WEBVTT
00:00:02.399 --> 00:00:09.359
Remember back in 2024 when knowing how to use a custom GPT was a differentiator?
00:00:09.679 --> 00:00:18.879
Or in 2025, when knowing the difference between a skill and a project and Claude could be a competitive advantage for your company's AI efforts.
00:00:19.120 --> 00:00:21.120
Well, those days are gone.
00:00:21.280 --> 00:00:30.559
And so is most of the useful lingo because controlling front-end chatbots and models and modes are now table stakes.
00:00:30.719 --> 00:00:35.520
The differentiator now is controlling long-running desktop agents.
00:00:35.679 --> 00:00:48.799
And it sounds easy in theory until you realize that the language you've been building up on the front-end AI chatbot era and the language of the long-running autonomous agents are not exactly compatible.
00:00:49.039 --> 00:01:02.000
So today we're giving you the primer and establishing some baseline vocabulary and concepts you'll need heading into 2027 when long-running AI agents become the new norm.
00:01:02.240 --> 00:01:04.959
So here is the big picture.
00:01:06.719 --> 00:01:11.439
Agent vocabulary is now an essential skill set.
00:01:11.599 --> 00:01:16.799
And yeah, it's changing all the time, which is one of the reasons why we do this thing every day.
00:01:16.959 --> 00:01:32.959
But agents can obviously read and write files, they can run tools, call their own, create their own apps and plugins, they can fix mistakes and work for hours unattended, which is both a good and a bad thing, depending on uh how active you are in your agents.
00:01:33.120 --> 00:01:40.079
And every new term now names a problem you only hit once an agent run runs long.
00:01:40.239 --> 00:01:40.799
Uh right.
00:01:40.879 --> 00:01:46.799
So so so much of the previous uh terminology that we use with AI chatbots, right?
00:01:46.879 --> 00:01:52.400
You had an instant feedback loop for yourself, whether it worked or not, and that is kind of gone.
00:01:52.480 --> 00:01:53.840
You really have to be paying attention.
00:01:54.000 --> 00:02:06.959
And not knowing the words now, well, means you might set a vague goal, you might put up weak guardrails, and that run that you think might fix a problem as you go take a walk, might not really go anywhere.
00:02:07.120 --> 00:02:15.280
And learning this new agentic language is becoming as essential as learning how to prompt, as that was helpful.
00:02:15.439 --> 00:02:24.719
Uh, so on today's show, we're going to learn what loops, goals, plans, and subagents actually mean without the jargon.
00:02:24.879 --> 00:02:33.520
Uh, we're gonna go over how all of these terms work in Codex and Claw Desktop so you can know how these features work together.
00:02:33.599 --> 00:02:47.759
Uh, you're gonna learn why fluency in these terms is the skill separating operators from spectators, and you're gonna know the mental model that makes the whole vocabulary click starting now.
00:02:48.080 --> 00:02:50.560
All right, if you're new here, welcome.
00:02:50.719 --> 00:02:53.120
This is the Start Here series.
00:02:54.319 --> 00:03:12.159
The Start Here series is part of everyday AI's ongoing effort to give whether you're a new listener or a seasoned AI expert to give everyone an essential podcast series to both learn the AI basics and double down on your AI knowledge.
00:03:12.319 --> 00:03:14.639
So if that's what you're trying to do, sweet, me too.
00:03:14.719 --> 00:03:15.439
Let's do it together.
00:03:15.599 --> 00:03:17.599
Make sure to go to starthirseries.com.
00:03:17.759 --> 00:03:31.840
That's going to give you exclusive access to our inner circle community and you're going to be redirected straight into our Start Here Series space, which has a playlist uh for all of these episodes, all in order.
00:03:31.919 --> 00:03:40.000
So it's easy to uh go through them all, as well as um you can read about them all on the page and connect with other people who are along the journey with you.
00:03:40.159 --> 00:03:53.360
So if you missed our last Start Here series, that was volume 29, where we talked about the open source surge and if models like GLM 5.2 uh make open source an enterprise priority.
00:03:53.520 --> 00:03:58.560
But today we are talking about the desktop agent lingo simplified.
00:03:58.800 --> 00:04:03.280
So first let's zoom out completely.
00:04:03.439 --> 00:04:03.919
All right.
00:04:04.159 --> 00:04:16.079
So if you are not someone that's using codex or clawed desktop or anti-gravity or cursor, uh, some of this might not make a ton of sense.
00:04:16.160 --> 00:04:16.959
And that's okay.
00:04:17.199 --> 00:04:20.959
And maybe this show is more for you than anyone else.
00:04:21.040 --> 00:04:31.920
Uh, because if you are using something like Codex or uh Claw Desktop every single day for hours, this episode will probably be a review at best, but I think still helpful.
00:04:32.000 --> 00:04:39.519
Uh so if you're like, okay, I don't use these tools, no, you need to listen up uh and you should start using them.
00:04:39.680 --> 00:04:51.040
Uh, but I want to talk about the shift, obviously, from the uh AI chatbot that's just very reactive versus the proactive autonomous desktop worker.
00:04:51.120 --> 00:04:54.399
Um, and really what separates them is the harness, right?
00:04:54.560 --> 00:05:05.199
So uh sometimes when you talk about a model, you talk a lot about the harness or where it lives and how it accesses how it how they access all of these tools, uh right.
00:05:05.680 --> 00:05:09.680
So let's just use an example codex from OpenAI, uh, right?
00:05:09.839 --> 00:05:11.120
It's my favorite harness.
00:05:11.199 --> 00:05:13.680
It's the one I use most, but a lot of people don't even know.
00:05:13.839 --> 00:05:16.959
You can use other models inside codecs.
00:05:17.120 --> 00:05:19.199
Codex is the harness, uh, right?
00:05:19.279 --> 00:05:24.160
You can't do that with Claude Code, you can do that with anti-gravity, codecs and some others, cursor as well.
00:05:24.319 --> 00:05:34.160
Uh, so the harness is how all of these uh agenc tools come together and they can work over long term, uh right.
00:05:34.240 --> 00:05:39.439
When you see all these stories, you're like, oh, this person had an agent running over the weekend or overnight.
00:05:39.600 --> 00:05:47.680
Well, that's really made possible by the harnesses that these companies create that give the models and the tools essentially a sandbox to do all this.
00:05:47.839 --> 00:05:55.759
And you know, it's kind of uh today's episode is really a primer on understanding what happens under this, the hood of the harness, so to speak.
00:05:55.920 --> 00:05:57.759
So we're not gonna get super technical.
00:05:57.920 --> 00:06:04.319
This is uh for beginners and maybe intermediate people as well, but that's what we're truly trying to understand.
00:06:04.560 --> 00:06:13.680
And like I said, even if you're not using these desktop tools now, I think most people will be using them uh come 2027, right?
00:06:13.920 --> 00:06:18.399
We've heard from Microsoft that they're coming out with their super app, uh, right?
00:06:18.480 --> 00:06:23.199
We tackled uh super apps uh on the show and on the start here series.
00:06:23.360 --> 00:06:24.560
Uh, what episode was that?
00:06:24.720 --> 00:06:25.600
Bringing that up here.
00:06:25.759 --> 00:06:32.079
That was our episode uh 799 or volume 28 of the Start Here series.
00:06:32.240 --> 00:06:34.560
So, you know, Microsoft is bringing out a super app.
00:06:34.639 --> 00:06:38.319
So these super apps uh or these agentic harnesses, right?
00:06:38.480 --> 00:06:48.319
If you want to get a more technical term, essentially they are a much more powerful version of a web-based chatbot that can run autonomously on your computer.
00:06:48.399 --> 00:06:50.800
Uh, you can run them in loops, you can run goals.
00:06:50.879 --> 00:06:53.600
We're gonna go over all these things, you can schedule automations.
00:06:53.839 --> 00:07:00.240
So the big difference is well, it can act autonomously, it can share the context uh across different chats.
00:07:00.560 --> 00:07:04.560
Um, and it can read and write uh to your computer.
00:07:04.639 --> 00:07:11.279
So any file, just like a human really would, it can use your actual computer with computer use, it can use your browsers, right?
00:07:11.360 --> 00:07:13.040
Uh, has built-in browsers.
00:07:13.199 --> 00:07:21.759
So uh, you know, definitely go listen to the uh AI super apps episode uh if you want a little bit more primer, but let's get straight into it now.
00:07:22.079 --> 00:07:24.160
So let's talk about plans.
00:07:24.480 --> 00:07:34.240
And I think plans are actually one of the more underrated features of these harnesses, specifically in Claude Code and in Codecs, right?
00:07:34.319 --> 00:07:46.079
Uh every once in a while I'll ask people how they're using uh these different harnesses, and I people don't really talk about or use plans as much as they should.
00:07:46.240 --> 00:07:52.160
So a plan essentially just reveals the route before an agent asks uh or acts.
00:07:52.319 --> 00:08:00.879
So a plan shows the intended steps before, you know, the files, the tools, the app changes all of those things, and planning exposes the assumptions, uh, right.
00:08:01.040 --> 00:08:04.240
So likely files, approval points, and verification steps.
00:08:04.319 --> 00:08:11.199
So plan modes matter because desktop agents can change work fast and they can work for a very long time.
00:08:11.360 --> 00:08:11.839
Uh, right.
00:08:11.920 --> 00:08:15.600
So it's kind of like uh a blueprint for a building, right?
00:08:15.759 --> 00:08:22.160
You wouldn't just, if you had unlimited resources and you wanted to build a building, you wouldn't just go to someone and be like, hey, go build a building.
00:08:22.240 --> 00:08:25.920
Or if you're building a custom house, you wouldn't just be like, hey, I want a house and make it awesome.
00:08:26.079 --> 00:08:26.480
Uh right.
00:08:26.560 --> 00:08:34.720
You'd probably sit down with an architect and you would go over the blueprints, the floor plans, uh, zoning restrictions, requirements, all those things, right?
00:08:34.879 --> 00:08:44.240
It's might sound tedious, but you're probably gonna get a much better result in the long run if you sit down and have the conversation with said architect or uh general contractor, right?
00:08:44.320 --> 00:08:44.879
Whatever it is.
00:08:45.039 --> 00:08:46.639
That's the same thing a plan is.
00:08:46.720 --> 00:08:48.159
It is literally a plan.
00:08:48.240 --> 00:08:58.080
Uh so a lot of people will usually just point uh, you know, clawed desktop or codex uh to a folder or share a little bit of context and say, get to work, buddy.
00:08:58.159 --> 00:08:59.279
Uh, not a good idea.
00:08:59.440 --> 00:09:01.919
So the plans are essential.
00:09:02.159 --> 00:09:08.799
So codex, right, uses uh the kind of plan, pair, and execute as collaboration gears.
00:09:08.879 --> 00:09:16.080
Uh so it reads, it analyzes, and then it proposes things to you, and then it waits for you to approve it before implementation.
00:09:16.240 --> 00:09:20.320
Uh so similarly, uh, that's how Claude Desktop works.
00:09:20.480 --> 00:09:26.320
Uh, it actually works out nicely that these work very similarly in Claude Desktop and in Codex.
00:09:26.480 --> 00:09:32.000
Uh, so if you're actually using these and if you want to see like how does this work, where's the button, right?
00:09:32.159 --> 00:09:34.000
Codex has a plan button.
00:09:34.080 --> 00:09:41.519
I don't believe Claude Desktop has it, but you can invoke them each the same way, which is just a backslash and then a plan, uh, right.
00:09:41.600 --> 00:09:44.639
And then they'll kind of walk you through and then you will approve the plan.
00:09:44.799 --> 00:09:57.919
Uh, so in the same way, how I am a very uh adamant telling people like, make sure you read the chain of thought after you've gone through uh, you know, and and gotten an output out of a large language model.
00:09:58.080 --> 00:10:01.600
I am that much um in favor of using plan mode.
00:10:01.679 --> 00:10:03.279
It is the equivalent on the front end.
00:10:03.360 --> 00:10:03.840
Uh, right.
00:10:03.919 --> 00:10:12.000
So if you've never used these harnesses, think of something like Google's uh Google Gemini's deep research actually does a really good job of this, right?
00:10:12.080 --> 00:10:20.559
Uh before you go out and do a deep research with Google Gemini, it actually gives you a plan to approve so you can see it and if something's wrong, you can modify it.
00:10:20.720 --> 00:10:32.799
And this is really important because when we talk about the the shift from token maxing to token efficiency, right, depending on how you set these agents up and what kind of task you're giving them, they might run for a long time.
00:10:32.960 --> 00:10:38.960
And you know, if you are using these on a company plan, chances are you're paying via the API.
00:10:39.120 --> 00:10:48.080
Uh so not having a plan in the same way telling uh a builder to build you an amazing house can be a very expensive step to overlook.
00:10:48.320 --> 00:10:50.320
Plan plan mode, you have to think.
00:10:50.399 --> 00:10:51.519
It's not slowing you down.
00:10:51.600 --> 00:10:54.799
A lot of people are just like, I want to build, I want to break stuff, I want to burn tokens.
00:10:54.960 --> 00:10:56.879
No, it's a guardrail, right?
00:10:57.039 --> 00:10:58.080
Not a slowdown.
00:10:58.159 --> 00:11:06.559
So planning separates thinking from doing, uh, but it's not, you know, it's not enough because risky work still needs, right?
00:11:06.639 --> 00:11:09.919
That read-only access, the work trees, the checkpoints.
00:11:10.080 --> 00:11:16.639
And then once your route is approved, you know, then the permissions kind of define the real blast radius.
00:11:16.799 --> 00:11:29.360
Uh, so don't just rush into giving a long-running agent uh keys to the castle or you know, giving it access to just run for hours and burn through your API bill.
00:11:29.679 --> 00:11:32.960
All right, next, let's talk about goals.
00:11:33.120 --> 00:11:33.600
All right.
00:11:33.759 --> 00:11:38.240
And I I kind of put these in an order that I think might make sense for most.
00:11:38.399 --> 00:11:40.960
Um, I generally will start with a plan.
00:11:41.039 --> 00:11:49.440
And the way I actually do it is I use a plan and then uh parlay uh the findings of that into a goal.
00:11:49.600 --> 00:11:54.159
So it's a little confusing and uh Claude Code and Codex work a little bit differently.
00:11:54.240 --> 00:12:03.679
Uh uh in for plan mode, it will still work through your plan, but a lot of times it will stop once it's gone through the steps.
00:12:03.840 --> 00:12:16.480
Uh so the biggest difference between a plan and a goal is on the front end, uh, a plan literally just outlines it and you can see uh codecs or Claude code work through each step, and there is a visual indicator, which is great.
00:12:16.639 --> 00:12:17.840
Goal is a little bit different.
00:12:18.080 --> 00:12:24.799
Goal is you literally give it an end goal and it will not stop and in sometimes loop until it hits that goal, right?
00:12:24.879 --> 00:12:44.320
So uh a lot of times you have to be careful using goal if you don't go through a plan mode first, or if you don't go through an old school best practice prompt engineering, context engineering, like we used to teach with Prime Prompt Polish, to really share that context and have an understanding with the model of what ultimately the output looks like.
00:12:44.480 --> 00:12:46.960
So if you just blindly, and a lot of people do this, right?
00:12:47.039 --> 00:12:53.919
They just blindly throw a bunch of context at Codecs or Claude Desktop and they give it a goal and it will keep going until it hits that goal.
00:12:54.080 --> 00:13:07.120
And if that goal is maybe unattainable, uh, yeah, that's where you can have it work for uh hours on end, or maybe a day or longer, and all of a sudden, yeah, your uh your your bills through the roof.
00:13:07.440 --> 00:13:10.480
So uh let's talk a little bit about goals.
00:13:10.559 --> 00:13:14.080
So it's really defining the finish line before you start any motion.
00:13:14.240 --> 00:13:17.679
And like I said, I always go through a planning phase first.
00:13:17.840 --> 00:13:26.000
I usually make the plan kind of big uh to work through it, uh, let you know, Codex or uh Claude go through and do some work.
00:13:26.080 --> 00:13:31.039
Then I'll see, you know, there's always gonna be shortcomings the first time you give something uh something to someone.
00:13:31.279 --> 00:13:33.039
So I'll see where went wrong.
00:13:33.120 --> 00:13:38.639
I'll go through, read the summarized chain of thought, then I will kind of reprompt it and then make that as a goal.
00:13:38.799 --> 00:13:39.039
All right.
00:13:39.200 --> 00:13:48.399
And sometimes I'll literally just copy and paste the original plan, make some modifications, obviously, because some things on the list or on the steps of the goals are gonna get crossed off.
00:13:48.559 --> 00:13:49.519
Some things aren't.
00:13:49.600 --> 00:13:56.559
Uh so I usually do a lot of manual nitpicking, uh, transitioning from a uh a plan to the goal.
00:13:56.799 --> 00:13:59.919
But the goal is the outcome and agent checks across the long run.
00:14:00.080 --> 00:14:11.279
And strong goals specify the audience deliverable source material in the done condition, which is why I think it is usually best in general terms for you know, general knowledge work.
00:14:11.360 --> 00:14:13.919
And again, I'm coming at this from like a general knowledge work.
00:14:14.080 --> 00:14:26.000
Yes, I do some coding software dev stuff, building myself, uh, you know, cool projects and uh uh the uh products, extensions, all those things, but I am talking about this through general knowledge work.
00:14:26.080 --> 00:14:33.919
So even if you're not, you know, writing software or you know, vibe coding something, let's just say your vibe working, I still think that this holds true.
00:14:34.159 --> 00:14:46.000
Um, so weak goals make plans, loops, and subagents drift quietly, which I think it's important to go through, especially if you're talking about uh a project that you want to do, right?
00:14:46.080 --> 00:14:54.639
If you're just trying to knock off a uh a task that might normally take you or a single agent on the web, you know, five, 10 minutes to go through.
00:14:54.799 --> 00:14:58.000
I don't think you necessarily have to go through the plan and the goal in combination.
00:14:58.159 --> 00:15:16.240
But if you're talking about knocking out an actual project, which is what these systems are capable of doing right now, that might take a human two, three, four, five hours or two, three, four, five days, that's when I think it uh, you know, when you're talking about your ROI and what you have to invest on the front end.
00:15:16.399 --> 00:15:21.360
This is one of those things where you're gonna have to build a bridge in order to save the four days driving around the mountain.
00:15:21.519 --> 00:15:26.879
Uh, so goals differ a little bit uh across Codecs and Claude.
00:15:27.120 --> 00:15:33.039
So Codec's goals are a little bit more persistent and editable and also visual in the composer.
00:15:33.120 --> 00:15:41.039
There's a nice little toggle uh button in the goal first threads uh in codecs, kind of keep the objective across turns and sessions.
00:15:41.200 --> 00:15:46.320
Uh Clawed code uh via the desktop version, not the uh CLI.
00:15:46.480 --> 00:15:50.000
It does still support the backslash goal in the same way.
00:15:50.080 --> 00:15:53.120
Uh, but uh cowork is a little bit differently.
00:15:53.200 --> 00:15:54.399
Uh works a little bit different.
00:15:54.480 --> 00:15:55.039
That's the other thing.
00:15:55.200 --> 00:16:04.240
People uh, you know, and for me, I personally use uh, you know, codecs, and I do have to tell people this because if not, it would be extremely irresponsible of me.
00:16:04.399 --> 00:16:10.080
Uh, you know, the big difference between codex and claw desktop is clawed desktop is fragmented.
00:16:10.240 --> 00:16:13.360
There is a chat, a co-work, and a code tab.
00:16:13.600 --> 00:16:16.320
And those tabs have no clue what the other is doing.
00:16:16.399 --> 00:16:27.200
So there's, you know, uh certain things that work really well in cowork goal, uh, such as the visual indicators that don't always work as well in the code version of goal.
00:16:27.440 --> 00:16:45.279
So, you know, even as we go over individual features, uh, because clawed code by default, uh or sorry, clawed desktop by default is siloed uh between the chat, the co-work uh and the code, you know, even things like goal work a little bit differently uh in co-work and in clawed code.
00:16:45.519 --> 00:16:51.600
All right, another thing to keep in mind um a goal is not just a step-by-step prompt, right?
00:16:51.759 --> 00:16:56.159
So over-specified steps can actually fight the agent's own planning process.
00:16:56.320 --> 00:16:59.759
So you don't have to know everything going into it, right?
00:16:59.840 --> 00:17:04.559
You don't have to like be like, oh my gosh, I can't use this plan thing and the goal thing because I don't have experience.
00:17:04.640 --> 00:17:05.680
I'm not a software engineer.
00:17:05.839 --> 00:17:06.799
Absolutely not, right?
00:17:06.960 --> 00:17:13.200
You have to be able to communicate and understand what you want and then work with an agent to you know go through the steps to get there.
00:17:13.359 --> 00:17:18.960
But under specified outcomes, make the agent invent missing like success criteria.
00:17:19.039 --> 00:17:21.599
And that's what you absolutely want to avoid.
00:17:21.920 --> 00:17:22.559
All right.
00:17:23.119 --> 00:17:26.720
So now let's talk about loops, right?
00:17:26.799 --> 00:17:36.079
So loops are kind of the uh new-ish trend, uh, right, even though we had the Ralph loops, you know, many, many quarters ago.
00:17:36.160 --> 00:17:40.559
Uh, now loops are making their uh viral reappearance again.
00:17:40.720 --> 00:17:46.880
So loops uh essentially can turn plans, approved plans, into agent work.
00:17:47.039 --> 00:17:47.519
Uh right.
00:17:47.599 --> 00:17:51.359
So similarly, like a heartbeat if you're using open claw, uh, right.
00:17:51.440 --> 00:17:57.440
So think of it like this it's it's something that you can schedule to happen over and over and over and over again.
00:17:57.680 --> 00:18:02.880
So a loop means you you know the agent is going to observe, plan, act, check, adjust, and repeat.
00:18:02.960 --> 00:18:04.720
It is literally a loop.
00:18:04.799 --> 00:18:12.480
Uh, so a chatbot would run once, while a long-running agent can loop many times if you give it that instruction.
00:18:12.640 --> 00:18:16.880
So the loop only becomes useful when the agent verifies each step.
00:18:17.039 --> 00:18:25.920
Otherwise, you're just burning tokens like you're still trying to climb the uh internal meta um, you know, token burning leaderboard.
00:18:26.400 --> 00:18:29.599
So uh, you know, loop review looks a little bit different in each product.
00:18:29.680 --> 00:18:34.079
So codex uh runs every task loop inside a project organized thread.
00:18:34.160 --> 00:18:40.160
Uh so you know, you can also uh build loops as skills or automations, right?
00:18:40.240 --> 00:18:43.759
So you can just tell them, hey, here's what I want you to do, right?
00:18:43.920 --> 00:18:57.279
Every, you know, every hour I want you to go through, you know, triage my email, my calendar uh and my drive, uh, you know, respond to any emails that I might want to update any decks, and I want you to do this every hour.
00:18:57.359 --> 00:18:59.279
I want you to do this, you know, every day.
00:18:59.359 --> 00:18:59.839
Uh, right.
00:18:59.920 --> 00:19:04.720
So that's an example of a very oversimplified example of what could be considered a loop.
00:19:04.880 --> 00:19:05.039
Right.
00:19:05.119 --> 00:19:07.039
So Claude Cowork, again, a little bit different.
00:19:07.119 --> 00:19:11.440
It shows the steps with citations to the files and the messages, which I really like.
00:19:11.519 --> 00:19:18.880
I like that uh that piece in co-work uh a little bit more than I like it in Claude Code because it has that dedicated side panel on the right.
00:19:19.039 --> 00:19:23.920
So even if you are running technically a loop, you can see those steps visually get checked off.
00:19:24.000 --> 00:19:29.599
Uh Claude Code surfaces the plans, the tasks, the sub agents, the diffs in the progress panels.
00:19:29.759 --> 00:19:36.960
The biggest thing to talk about when uh you know talking about loops is having very clear verification, uh, right?
00:19:37.039 --> 00:19:40.880
Whether you want to run something uh a first time, right?
00:19:41.200 --> 00:19:43.839
I would literally, and I've done this before.
00:19:44.000 --> 00:19:52.000
You can work through, uh just go through a natural language, work with codecs, work with Claude Code, and say, hey, I want to create a loop.
00:19:52.079 --> 00:19:59.599
I want to ultimately save this as a skill earn automation, uh or you know, usually both save it as a skill and an automation that you can schedule and run.
00:19:59.680 --> 00:20:03.599
But say, hey, instead of giving you the full loop, I want to work with you uh through this.
00:20:03.759 --> 00:20:06.319
Here, you know, here's ultimately what I want to do.
00:20:06.400 --> 00:20:11.839
I want to, you know, check this website uh, you know, twice a day.
00:20:12.000 --> 00:20:16.079
Uh, you know, this is my whole company, we live and die by this website, right?
00:20:16.160 --> 00:20:21.599
It could be industry news, could be uh I don't know, stocks, finances, um, etc.
00:20:21.839 --> 00:20:22.000
Right.
00:20:22.079 --> 00:20:31.920
But think if if if you are someone that has to pay very close attention to the market, something in healthcare, I don't know, FDA regulations, and there's all these things pinging all the time.
00:20:32.079 --> 00:20:32.559
Uh, right.
00:20:32.640 --> 00:20:41.440
Maybe you want to create a loop that you train or you help build a skill set with codecs or clawed code that goes in there every so often.
00:20:41.599 --> 00:20:46.480
You know, you see what's new, you run it through your context, your decision-making process, etc.
00:20:46.799 --> 00:20:49.519
Work through those steps in the loop one by one.
00:20:49.759 --> 00:20:59.200
So the um, so this harness or the model in the harness can understand what verification looks like, what success looks like at each step of the loop.
00:20:59.359 --> 00:21:01.920
If not, bad loops, bad, right?
00:21:02.079 --> 00:21:09.519
Bad loops turn into producing uh, you know, overly polished work that just is maybe wrong.
00:21:09.759 --> 00:21:15.440
Because when one loop gets overloaded, then sub agents are maybe going to split the work.
00:21:15.599 --> 00:21:15.920
All right.
00:21:16.079 --> 00:21:17.680
So loops are great.
00:21:17.839 --> 00:21:20.160
You can save them as a skill in automation.
00:21:20.319 --> 00:21:22.400
Think of it similar to like a heartbeat.
00:21:22.559 --> 00:21:27.119
Uh, if you've used OpenClaw, the great thing, all it takes is natural language, uh, right.
00:21:27.200 --> 00:21:31.119
I do think in this aspect, codex is a little bit better uh at setting these up.
00:21:31.200 --> 00:21:40.240
I think that they're uh because not being fragmented, uh, it's a little bit easier to just chat uh with codecs in natural language to set up those loops.
00:21:40.480 --> 00:21:44.319
All right, and then last but not least, we have sub agents.
00:21:44.720 --> 00:21:56.319
So all right, sub agents like loops are where things get very interesting and potentially dangerous if you are not uh being hands on.
00:21:56.640 --> 00:21:59.519
So if you're being uh uh laissez faire.
00:22:00.160 --> 00:22:02.720
Human in the loop and setting all these things off.
00:22:02.880 --> 00:22:08.400
I wouldn't recommend going too heavy and the pains on uh loops and subagents.
00:22:08.480 --> 00:22:09.920
It will get expensive quickly.
00:22:10.000 --> 00:22:13.839
If you are very hands-on, uh, I think loops and subagents are great.
00:22:13.920 --> 00:22:15.519
So here's what subagents are.
00:22:15.680 --> 00:22:21.119
Well, they're just helper agents with focused assignments and separate context windows.
00:22:21.279 --> 00:22:24.799
So they help with parallel work, not just vague requests to think harder.
00:22:24.960 --> 00:22:29.359
The real value, though, is context hygiene before the consolidation begins.
00:22:29.519 --> 00:22:38.240
So subagents control um kind of uh a different aspect of a job.
00:22:38.480 --> 00:22:40.640
So one easy thing, right?
00:22:40.720 --> 00:22:46.559
Uh an example of how I use subagents uh in codecs uh and in Claude.
00:22:47.279 --> 00:22:52.480
So you can just use them um by invoking them.
00:22:52.720 --> 00:22:54.160
Say use sub agents for this.
00:22:54.240 --> 00:22:56.000
A lot of people are like, How do you use sub agents?
00:22:56.079 --> 00:22:58.559
Well, you natural language, use sub agents for this.
00:22:58.720 --> 00:23:03.599
So, as an example, one thing I like to do is I like to build with codecs.
00:23:03.680 --> 00:23:03.920
All right.
00:23:04.000 --> 00:23:05.279
So I build with codecs.
00:23:05.359 --> 00:23:07.440
A lot of people are like, Oh, Claude is better at front end.
00:23:07.519 --> 00:23:09.839
Yeah, it's way better at front end, but I don't know.
00:23:09.920 --> 00:23:15.759
Uh, codecs has this thing called the most powerful AI um image model in the world.
00:23:15.920 --> 00:23:28.880
So start with the uh AI image gen in codecs, uh, use the front end skill, takes three seconds, and you generally will have a much better front end uh if you are building, you know, a piece of software that's just what I'm using as an example, right?
00:23:29.039 --> 00:23:35.200
If you use the image gen in the front end design skill, you're gonna have way better than what you would get out of Claude Code if you do that.
00:23:35.279 --> 00:23:37.680
If not, obviously uh Cloud Code is gonna be better.
00:23:37.759 --> 00:23:46.079
But one reason or one way I use subagents, uh I I love Claude's ability to uh really investigate and use subagents across multiple things.
00:23:46.160 --> 00:23:50.400
So I'll sometimes I'll assign, you know, I'll say, hey Claude, I want you to do these sub agents.
00:23:50.480 --> 00:23:53.759
I want you to look at, you know, one, look at the entire uh repo.
00:23:53.920 --> 00:23:58.079
Number, you know, two, look at the feature set, three, look at the language, right?
00:23:58.160 --> 00:24:07.119
So then each subagent uh can really have a more defined and refined um kind of uh task, right?
00:24:07.279 --> 00:24:08.559
So it's the same thing.
00:24:08.720 --> 00:24:16.160
If you just walked into a room of 20 employees and they're there to help you, and you say, Hey, 20 employees, uh, go help me with this project.
00:24:16.240 --> 00:24:18.559
Here's where we're at, have fun, uh, right.
00:24:18.720 --> 00:24:21.119
They might decide upon themselves, okay, let's split this up.
00:24:21.200 --> 00:24:30.559
But if you say, hey, designer, go look at the design, hey copywriter, go look at the copy, uh, hey engineer, go look at the, I don't know, the security on the back end, make sure everything's you know tightened up.
00:24:30.640 --> 00:24:33.759
So all you have to do is well say, go use sub agents.
00:24:33.920 --> 00:24:46.000
Obviously, these models are smart enough where you can just say that broadly, um, and they will uh, you know, usually uh, you know, assign one sub agent to a specific task.
00:24:46.160 --> 00:24:50.880
Uh, usually I'll just broadly say go use subagents and poke holes in ABC.
00:24:50.960 --> 00:24:52.880
And then I'll see exactly how they did it.
00:24:52.960 --> 00:25:00.319
Uh Claude and Codecs allow you to click uh onto a subagent and see exactly what they're working on in real time, which is cool.
00:25:00.400 --> 00:25:01.759
I love how Codecs names them.
00:25:01.839 --> 00:25:11.200
Uh the Claude uh names, I forgot how they name them, but um, you know, usually I'll do a quick general subagent run if it is a big project that's super important, right?
00:25:11.279 --> 00:25:13.279
That I can spend kind of the token budget on.
00:25:13.359 --> 00:25:17.920
Uh, I'll see what they did, what went wrong, what didn't, uh, you know, how it can be improved.
00:25:18.000 --> 00:25:22.559
And then I'll probably assign roles the second time I have another group of subagents do it.
00:25:22.720 --> 00:25:30.880
Sometimes I'll have sub agents work on the front end before the work actually starts to scope it and make sure it makes sense on where they're spending their time.
00:25:31.039 --> 00:25:35.039
Sometimes I'll have them work on the back end after the work is done, or a combination of the both.
00:25:35.200 --> 00:25:44.240
And the best thing is well, you can just tell Codex or Claude, like, hey, have sub-agents work on the front end before you start this project to make sure everything is correct.
00:25:44.319 --> 00:25:49.440
Uh, then go do the work, then have a separate group of sub-agents really tear it apart on the back end, right?
00:25:49.519 --> 00:25:51.920
That's all you really need to say, and they will do that.
00:25:52.000 --> 00:25:59.440
It is kind of like this having this parallel work stream that can disagree, duplicate, uh, you know, miss shared constraints.
00:25:59.519 --> 00:26:00.160
And that's great.
00:26:00.240 --> 00:26:04.720
And then the main agent can pair findings, resolve conflicts, synthesize.
00:26:04.880 --> 00:26:10.720
I've even set up a skill that has like a three-tier uh kind of uh agentic management system.
00:26:10.799 --> 00:26:12.880
So I don't even have to go through and type this, right?
00:26:12.960 --> 00:26:23.279
So if I know a project is pretty big, or if I'm gonna be working, you know, it's like, hey, I'm setting this off to work overnight, uh, it's it's an ongoing project, and I'm not gonna go through the the planning and goal.
00:26:23.440 --> 00:26:37.119
I know I have this, you know, subagent uh system where I, you know, say, hey, this one's gonna go in and be a contrary uh or uh contrarian and you know, tear apart all this and second guess every single thing that we do before it goes to production, whatever it may be.
00:26:37.279 --> 00:26:41.519
Uh so that's just uh a way to go through subagents.
00:26:41.599 --> 00:26:42.000
All right.
00:26:42.160 --> 00:26:45.920
So now as we wrap up, the real constraint here is actually your machine.
00:26:46.079 --> 00:26:52.559
This is the big thing to keep in mind because your machine has to be on, it has to be uh right, you have to have these programs open.
00:26:52.720 --> 00:26:56.160
So yeah, you can do all these great things like control your computer, right?
00:26:56.240 --> 00:26:57.839
And and control the browser.
00:26:58.000 --> 00:27:07.759
And think of that as well as we talk about uh, you know, the desktop lingo is it's literally as if you were talking to a human sitting in front of the computer, but it has to be on, right?
00:27:07.839 --> 00:27:10.480
A sleeping laptop can break remote steering.
00:27:10.720 --> 00:27:16.240
Uh, you know, if you run out of power, if your internet goes off, everything, well, from an agentic perspective stops.
00:27:16.400 --> 00:27:17.519
Where that's a little bit different.
00:27:17.680 --> 00:27:24.400
The advantage of using on the front end chatbot, uh, those you know, singular tasks, those singular one-off prompts can continue to go on.
00:27:24.559 --> 00:27:27.359
But the beginner mistake is not managing anything, uh, right?
00:27:27.440 --> 00:27:30.480
The beginner mistake is being like, wow, these things are super powerful, these agents.
00:27:30.640 --> 00:27:35.519
I can just go in and you know, drop it some context and say, go do this work and wow, look at it.
00:27:35.680 --> 00:27:39.519
You're gonna get just slop work almost every time, uh, right.
00:27:39.599 --> 00:27:45.200
Because also bad handoffs say research this, organize that, make that better, and then you just use the output.
00:27:45.279 --> 00:27:45.759
Uh, right.
00:27:45.920 --> 00:27:53.119
Good handoffs are well defining the goal, the plan, the permissions, the workflow, the verification, the subagents, the loops, uh, right.
00:27:53.200 --> 00:27:59.680
That's the differentiator now, not just saying, oh, I use codecs, oh, I use Claude Code Desktop, right?
00:27:59.759 --> 00:28:04.880
Uh, the handoff turns the agent lingo into repeatable desktop delegation.
00:28:05.359 --> 00:28:16.400
So, like I said, the new skill is number one, understanding how to talk to your agents, but then using those skills to supervise work, not just prompt, right?
00:28:16.480 --> 00:28:19.359
The agentic layer is growing thicker and thicker by the day.
00:28:19.519 --> 00:28:25.119
That doesn't mean the human layer, uh, both uh, you know, I call it the uh, you know, the agentic human sandwich, right?
00:28:25.200 --> 00:28:25.839
We are the bun.
00:28:25.920 --> 00:28:30.160
We are providing uh the input and the context on the front end, and then ultimately the verification.
00:28:30.240 --> 00:28:32.480
But that doesn't mean we're hands-off in the agentic layer.
00:28:32.559 --> 00:28:35.039
We have to constantly be monitoring, uh improving.
00:28:35.119 --> 00:28:42.480
So we don't just, you know, blindly let these agents loop and uh you know, blindly assign work to subagents, uh, right?
00:28:42.640 --> 00:28:49.200
We can just spend more time uh as the bun and far less time as the meat, to use my old saying there.
00:28:49.359 --> 00:29:03.200
All right, I hope this one was helpful going over the basics of desktop agent lingo, simplifying goals, loops, plans, uh, and sub agents and a little bit of how they work in to in codex and in clawed code on the desktop.
00:29:03.279 --> 00:29:06.480
So if this was helpful, please go to starthirseries.com.
00:29:06.720 --> 00:29:09.519
If you haven't already, make sure to subscribe to the podcast.
00:29:09.680 --> 00:29:10.880
I would really appreciate that.
00:29:11.039 --> 00:29:11.839
Thanks for tuning in.
00:29:12.000 --> 00:29:14.720
Hope to see you back tomorrow and every day for more everyday AI.
00:29:14.799 --> 00:29:15.680
Thanks, y'all.
00:00:02.399 --> 00:00:09.359
Remember back in 2024 when knowing how to use a custom GPT was a differentiator?
00:00:09.679 --> 00:00:18.879
Or in 2025, when knowing the difference between a skill and a project and Claude could be a competitive advantage for your company's AI efforts.
00:00:19.120 --> 00:00:21.120
Well, those days are gone.
00:00:21.280 --> 00:00:30.559
And so is most of the useful lingo because controlling front-end chatbots and models and modes are now table stakes.
00:00:30.719 --> 00:00:35.520
The differentiator now is controlling long-running desktop agents.
00:00:35.679 --> 00:00:48.799
And it sounds easy in theory until you realize that the language you've been building up on the front-end AI chatbot era and the language of the long-running autonomous agents are not exactly compatible.
00:00:49.039 --> 00:01:02.000
So today we're giving you the primer and establishing some baseline vocabulary and concepts you'll need heading into 2027 when long-running AI agents become the new norm.
00:01:02.240 --> 00:01:04.959
So here is the big picture.
00:01:06.719 --> 00:01:11.439
Agent vocabulary is now an essential skill set.
00:01:11.599 --> 00:01:16.799
And yeah, it's changing all the time, which is one of the reasons why we do this thing every day.
00:01:16.959 --> 00:01:32.959
But agents can obviously read and write files, they can run tools, call their own, create their own apps and plugins, they can fix mistakes and work for hours unattended, which is both a good and a bad thing, depending on uh how active you are in your agents.
00:01:33.120 --> 00:01:40.079
And every new term now names a problem you only hit once an agent run runs long.
00:01:40.239 --> 00:01:40.799
Uh right.
00:01:40.879 --> 00:01:46.799
So so so much of the previous uh terminology that we use with AI chatbots, right?
00:01:46.879 --> 00:01:52.400
You had an instant feedback loop for yourself, whether it worked or not, and that is kind of gone.
00:01:52.480 --> 00:01:53.840
You really have to be paying attention.
00:01:54.000 --> 00:02:06.959
And not knowing the words now, well, means you might set a vague goal, you might put up weak guardrails, and that run that you think might fix a problem as you go take a walk, might not really go anywhere.
00:02:07.120 --> 00:02:15.280
And learning this new agentic language is becoming as essential as learning how to prompt, as that was helpful.
00:02:15.439 --> 00:02:24.719
Uh, so on today's show, we're going to learn what loops, goals, plans, and subagents actually mean without the jargon.
00:02:24.879 --> 00:02:33.520
Uh, we're gonna go over how all of these terms work in Codex and Claw Desktop so you can know how these features work together.
00:02:33.599 --> 00:02:47.759
Uh, you're gonna learn why fluency in these terms is the skill separating operators from spectators, and you're gonna know the mental model that makes the whole vocabulary click starting now.
00:02:48.080 --> 00:02:50.560
All right, if you're new here, welcome.
00:02:50.719 --> 00:02:53.120
This is the Start Here series.
00:02:54.319 --> 00:03:12.159
The Start Here series is part of everyday AI's ongoing effort to give whether you're a new listener or a seasoned AI expert to give everyone an essential podcast series to both learn the AI basics and double down on your AI knowledge.
00:03:12.319 --> 00:03:14.639
So if that's what you're trying to do, sweet, me too.
00:03:14.719 --> 00:03:15.439
Let's do it together.
00:03:15.599 --> 00:03:17.599
Make sure to go to starthirseries.com.
00:03:17.759 --> 00:03:31.840
That's going to give you exclusive access to our inner circle community and you're going to be redirected straight into our Start Here Series space, which has a playlist uh for all of these episodes, all in order.
00:03:31.919 --> 00:03:40.000
So it's easy to uh go through them all, as well as um you can read about them all on the page and connect with other people who are along the journey with you.
00:03:40.159 --> 00:03:53.360
So if you missed our last Start Here series, that was volume 29, where we talked about the open source surge and if models like GLM 5.2 uh make open source an enterprise priority.
00:03:53.520 --> 00:03:58.560
But today we are talking about the desktop agent lingo simplified.
00:03:58.800 --> 00:04:03.280
So first let's zoom out completely.
00:04:03.439 --> 00:04:03.919
All right.
00:04:04.159 --> 00:04:16.079
So if you are not someone that's using codex or clawed desktop or anti-gravity or cursor, uh, some of this might not make a ton of sense.
00:04:16.160 --> 00:04:16.959
And that's okay.
00:04:17.199 --> 00:04:20.959
And maybe this show is more for you than anyone else.
00:04:21.040 --> 00:04:31.920
Uh, because if you are using something like Codex or uh Claw Desktop every single day for hours, this episode will probably be a review at best, but I think still helpful.
00:04:32.000 --> 00:04:39.519
Uh so if you're like, okay, I don't use these tools, no, you need to listen up uh and you should start using them.
00:04:39.680 --> 00:04:51.040
Uh, but I want to talk about the shift, obviously, from the uh AI chatbot that's just very reactive versus the proactive autonomous desktop worker.
00:04:51.120 --> 00:04:54.399
Um, and really what separates them is the harness, right?
00:04:54.560 --> 00:05:05.199
So uh sometimes when you talk about a model, you talk a lot about the harness or where it lives and how it accesses how it how they access all of these tools, uh right.
00:05:05.680 --> 00:05:09.680
So let's just use an example codex from OpenAI, uh, right?
00:05:09.839 --> 00:05:11.120
It's my favorite harness.
00:05:11.199 --> 00:05:13.680
It's the one I use most, but a lot of people don't even know.
00:05:13.839 --> 00:05:16.959
You can use other models inside codecs.
00:05:17.120 --> 00:05:19.199
Codex is the harness, uh, right?
00:05:19.279 --> 00:05:24.160
You can't do that with Claude Code, you can do that with anti-gravity, codecs and some others, cursor as well.
00:05:24.319 --> 00:05:34.160
Uh, so the harness is how all of these uh agenc tools come together and they can work over long term, uh right.
00:05:34.240 --> 00:05:39.439
When you see all these stories, you're like, oh, this person had an agent running over the weekend or overnight.
00:05:39.600 --> 00:05:47.680
Well, that's really made possible by the harnesses that these companies create that give the models and the tools essentially a sandbox to do all this.
00:05:47.839 --> 00:05:55.759
And you know, it's kind of uh today's episode is really a primer on understanding what happens under this, the hood of the harness, so to speak.
00:05:55.920 --> 00:05:57.759
So we're not gonna get super technical.
00:05:57.920 --> 00:06:04.319
This is uh for beginners and maybe intermediate people as well, but that's what we're truly trying to understand.
00:06:04.560 --> 00:06:13.680
And like I said, even if you're not using these desktop tools now, I think most people will be using them uh come 2027, right?
00:06:13.920 --> 00:06:18.399
We've heard from Microsoft that they're coming out with their super app, uh, right?
00:06:18.480 --> 00:06:23.199
We tackled uh super apps uh on the show and on the start here series.
00:06:23.360 --> 00:06:24.560
Uh, what episode was that?
00:06:24.720 --> 00:06:25.600
Bringing that up here.
00:06:25.759 --> 00:06:32.079
That was our episode uh 799 or volume 28 of the Start Here series.
00:06:32.240 --> 00:06:34.560
So, you know, Microsoft is bringing out a super app.
00:06:34.639 --> 00:06:38.319
So these super apps uh or these agentic harnesses, right?
00:06:38.480 --> 00:06:48.319
If you want to get a more technical term, essentially they are a much more powerful version of a web-based chatbot that can run autonomously on your computer.
00:06:48.399 --> 00:06:50.800
Uh, you can run them in loops, you can run goals.
00:06:50.879 --> 00:06:53.600
We're gonna go over all these things, you can schedule automations.
00:06:53.839 --> 00:07:00.240
So the big difference is well, it can act autonomously, it can share the context uh across different chats.
00:07:00.560 --> 00:07:04.560
Um, and it can read and write uh to your computer.
00:07:04.639 --> 00:07:11.279
So any file, just like a human really would, it can use your actual computer with computer use, it can use your browsers, right?
00:07:11.360 --> 00:07:13.040
Uh, has built-in browsers.
00:07:13.199 --> 00:07:21.759
So uh, you know, definitely go listen to the uh AI super apps episode uh if you want a little bit more primer, but let's get straight into it now.
00:07:22.079 --> 00:07:24.160
So let's talk about plans.
00:07:24.480 --> 00:07:34.240
And I think plans are actually one of the more underrated features of these harnesses, specifically in Claude Code and in Codecs, right?
00:07:34.319 --> 00:07:46.079
Uh every once in a while I'll ask people how they're using uh these different harnesses, and I people don't really talk about or use plans as much as they should.
00:07:46.240 --> 00:07:52.160
So a plan essentially just reveals the route before an agent asks uh or acts.
00:07:52.319 --> 00:08:00.879
So a plan shows the intended steps before, you know, the files, the tools, the app changes all of those things, and planning exposes the assumptions, uh, right.
00:08:01.040 --> 00:08:04.240
So likely files, approval points, and verification steps.
00:08:04.319 --> 00:08:11.199
So plan modes matter because desktop agents can change work fast and they can work for a very long time.
00:08:11.360 --> 00:08:11.839
Uh, right.
00:08:11.920 --> 00:08:15.600
So it's kind of like uh a blueprint for a building, right?
00:08:15.759 --> 00:08:22.160
You wouldn't just, if you had unlimited resources and you wanted to build a building, you wouldn't just go to someone and be like, hey, go build a building.
00:08:22.240 --> 00:08:25.920
Or if you're building a custom house, you wouldn't just be like, hey, I want a house and make it awesome.
00:08:26.079 --> 00:08:26.480
Uh right.
00:08:26.560 --> 00:08:34.720
You'd probably sit down with an architect and you would go over the blueprints, the floor plans, uh, zoning restrictions, requirements, all those things, right?
00:08:34.879 --> 00:08:44.240
It's might sound tedious, but you're probably gonna get a much better result in the long run if you sit down and have the conversation with said architect or uh general contractor, right?
00:08:44.320 --> 00:08:44.879
Whatever it is.
00:08:45.039 --> 00:08:46.639
That's the same thing a plan is.
00:08:46.720 --> 00:08:48.159
It is literally a plan.
00:08:48.240 --> 00:08:58.080
Uh so a lot of people will usually just point uh, you know, clawed desktop or codex uh to a folder or share a little bit of context and say, get to work, buddy.
00:08:58.159 --> 00:08:59.279
Uh, not a good idea.
00:08:59.440 --> 00:09:01.919
So the plans are essential.
00:09:02.159 --> 00:09:08.799
So codex, right, uses uh the kind of plan, pair, and execute as collaboration gears.
00:09:08.879 --> 00:09:16.080
Uh so it reads, it analyzes, and then it proposes things to you, and then it waits for you to approve it before implementation.
00:09:16.240 --> 00:09:20.320
Uh so similarly, uh, that's how Claude Desktop works.
00:09:20.480 --> 00:09:26.320
Uh, it actually works out nicely that these work very similarly in Claude Desktop and in Codex.
00:09:26.480 --> 00:09:32.000
Uh, so if you're actually using these and if you want to see like how does this work, where's the button, right?
00:09:32.159 --> 00:09:34.000
Codex has a plan button.
00:09:34.080 --> 00:09:41.519
I don't believe Claude Desktop has it, but you can invoke them each the same way, which is just a backslash and then a plan, uh, right.
00:09:41.600 --> 00:09:44.639
And then they'll kind of walk you through and then you will approve the plan.
00:09:44.799 --> 00:09:57.919
Uh, so in the same way, how I am a very uh adamant telling people like, make sure you read the chain of thought after you've gone through uh, you know, and and gotten an output out of a large language model.
00:09:58.080 --> 00:10:01.600
I am that much um in favor of using plan mode.
00:10:01.679 --> 00:10:03.279
It is the equivalent on the front end.
00:10:03.360 --> 00:10:03.840
Uh, right.
00:10:03.919 --> 00:10:12.000
So if you've never used these harnesses, think of something like Google's uh Google Gemini's deep research actually does a really good job of this, right?
00:10:12.080 --> 00:10:20.559
Uh before you go out and do a deep research with Google Gemini, it actually gives you a plan to approve so you can see it and if something's wrong, you can modify it.
00:10:20.720 --> 00:10:32.799
And this is really important because when we talk about the the shift from token maxing to token efficiency, right, depending on how you set these agents up and what kind of task you're giving them, they might run for a long time.
00:10:32.960 --> 00:10:38.960
And you know, if you are using these on a company plan, chances are you're paying via the API.
00:10:39.120 --> 00:10:48.080
Uh so not having a plan in the same way telling uh a builder to build you an amazing house can be a very expensive step to overlook.
00:10:48.320 --> 00:10:50.320
Plan plan mode, you have to think.
00:10:50.399 --> 00:10:51.519
It's not slowing you down.
00:10:51.600 --> 00:10:54.799
A lot of people are just like, I want to build, I want to break stuff, I want to burn tokens.
00:10:54.960 --> 00:10:56.879
No, it's a guardrail, right?
00:10:57.039 --> 00:10:58.080
Not a slowdown.
00:10:58.159 --> 00:11:06.559
So planning separates thinking from doing, uh, but it's not, you know, it's not enough because risky work still needs, right?
00:11:06.639 --> 00:11:09.919
That read-only access, the work trees, the checkpoints.
00:11:10.080 --> 00:11:16.639
And then once your route is approved, you know, then the permissions kind of define the real blast radius.
00:11:16.799 --> 00:11:29.360
Uh, so don't just rush into giving a long-running agent uh keys to the castle or you know, giving it access to just run for hours and burn through your API bill.
00:11:29.679 --> 00:11:32.960
All right, next, let's talk about goals.
00:11:33.120 --> 00:11:33.600
All right.
00:11:33.759 --> 00:11:38.240
And I I kind of put these in an order that I think might make sense for most.
00:11:38.399 --> 00:11:40.960
Um, I generally will start with a plan.
00:11:41.039 --> 00:11:49.440
And the way I actually do it is I use a plan and then uh parlay uh the findings of that into a goal.
00:11:49.600 --> 00:11:54.159
So it's a little confusing and uh Claude Code and Codex work a little bit differently.
00:11:54.240 --> 00:12:03.679
Uh uh in for plan mode, it will still work through your plan, but a lot of times it will stop once it's gone through the steps.
00:12:03.840 --> 00:12:16.480
Uh so the biggest difference between a plan and a goal is on the front end, uh, a plan literally just outlines it and you can see uh codecs or Claude code work through each step, and there is a visual indicator, which is great.
00:12:16.639 --> 00:12:17.840
Goal is a little bit different.
00:12:18.080 --> 00:12:24.799
Goal is you literally give it an end goal and it will not stop and in sometimes loop until it hits that goal, right?
00:12:24.879 --> 00:12:44.320
So uh a lot of times you have to be careful using goal if you don't go through a plan mode first, or if you don't go through an old school best practice prompt engineering, context engineering, like we used to teach with Prime Prompt Polish, to really share that context and have an understanding with the model of what ultimately the output looks like.
00:12:44.480 --> 00:12:46.960
So if you just blindly, and a lot of people do this, right?
00:12:47.039 --> 00:12:53.919
They just blindly throw a bunch of context at Codecs or Claude Desktop and they give it a goal and it will keep going until it hits that goal.
00:12:54.080 --> 00:13:07.120
And if that goal is maybe unattainable, uh, yeah, that's where you can have it work for uh hours on end, or maybe a day or longer, and all of a sudden, yeah, your uh your your bills through the roof.
00:13:07.440 --> 00:13:10.480
So uh let's talk a little bit about goals.
00:13:10.559 --> 00:13:14.080
So it's really defining the finish line before you start any motion.
00:13:14.240 --> 00:13:17.679
And like I said, I always go through a planning phase first.
00:13:17.840 --> 00:13:26.000
I usually make the plan kind of big uh to work through it, uh, let you know, Codex or uh Claude go through and do some work.
00:13:26.080 --> 00:13:31.039
Then I'll see, you know, there's always gonna be shortcomings the first time you give something uh something to someone.
00:13:31.279 --> 00:13:33.039
So I'll see where went wrong.
00:13:33.120 --> 00:13:38.639
I'll go through, read the summarized chain of thought, then I will kind of reprompt it and then make that as a goal.
00:13:38.799 --> 00:13:39.039
All right.
00:13:39.200 --> 00:13:48.399
And sometimes I'll literally just copy and paste the original plan, make some modifications, obviously, because some things on the list or on the steps of the goals are gonna get crossed off.
00:13:48.559 --> 00:13:49.519
Some things aren't.
00:13:49.600 --> 00:13:56.559
Uh so I usually do a lot of manual nitpicking, uh, transitioning from a uh a plan to the goal.
00:13:56.799 --> 00:13:59.919
But the goal is the outcome and agent checks across the long run.
00:14:00.080 --> 00:14:11.279
And strong goals specify the audience deliverable source material in the done condition, which is why I think it is usually best in general terms for you know, general knowledge work.
00:14:11.360 --> 00:14:13.919
And again, I'm coming at this from like a general knowledge work.
00:14:14.080 --> 00:14:26.000
Yes, I do some coding software dev stuff, building myself, uh, you know, cool projects and uh uh the uh products, extensions, all those things, but I am talking about this through general knowledge work.
00:14:26.080 --> 00:14:33.919
So even if you're not, you know, writing software or you know, vibe coding something, let's just say your vibe working, I still think that this holds true.
00:14:34.159 --> 00:14:46.000
Um, so weak goals make plans, loops, and subagents drift quietly, which I think it's important to go through, especially if you're talking about uh a project that you want to do, right?
00:14:46.080 --> 00:14:54.639
If you're just trying to knock off a uh a task that might normally take you or a single agent on the web, you know, five, 10 minutes to go through.
00:14:54.799 --> 00:14:58.000
I don't think you necessarily have to go through the plan and the goal in combination.
00:14:58.159 --> 00:15:16.240
But if you're talking about knocking out an actual project, which is what these systems are capable of doing right now, that might take a human two, three, four, five hours or two, three, four, five days, that's when I think it uh, you know, when you're talking about your ROI and what you have to invest on the front end.
00:15:16.399 --> 00:15:21.360
This is one of those things where you're gonna have to build a bridge in order to save the four days driving around the mountain.
00:15:21.519 --> 00:15:26.879
Uh, so goals differ a little bit uh across Codecs and Claude.
00:15:27.120 --> 00:15:33.039
So Codec's goals are a little bit more persistent and editable and also visual in the composer.
00:15:33.120 --> 00:15:41.039
There's a nice little toggle uh button in the goal first threads uh in codecs, kind of keep the objective across turns and sessions.
00:15:41.200 --> 00:15:46.320
Uh Clawed code uh via the desktop version, not the uh CLI.
00:15:46.480 --> 00:15:50.000
It does still support the backslash goal in the same way.
00:15:50.080 --> 00:15:53.120
Uh, but uh cowork is a little bit differently.
00:15:53.200 --> 00:15:54.399
Uh works a little bit different.
00:15:54.480 --> 00:15:55.039
That's the other thing.
00:15:55.200 --> 00:16:04.240
People uh, you know, and for me, I personally use uh, you know, codecs, and I do have to tell people this because if not, it would be extremely irresponsible of me.
00:16:04.399 --> 00:16:10.080
Uh, you know, the big difference between codex and claw desktop is clawed desktop is fragmented.
00:16:10.240 --> 00:16:13.360
There is a chat, a co-work, and a code tab.
00:16:13.600 --> 00:16:16.320
And those tabs have no clue what the other is doing.
00:16:16.399 --> 00:16:27.200
So there's, you know, uh certain things that work really well in cowork goal, uh, such as the visual indicators that don't always work as well in the code version of goal.
00:16:27.440 --> 00:16:45.279
So, you know, even as we go over individual features, uh, because clawed code by default, uh or sorry, clawed desktop by default is siloed uh between the chat, the co-work uh and the code, you know, even things like goal work a little bit differently uh in co-work and in clawed code.
00:16:45.519 --> 00:16:51.600
All right, another thing to keep in mind um a goal is not just a step-by-step prompt, right?
00:16:51.759 --> 00:16:56.159
So over-specified steps can actually fight the agent's own planning process.
00:16:56.320 --> 00:16:59.759
So you don't have to know everything going into it, right?
00:16:59.840 --> 00:17:04.559
You don't have to like be like, oh my gosh, I can't use this plan thing and the goal thing because I don't have experience.
00:17:04.640 --> 00:17:05.680
I'm not a software engineer.
00:17:05.839 --> 00:17:06.799
Absolutely not, right?
00:17:06.960 --> 00:17:13.200
You have to be able to communicate and understand what you want and then work with an agent to you know go through the steps to get there.
00:17:13.359 --> 00:17:18.960
But under specified outcomes, make the agent invent missing like success criteria.
00:17:19.039 --> 00:17:21.599
And that's what you absolutely want to avoid.
00:17:21.920 --> 00:17:22.559
All right.
00:17:23.119 --> 00:17:26.720
So now let's talk about loops, right?
00:17:26.799 --> 00:17:36.079
So loops are kind of the uh new-ish trend, uh, right, even though we had the Ralph loops, you know, many, many quarters ago.
00:17:36.160 --> 00:17:40.559
Uh, now loops are making their uh viral reappearance again.
00:17:40.720 --> 00:17:46.880
So loops uh essentially can turn plans, approved plans, into agent work.
00:17:47.039 --> 00:17:47.519
Uh right.
00:17:47.599 --> 00:17:51.359
So similarly, like a heartbeat if you're using open claw, uh, right.
00:17:51.440 --> 00:17:57.440
So think of it like this it's it's something that you can schedule to happen over and over and over and over again.
00:17:57.680 --> 00:18:02.880
So a loop means you you know the agent is going to observe, plan, act, check, adjust, and repeat.
00:18:02.960 --> 00:18:04.720
It is literally a loop.
00:18:04.799 --> 00:18:12.480
Uh, so a chatbot would run once, while a long-running agent can loop many times if you give it that instruction.
00:18:12.640 --> 00:18:16.880
So the loop only becomes useful when the agent verifies each step.
00:18:17.039 --> 00:18:25.920
Otherwise, you're just burning tokens like you're still trying to climb the uh internal meta um, you know, token burning leaderboard.
00:18:26.400 --> 00:18:29.599
So uh, you know, loop review looks a little bit different in each product.
00:18:29.680 --> 00:18:34.079
So codex uh runs every task loop inside a project organized thread.
00:18:34.160 --> 00:18:40.160
Uh so you know, you can also uh build loops as skills or automations, right?
00:18:40.240 --> 00:18:43.759
So you can just tell them, hey, here's what I want you to do, right?
00:18:43.920 --> 00:18:57.279
Every, you know, every hour I want you to go through, you know, triage my email, my calendar uh and my drive, uh, you know, respond to any emails that I might want to update any decks, and I want you to do this every hour.
00:18:57.359 --> 00:18:59.279
I want you to do this, you know, every day.
00:18:59.359 --> 00:18:59.839
Uh, right.
00:18:59.920 --> 00:19:04.720
So that's an example of a very oversimplified example of what could be considered a loop.
00:19:04.880 --> 00:19:05.039
Right.
00:19:05.119 --> 00:19:07.039
So Claude Cowork, again, a little bit different.
00:19:07.119 --> 00:19:11.440
It shows the steps with citations to the files and the messages, which I really like.
00:19:11.519 --> 00:19:18.880
I like that uh that piece in co-work uh a little bit more than I like it in Claude Code because it has that dedicated side panel on the right.
00:19:19.039 --> 00:19:23.920
So even if you are running technically a loop, you can see those steps visually get checked off.
00:19:24.000 --> 00:19:29.599
Uh Claude Code surfaces the plans, the tasks, the sub agents, the diffs in the progress panels.
00:19:29.759 --> 00:19:36.960
The biggest thing to talk about when uh you know talking about loops is having very clear verification, uh, right?
00:19:37.039 --> 00:19:40.880
Whether you want to run something uh a first time, right?
00:19:41.200 --> 00:19:43.839
I would literally, and I've done this before.
00:19:44.000 --> 00:19:52.000
You can work through, uh just go through a natural language, work with codecs, work with Claude Code, and say, hey, I want to create a loop.
00:19:52.079 --> 00:19:59.599
I want to ultimately save this as a skill earn automation, uh or you know, usually both save it as a skill and an automation that you can schedule and run.
00:19:59.680 --> 00:20:03.599
But say, hey, instead of giving you the full loop, I want to work with you uh through this.
00:20:03.759 --> 00:20:06.319
Here, you know, here's ultimately what I want to do.
00:20:06.400 --> 00:20:11.839
I want to, you know, check this website uh, you know, twice a day.
00:20:12.000 --> 00:20:16.079
Uh, you know, this is my whole company, we live and die by this website, right?
00:20:16.160 --> 00:20:21.599
It could be industry news, could be uh I don't know, stocks, finances, um, etc.
00:20:21.839 --> 00:20:22.000
Right.
00:20:22.079 --> 00:20:31.920
But think if if if you are someone that has to pay very close attention to the market, something in healthcare, I don't know, FDA regulations, and there's all these things pinging all the time.
00:20:32.079 --> 00:20:32.559
Uh, right.
00:20:32.640 --> 00:20:41.440
Maybe you want to create a loop that you train or you help build a skill set with codecs or clawed code that goes in there every so often.
00:20:41.599 --> 00:20:46.480
You know, you see what's new, you run it through your context, your decision-making process, etc.
00:20:46.799 --> 00:20:49.519
Work through those steps in the loop one by one.
00:20:49.759 --> 00:20:59.200
So the um, so this harness or the model in the harness can understand what verification looks like, what success looks like at each step of the loop.
00:20:59.359 --> 00:21:01.920
If not, bad loops, bad, right?
00:21:02.079 --> 00:21:09.519
Bad loops turn into producing uh, you know, overly polished work that just is maybe wrong.
00:21:09.759 --> 00:21:15.440
Because when one loop gets overloaded, then sub agents are maybe going to split the work.
00:21:15.599 --> 00:21:15.920
All right.
00:21:16.079 --> 00:21:17.680
So loops are great.
00:21:17.839 --> 00:21:20.160
You can save them as a skill in automation.
00:21:20.319 --> 00:21:22.400
Think of it similar to like a heartbeat.
00:21:22.559 --> 00:21:27.119
Uh, if you've used OpenClaw, the great thing, all it takes is natural language, uh, right.
00:21:27.200 --> 00:21:31.119
I do think in this aspect, codex is a little bit better uh at setting these up.
00:21:31.200 --> 00:21:40.240
I think that they're uh because not being fragmented, uh, it's a little bit easier to just chat uh with codecs in natural language to set up those loops.
00:21:40.480 --> 00:21:44.319
All right, and then last but not least, we have sub agents.
00:21:44.720 --> 00:21:56.319
So all right, sub agents like loops are where things get very interesting and potentially dangerous if you are not uh being hands on.
00:21:56.640 --> 00:21:59.519
So if you're being uh uh laissez faire.
00:22:00.160 --> 00:22:02.720
Human in the loop and setting all these things off.
00:22:02.880 --> 00:22:08.400
I wouldn't recommend going too heavy and the pains on uh loops and subagents.
00:22:08.480 --> 00:22:09.920
It will get expensive quickly.
00:22:10.000 --> 00:22:13.839
If you are very hands-on, uh, I think loops and subagents are great.
00:22:13.920 --> 00:22:15.519
So here's what subagents are.
00:22:15.680 --> 00:22:21.119
Well, they're just helper agents with focused assignments and separate context windows.
00:22:21.279 --> 00:22:24.799
So they help with parallel work, not just vague requests to think harder.
00:22:24.960 --> 00:22:29.359
The real value, though, is context hygiene before the consolidation begins.
00:22:29.519 --> 00:22:38.240
So subagents control um kind of uh a different aspect of a job.
00:22:38.480 --> 00:22:40.640
So one easy thing, right?
00:22:40.720 --> 00:22:46.559
Uh an example of how I use subagents uh in codecs uh and in Claude.
00:22:47.279 --> 00:22:52.480
So you can just use them um by invoking them.
00:22:52.720 --> 00:22:54.160
Say use sub agents for this.
00:22:54.240 --> 00:22:56.000
A lot of people are like, How do you use sub agents?
00:22:56.079 --> 00:22:58.559
Well, you natural language, use sub agents for this.
00:22:58.720 --> 00:23:03.599
So, as an example, one thing I like to do is I like to build with codecs.
00:23:03.680 --> 00:23:03.920
All right.
00:23:04.000 --> 00:23:05.279
So I build with codecs.
00:23:05.359 --> 00:23:07.440
A lot of people are like, Oh, Claude is better at front end.
00:23:07.519 --> 00:23:09.839
Yeah, it's way better at front end, but I don't know.
00:23:09.920 --> 00:23:15.759
Uh, codecs has this thing called the most powerful AI um image model in the world.
00:23:15.920 --> 00:23:28.880
So start with the uh AI image gen in codecs, uh, use the front end skill, takes three seconds, and you generally will have a much better front end uh if you are building, you know, a piece of software that's just what I'm using as an example, right?
00:23:29.039 --> 00:23:35.200
If you use the image gen in the front end design skill, you're gonna have way better than what you would get out of Claude Code if you do that.
00:23:35.279 --> 00:23:37.680
If not, obviously uh Cloud Code is gonna be better.
00:23:37.759 --> 00:23:46.079
But one reason or one way I use subagents, uh I I love Claude's ability to uh really investigate and use subagents across multiple things.
00:23:46.160 --> 00:23:50.400
So I'll sometimes I'll assign, you know, I'll say, hey Claude, I want you to do these sub agents.
00:23:50.480 --> 00:23:53.759
I want you to look at, you know, one, look at the entire uh repo.
00:23:53.920 --> 00:23:58.079
Number, you know, two, look at the feature set, three, look at the language, right?
00:23:58.160 --> 00:24:07.119
So then each subagent uh can really have a more defined and refined um kind of uh task, right?
00:24:07.279 --> 00:24:08.559
So it's the same thing.
00:24:08.720 --> 00:24:16.160
If you just walked into a room of 20 employees and they're there to help you, and you say, Hey, 20 employees, uh, go help me with this project.
00:24:16.240 --> 00:24:18.559
Here's where we're at, have fun, uh, right.
00:24:18.720 --> 00:24:21.119
They might decide upon themselves, okay, let's split this up.
00:24:21.200 --> 00:24:30.559
But if you say, hey, designer, go look at the design, hey copywriter, go look at the copy, uh, hey engineer, go look at the, I don't know, the security on the back end, make sure everything's you know tightened up.
00:24:30.640 --> 00:24:33.759
So all you have to do is well say, go use sub agents.
00:24:33.920 --> 00:24:46.000
Obviously, these models are smart enough where you can just say that broadly, um, and they will uh, you know, usually uh, you know, assign one sub agent to a specific task.
00:24:46.160 --> 00:24:50.880
Uh, usually I'll just broadly say go use subagents and poke holes in ABC.
00:24:50.960 --> 00:24:52.880
And then I'll see exactly how they did it.
00:24:52.960 --> 00:25:00.319
Uh Claude and Codecs allow you to click uh onto a subagent and see exactly what they're working on in real time, which is cool.
00:25:00.400 --> 00:25:01.759
I love how Codecs names them.
00:25:01.839 --> 00:25:11.200
Uh the Claude uh names, I forgot how they name them, but um, you know, usually I'll do a quick general subagent run if it is a big project that's super important, right?
00:25:11.279 --> 00:25:13.279
That I can spend kind of the token budget on.
00:25:13.359 --> 00:25:17.920
Uh, I'll see what they did, what went wrong, what didn't, uh, you know, how it can be improved.
00:25:18.000 --> 00:25:22.559
And then I'll probably assign roles the second time I have another group of subagents do it.
00:25:22.720 --> 00:25:30.880
Sometimes I'll have sub agents work on the front end before the work actually starts to scope it and make sure it makes sense on where they're spending their time.
00:25:31.039 --> 00:25:35.039
Sometimes I'll have them work on the back end after the work is done, or a combination of the both.
00:25:35.200 --> 00:25:44.240
And the best thing is well, you can just tell Codex or Claude, like, hey, have sub-agents work on the front end before you start this project to make sure everything is correct.
00:25:44.319 --> 00:25:49.440
Uh, then go do the work, then have a separate group of sub-agents really tear it apart on the back end, right?
00:25:49.519 --> 00:25:51.920
That's all you really need to say, and they will do that.
00:25:52.000 --> 00:25:59.440
It is kind of like this having this parallel work stream that can disagree, duplicate, uh, you know, miss shared constraints.
00:25:59.519 --> 00:26:00.160
And that's great.
00:26:00.240 --> 00:26:04.720
And then the main agent can pair findings, resolve conflicts, synthesize.
00:26:04.880 --> 00:26:10.720
I've even set up a skill that has like a three-tier uh kind of uh agentic management system.
00:26:10.799 --> 00:26:12.880
So I don't even have to go through and type this, right?
00:26:12.960 --> 00:26:23.279
So if I know a project is pretty big, or if I'm gonna be working, you know, it's like, hey, I'm setting this off to work overnight, uh, it's it's an ongoing project, and I'm not gonna go through the the planning and goal.
00:26:23.440 --> 00:26:37.119
I know I have this, you know, subagent uh system where I, you know, say, hey, this one's gonna go in and be a contrary uh or uh contrarian and you know, tear apart all this and second guess every single thing that we do before it goes to production, whatever it may be.
00:26:37.279 --> 00:26:41.519
Uh so that's just uh a way to go through subagents.
00:26:41.599 --> 00:26:42.000
All right.
00:26:42.160 --> 00:26:45.920
So now as we wrap up, the real constraint here is actually your machine.
00:26:46.079 --> 00:26:52.559
This is the big thing to keep in mind because your machine has to be on, it has to be uh right, you have to have these programs open.
00:26:52.720 --> 00:26:56.160
So yeah, you can do all these great things like control your computer, right?
00:26:56.240 --> 00:26:57.839
And and control the browser.
00:26:58.000 --> 00:27:07.759
And think of that as well as we talk about uh, you know, the desktop lingo is it's literally as if you were talking to a human sitting in front of the computer, but it has to be on, right?
00:27:07.839 --> 00:27:10.480
A sleeping laptop can break remote steering.
00:27:10.720 --> 00:27:16.240
Uh, you know, if you run out of power, if your internet goes off, everything, well, from an agentic perspective stops.
00:27:16.400 --> 00:27:17.519
Where that's a little bit different.
00:27:17.680 --> 00:27:24.400
The advantage of using on the front end chatbot, uh, those you know, singular tasks, those singular one-off prompts can continue to go on.
00:27:24.559 --> 00:27:27.359
But the beginner mistake is not managing anything, uh, right?
00:27:27.440 --> 00:27:30.480
The beginner mistake is being like, wow, these things are super powerful, these agents.
00:27:30.640 --> 00:27:35.519
I can just go in and you know, drop it some context and say, go do this work and wow, look at it.
00:27:35.680 --> 00:27:39.519
You're gonna get just slop work almost every time, uh, right.
00:27:39.599 --> 00:27:45.200
Because also bad handoffs say research this, organize that, make that better, and then you just use the output.
00:27:45.279 --> 00:27:45.759
Uh, right.
00:27:45.920 --> 00:27:53.119
Good handoffs are well defining the goal, the plan, the permissions, the workflow, the verification, the subagents, the loops, uh, right.
00:27:53.200 --> 00:27:59.680
That's the differentiator now, not just saying, oh, I use codecs, oh, I use Claude Code Desktop, right?
00:27:59.759 --> 00:28:04.880
Uh, the handoff turns the agent lingo into repeatable desktop delegation.
00:28:05.359 --> 00:28:16.400
So, like I said, the new skill is number one, understanding how to talk to your agents, but then using those skills to supervise work, not just prompt, right?
00:28:16.480 --> 00:28:19.359
The agentic layer is growing thicker and thicker by the day.
00:28:19.519 --> 00:28:25.119
That doesn't mean the human layer, uh, both uh, you know, I call it the uh, you know, the agentic human sandwich, right?
00:28:25.200 --> 00:28:25.839
We are the bun.
00:28:25.920 --> 00:28:30.160
We are providing uh the input and the context on the front end, and then ultimately the verification.
00:28:30.240 --> 00:28:32.480
But that doesn't mean we're hands-off in the agentic layer.
00:28:32.559 --> 00:28:35.039
We have to constantly be monitoring, uh improving.
00:28:35.119 --> 00:28:42.480
So we don't just, you know, blindly let these agents loop and uh you know, blindly assign work to subagents, uh, right?
00:28:42.640 --> 00:28:49.200
We can just spend more time uh as the bun and far less time as the meat, to use my old saying there.
00:28:49.359 --> 00:29:03.200
All right, I hope this one was helpful going over the basics of desktop agent lingo, simplifying goals, loops, plans, uh, and sub agents and a little bit of how they work in to in codex and in clawed code on the desktop.
00:29:03.279 --> 00:29:06.480
So if this was helpful, please go to starthirseries.com.
00:29:06.720 --> 00:29:09.519
If you haven't already, make sure to subscribe to the podcast.
00:29:09.680 --> 00:29:10.880
I would really appreciate that.
00:29:11.039 --> 00:29:11.839
Thanks for tuning in.
00:29:12.000 --> 00:29:14.720
Hope to see you back tomorrow and every day for more everyday AI.
00:29:14.799 --> 00:29:15.680
Thanks, y'all.
