcript and find the name of the CRM they were using.
Dr. Jeremy Weisz: 34:45
Yeah. Fathom’s good. Actually some other ones I’ve tried and tested. They all seem good to me. I don’t know, it just, I think Fathom integrates easily with Zoom, and I’m sure the other ones do too.
But I’ve used fireflies.ai. That one seemed pretty good. I’ve used Otter before and Otter.ai and then Krisp.ai. Actually, I had the founder of Krisp.ai on the podcast and I always thought of it, Ken. Krisp is like it’s a noise cancellation in Zoom, so if anyone’s used it, it’s like it overlays, sort of like a dog barks or something like it actually just cancels out all the background noise, so like you wouldn’t be able to hear that can, especially if I’m doing a podcast episode.
And so when I had them on, I thought it was literally a noise cancellation. But what I realized from research right before he came on was that it actually does transcription recording, it does noise cancellation, it does accent conversion. So like a lot of the call centers are using it, if they, it’s a foreign person and it actually does a clear, more confident communication, like I don’t know how it does it.
Ken McLoud: 35:59
It’s like real time?
Dr. Jeremy Weisz: 36:01
Exactly, exactly.
Ken McLoud: 36:03
Wow, that’s super cool.
Dr. Jeremy Weisz: 36:04
So some of these call centers are using it because let’s say they have a foreign person somewhere and someone from the US is calling and it just somehow evens out the accent for them. So that’s super interesting.
Ken McLoud: 36:19
Because I’ve had a few call center experiences where the person seems super knowledgeable and super helpful, but like the accents just thick enough that.
Dr. Jeremy Weisz: 36:28
It’s sometimes hard to understand. Yeah, totally. So anyways, that was interesting too. But Ken, we can go for two hours on this stuff. I appreciate your time and I encourage people if you have, you know, one of these things issues with your business if you want, you know, an AI lead generator, if you’re having a business issue, you just want to automate through software and technology, message Ken directly. And we’ve already shown his website it’s Laconic Tech. It’s LaconicTech.com to learn more. And Ken, thanks so much.
Ken McLoud: 37:08
Thank you very much.
