← Behind the Craft
Behind the Craft
How I Plan, Build, and Run Loops with Claude Code in 40 Minutes | Thariq Shihipar
00:41:17
Loading summary
Transcript136 lines
- [00:00]
A
As the models have gotten smarter, they need less direction, fewer constraints, and fewer examples. We've got loop, we've got goal, and we've got workflows. These are all geared at trying to get the agent to run for long periods of time. This was just one shot, like win a one prompt. Sorry. It's created the caption and it's created the little overlay, and now it's going to fade to black. My goal for this year is to be more productive, but work less.
- [00:28]
B
Hey, everyone. Today I'm really excited to welcome Tariq from the cloud code team. This has been a long time coming, so I'm going to ask Tharik to show us how to design loops and workflows to get cloud to work longer and how he personally uses cloud code and also the new cloud tag. Welcome, sir.
- [00:41]
A
Hey, Peter. Yeah, thanks for having me. I've been excited to join for a while.
- [00:45]
B
Yeah. All right, so maybe before you demo anything, maybe you can talk about at a high level how you and a cloud core team think about how do you go from you being a person prompting the agents to kind of design loops and some of the stuff to get the agent to kind of work by itself? How do you guys think about this?
- [00:59]
A
Totally. Yeah. So I think loops are a pretty general term, I think, for just different ways of having the agent get feedback or work for long periods of time in an orchestrated way. And so we've got loop, we've got goal, and we've got workflows. And these are all, like you said, geared at trying to get the agent to run for long periods of time. Slash. Goal is one of those things that it helps the agent remind itself what its exit condition is. Right. And only allows it to exit once you know it. And so goal is really great for when you have a complicated task and you really need to, you know, make sure that this is, like, done at the end. You really want to prevent any stopping. You kind of also want to give the agent feedback, a signal to basically say, like, hey, power through things. Right? And so I think one of the, like, if you were to step into Claude's shoes, sometimes you ask someone asks you to do a task, you run into a, like, maybe, you know, sort of a complication or something that is maybe not quite up to spec or, like, doesn't match what the user said, you might stop early and be like, hey, should I keep doing this or not, Right? And so goal is one way of just kind of the user indicating like, hey, I've. I've done Enough spec and exploration. Right. I understand the problem space, just go execute on it and if you run into something, you know, fill it in. Right. So that's how we think about goal. And then workflows are in my opinion maybe like the most powerful form of this where you can, you know, spin up sub agents to both do the work, paralyze the work and verify the work. And so I think especially for non technical work, this can be like a really great way of taking a non deterministic task and then breaking it down roughly into a deterministic. So yeah, I think that's how we think about some of these.
- [02:49]
B
And you know, I think I'm going to do a quick demo but like before that. Yeah, like just on the Clocko team, how do you guys use some of this stuff? Like do you do like a go make this match design or like go make this number go up or what do you use?
- [03:02]
A
Yeah, yeah. I mean everyone does different things, right? Based on like kind of the like based on. Yeah. What they're trying to do. I think like, you know, Jared's talked a lot about the, the bond rewrite in Rust and how that used workflows and he's going to be talking more about that. I think like anytime you have a deterministic sort of like signal, like for example, like latency, right. Like goal is like a great thing to just sort of like have the agent explore things sort of in like an auto research way. Right. And yeah, I think that like you like let's say you're talking about design, right. I think one way you want to think about this is like how well can the agent understand what your design is to begin with. Right. And so you might want to turn it into like if it has a spec that it can verify. Like if it's a FIGMA file, you used a FIGMA MCP and then you're like goal, make sure that the render design matches the FIGMA mcp. That's a lot easier than like a screenshot. Right. And so if you're a screenshot maybe you want to do a workflow where you like it's more squishy, you know, like there's like a rubric that you're evaluating against and you have a verification agent and things like that. So.
- [04:24]
B
Got it.
- [04:25]
A
Yeah, very dependent on the task.
- [04:26]
B
So basically the planning matters a lot. Right. To give it. Because I tried to give me a goal to like just build me an amazing game and then it kind of went off the rails because like, you know, just like One line.
- [04:36]
A
So yeah, I mean, I think the details of like, you know, there's a lot to figure out in what you want. You know, there's a lot of work that goes into that.
- [04:48]
B
Yeah, got it. Okay, so let's make this really concrete. You've been sharing really awesome videos on Twitter about like using Claude to make the video and I think you have like a video workflow that you want to share with us, right?
- [05:02]
A
Yeah, exactly. So I've had a few people ask about this and I think this is a good sort of demo, both for video editing, but also just how to think about a non technical work. So 10 minutes before I got on this call, I recorded a quick video. And so this is me just making a video. Basically what I'm saying is like, hey, it's me, I'm on the Peter Yang podcast. I'm going to point to where I want an overlay to show up and then I'm basically saying, hey, fade to black. This is a quick video that I put together. Now the prompt I gave it is this. This is a repo for a Peter Yang podcast. There's a sample video here called Peter Yang Recording. Transcribe it using Whisper, then use Remotion to create a UI that shows the transcript with each individual word being highlighted and different overlays. And yeah, goal, don't stop until the video is fully rendered. Right, nice. And so now this was just one shot, like when a one prompt. Right. And what it gave me is this. So it's transcribed. It obviously doesn't know my name is Tarik. Right. And it's like overlaid the text or. Sorry, it's created the caption and, and it's created the little overlay and now it's going to fade to black. Right.
- [06:26]
B
Oh, awesome. Okay, so you basically set out the instructions out loud, right?
- [06:29]
A
I said it out loud. Yeah, exactly. And of course there are many ways that you can do this, but it's like now in this directory, transcribed everything using Whisper. It's created the read motion stuff. Yeah, this is like the very basic start to video editing workflow that I have, which is, yeah, I think a good signal for how to do non technical work in cloud code.
- [06:52]
B
So you kind of did this manually with the prompt stuff, but then you probably saved it into a skill or something, right?
- [06:57]
A
Yeah, so right now I don't have a skill. I think one of the things I try and do first is really figure out what I want before I turn it into a skill. And so what are the edge cases here? One of the things I noticed that I had to prompt it towards is getting better at figuring out where my. Where my hand was, right? And so, like. Or where I was pointing towards. And even now I'm not. Not thrilled with, like, where this showed up, you know? And so, like, yeah, like, you could imagine that one direction I could take this in is I could be like, oh, I want to track my fingers or track my face and then get. Give the agent more metadata, right? So that it can then do interesting overlays.
- [07:42]
B
This episode is brought to you by WhisperFlow. Whisper Flow saves me at least three hours a week and is one of my favorite AI apps by far. It's just so much faster to dictate to AI using your voice than to type. You just talk naturally and it outputs clean, ready to send text. Whisperflow even removes filler words and formats your sentences for you. I use Whisper Flow for everything, including drafting newsletter posts, writing product specs, replying on Slack and more. It works on Mac, Windows, iPhone, and Android across all of your favorite apps. Try it free@whispersflow.com and use my code, Peter Whisper Flow, to get six months free. That's Peter Whisper Flow. Now, back to our episode. I think you had another tab that had, like a plan or something. Is that for the video skill? Yeah.
- [08:25]
A
Yeah. So I think I want to talk. I'd love to talk a little bit about how I got here, you know, I mean, so, like, I think, like, even to do this prompt, I think it took a bunch of planning and learning to get here, Right? And then I think that, like, one of the things that I think when we talk about plans, we often are talking about things where it's like just one shot, right? Like, you plan and then you do something and that's it. But I think planning is more of this iterative process of, like, exploring, investigating, finding out what you don't know what you want, right? And then, like, together, like, in the end, that sort of simplifies itself. But so, like, here's an example. But while we do this, I would love to do something where it's like, okay, I'm going to give it. Okay, I want to say, like, I want to update the UI of these overlays, right? And I want to use Peter Yang's style. Here's his blog. Create a HTML artifact for exploring different designs for the overlays and captions, design variations. Okay, so this is like an example to me. This is also planning, because what I'm doing is I'm trying to figure out what I want here it's more of exploration. You can see. One of the important things I'm doing is I'm giving it a reference. When I give it your website, it can now fetch the HTML and start doing this. This is a plan that will run while I go over the other, you know, kind of things I do when I'm planning. Right. So one of them is like, okay, what, how does transcription work? Right. And this is actually, I think pretty important to know because like it informs a lot of the edge cases. So this, this transcription uses Whisper and so, you know, like cloud code sort of put this explainer together, which is honestly like kind of amazing.
- [10:43]
B
Yeah.
- [10:43]
A
The important thing for me is like the like ways things can go wrong. And this is something I prompted it in. The plan was like, I want to explain whisper to me and understand what the edge cases are. Right. And so one of the things that says like silence can become thanks for watching. Right. Or yeah. A word can be split into two chunks. It doesn't know, it doesn't have speaker recognition. Right. And so there are a bunch of different like edge cases here where I think all of this is quite like good. And this sort of helped me build a confidence in using Whisper. Right?
- [11:17]
B
Yeah.
- [11:18]
A
But having these edge cases and knowing what the limits are upfront really helped me avoid this case where I build this complicated workflow during whisper. And then I realized there are things going wrong and I didn't have these unknown unknowns. This case is me sort of doing planning, but really it's like me discovering my unknowns. And I think that can take a lot of different shapes. It can be learning, it can be technical specs. Right. It can be mock ups and exploration. Yeah, I'm starting to like, I feel like the word plant is maybe too broad right now.
- [11:57]
B
So it's more like exploration and understanding like what you're trying to do. Basically.
- [12:02]
A
Yeah, I like to say like getting rid of your unknowns, you know, Like, I think whenever you have a task, it, it, it kind of like almost always there's a lot that you don't know either you don't know how things work or what you want. And it's very, very iterative, you know, and so it's not just like a, you write it all down once and then you implement it. I think there's like many steps and different passes, so.
- [12:27]
B
And yeah, and dude, this is like a beautiful plan, man. Like, do, do you have like, this is like the cloud template or something? Like it has images just using the
- [12:34]
A
front end Design plugin, I think, you know, like, I think like, yeah, it's not. Yeah, there are uglier plans here. Like, I've gotten this like remotion plan. I think that like, you know, the, the design of it doesn't matter so much as sort of like you really just want to make sure it's something that you really do read, you know, and kind of get a sense of like not all of it, but like there's important parts in it. And I think one failure mode I see is that people still sort of like glaze over, you know, the plans and explainers. So.
- [13:05]
B
Yeah, yeah, it becomes really. Because like, I can write all these, you know, crazy markdown files and they're usually pretty long. And then at some point, like, I just get lazy. I'm like, okay, you know, just, just, just do it. Just do it.
- [13:15]
A
Yeah, yeah, yeah, exactly. I think that's like the thing that, you know, it happens to all of us, right? Like the, like the prompt box can definitely just be a lazy button, right? Where you're just like, hey, just do the thing. But usually, you know, like, you end up paying for that, right? Because if you're, if you're trying to do something serious and you're taking the lazy step at each way, it'll end up taking longer, maybe costing longer too.
- [13:42]
B
So it's kind of this iterate. So you ask about the whisper and then you kind of learn a little bit more and then you ask it to make another plan and then at some point you have a plan that you can share, Right?
- [13:52]
A
Yeah, exactly. Yeah. So I did a lot of, for this video recognition stuff, I did a lot of research, research into how different video, like algorithms worked, right. Like, I think I did one on like video segmentation. I wanted to put like text behind a subject. And so this was like something I like explored and learned more about. And I ended up finding that like there wasn't something reliable enough here for me to use. But yeah, I think there's like just a lot of this, like, iterative process of like finding out what do I want, you know, and what is possible right before I can do it. And so it looks like this overlay is maybe almost done, so we'll see it soon.
- [14:32]
B
Yeah. And how about like on the team? So have you got the rest of the team to review HTML instead of markdown?
- [14:38]
A
Yeah, yeah, for sure. I mean, I think like everyone is different, right? So I think that like. But I think HTML artifacts, I mean, we launched artifacts, right? Currently it's only on Teams and enterprise, but hopefully coming to max and pro. And yeah, that's like how we share things now at the company. It's like we ask Claude to make an artifact and this can be of a plan of a PR that we've already done. It can be like a status report, like an incident report, things like that. So, yeah, it's definitely how we do it. Yeah.
- [15:13]
B
All right, let's see what it generated here.
- [15:15]
A
Yeah, let's do it. So let's say open overlay stuff. Okay. So, yeah, here are some of the, like, you know, the options it's made. I think it's probably maybe like, indexed more on the stack over or the substack brand versus your brand. You know what I mean?
- [15:33]
B
Yeah, Because I don't maybe I don't have a brand.
- [15:34]
A
Yeah, yeah, no, I think you do. Right? I think you do. Like, I think you've got the red and white. We can probably iterate on it there. But this is one of the things where I'm like, you know, there's quite a lot of difference between these different, like, you know, designs. Right. And yeah, one of the ways I like to. To plan is. Is to do this exploration, especially because I'm not a designer. So, like, I just really just only know it when I see it, you know?
- [16:01]
B
Got it, got it. Like, just to wrap up the plan conversation, like.
- [16:05]
A
Yeah.
- [16:05]
B
You know, I've been writing, like, product SPECs for, like, 10 years, and like, you know, they usually have, like, what's the problem we're solving, what's the solution, what's the goal? So on and so forth. Right. But I feel like part of this stuff is, like, read by agents now. And I feel like the different sections of your spec needs to change or maybe like the product spec and technical spec is like, one thing. Like, how do you guys do it on the team? They have like, one session for the humans to read and one session for the agents to read. Or how do you think about it?
- [16:29]
A
Yeah, good question. Good question. I think that it is very tied together. Right. Like, I think a spec kind of can evolve even towards, like, planning. Like, I think one of the things I think I'm showing here. Let me show you. See Implementation. Yeah. I think one of the things that you see sometimes is that when you run something, the model can find things that it didn't anticipate or you didn't anticipate when you're implementing.
- [17:01]
B
I see.
- [17:03]
A
I don't think of specking just happening at the start. It's like you. You start with the human request and then the agent does some technical exploration, it comes back, maybe you do some mock ups, some explainers trying to understand your unknowns, you refine that, you give it to the agent again, it might start implementing it. I ask it to keep implementation notes as it goes so that it finds out what are things that we were not expecting about this implementation. And, and once we have that we can actually respec if we need to. Right. Depending on how things go. So yeah, I think it's much less like there's one handoff of spec to implementation and more this back and forth process.
- [17:48]
B
Yeah. Because I guess it's pretty cheap to build now, so you can just maybe ask it to build a simplest version of this and there's probably a lot of bugs and issues and you're going to keep iterating, right?
- [17:57]
A
Yeah, exactly. Or like the prototype version of this. Right. That's how I think about these overlays. These are like prototypes of the, of the design and if we like it then we can do the more expensive version which is instead of HTML we can do it in React. That means you have to re render the video and do all the code changes. So what's the smallest step you can take to prove out the concept that you want to prove out the spec more?
- [18:27]
B
And I love how it takes implementation also. So the next time you start a new session it can just refer to the HTML. Right. It's like a living document.
- [18:35]
A
Yeah, exactly. Yeah.
- [18:37]
B
Got it. What percent of your work now is done through CloudTag on Slack versus in like the terminal or like you know, the cloud app?
- [18:45]
A
Yeah, I think that like the way for me it's trending is that like there's a lot of parallel stuff happening in cloud tag and so any multi clotting is often happening in cloud tag unless there's a reason why it needs to happen locally. You know, I think one of the things we put a lot of time into is like making sure that our environment can run in in remote and so yeah, I, I multi clotting is happening in cloud tag. You know, initial explorations, prax like trying to understand something is happening in cloud tag and then once I get into there's usually one thing I'm focusing on and that's happening in cloud code and it's like more, you know, like back and forth iterative. Right. And so yeah, I think that's how I'm using it but I think everyone at the company has sort of like different like approaches and mixes to it
- [19:35]
B
and when you say like, what do you mean by multi clouding?
- [19:38]
A
Oh yeah, like just like if I have like multiple tasks happening at once, especially if I'm like trying to involve someone else. Like if I am, you know, I've got a PR that I'm trying to get merged. You know, like one thing I'll do is like I'll, you know, ask it to babysit the PR fix test and then tag a reviewer and my like reviewer will get tagged in the same Slack channel and will like be able to talk there. But just generally multi clouding is like anything, any background work. Where before I was having, you know, like five different clouds in my cloud code. Now, like it's mostly one active cloud code session and then a bunch of cloud tag sessions.
- [20:16]
B
Oh, interesting. Okay, and so you talk to cloud tech both through like DMS on Slack and also like in share team channels, I'm guessing or.
- [20:26]
A
Yeah, yeah, in. In flack. Yeah, yeah, both in my private, like I have a. Yeah, my thoric Claude Slack channel where I do like most of my work and then yeah, we have like team channels like feedback and things like that where we or like project specific engineering channels as well. That's like a common pattern we see.
- [20:44]
B
Yeah, I guess it makes sense because I do think like these coding apps right now are like primary single player experiences. Like you're kind of talking to the agents through different threads and if you think about it like the cloud app actually kind of looks similar to Slack. There's like a bunch of threads like just like Slack has a bunch of channels. So I guess like the idea is like Slack is now the multiplayer cloud experience. Right. Because like everyone's there already.
- [21:05]
A
Yeah, I mean, I think like this is, you know, where it's starting. Right. Like, I think ultimately we think Claude will be, you know, this like sort of proactive agent that's sort of like meeting you where you are, you know. And I think Slack is where anthropic is. Right. And yeah, it's a very natural way to explore. And yeah, it's surprisingly good at coding. You can like, you know, there are some people who just do all of their coding almost completely in cloud tag.
- [21:27]
B
Yeah, but okay, maybe this is a dumb question, but how do you like, you know, in cloud code, like easily triggers skills and stuff. But like if I'm in cloud tag, do I just tag Claude and slash the skill or how's it.
- [21:39]
A
Yeah, you can just tell it to use the skill.
- [21:41]
B
Okay, that makes sense.
- [21:42]
A
Yeah, I think there's still UX iteration we're working on around all of this stuff.
- [21:48]
B
Yeah. Okay. I guess. Do you think. Do you agree? I feel like the future for this, like, because you're giving a talk, right. On human agent interaction. And I feel like the future is just like, the agent is just like another employee. You got an onboarding employee, you can talk to it through Slack. You can give it a phone. Phone call. What do you think?
- [22:08]
A
Yeah, I think it's a good question. I think these metaphors can sometimes be helpful, but also limiting in some ways. I think that one of the things when we talk about identity is that in CloudTag, every agent, every channel, has its own memory. Right. And so that is, like, I think one choice. Right. Like, you could also imagine that there are, like, you know, multiple clients you tag. Like, each one has a different Slack identity, and you're tagging them and things like that. And I'm not, like. I think it is kind of helpful to think of an agent that is, like, sort of has persistence and, you know, memory. But I think they're also different from co workers in some ways.
- [22:56]
B
Right.
- [22:56]
A
But I think they're, like, they're proactive and, like, you know. Yeah, it's proactive. It has memory, it has identity. But, yeah, it's like just sort of like an evolution of cloud code, and I think we want to see where, like, you know, where the models will. Will take us there versus sort of like putting it into a box, you know?
- [23:16]
B
Yeah, that's a good point. Yeah. Like, it definitely has a lot better memory than any human.
- [23:22]
A
Better and worse sometimes, you know what I mean? Like, it's spiky, right? So, yeah.
- [23:26]
B
Yeah, makes sense. Okay. All right, dude, well, why don't we do something fun? I want to show you my cloud code setup. I think we talked about this in person. I feel pretty proud of this. I built this podcast production skill, and what it does is it takes a transcript of an interview that I did. I actually interview your colleague Jesse, and it kind of generates a bunch of stuff for me. So it generates thumbnails and, like, stuff to cut. So basically I just tag it and I paste the transcript, and then it starts generating, like, you know, clickbaity YouTube thumbnails and stuff.
- [23:57]
A
Yeah, yeah.
- [23:58]
B
So generic stuff. And, like, I basically give it, like, my examples to try to keep it on track.
- [24:03]
A
Yeah.
- [24:04]
B
But I guess that this skill is kind of like, trying to do a lot of things, so maybe, like, I'm curious if you have any feedback or.
- [24:09]
A
Yeah, this is cool. Can I see what it produces? Like, the clips or like, oh, what it produces.
- [24:14]
B
Yeah, so some of it. What it produces is just like pure text. It's like, here's a news article. Some takeaways.
- [24:19]
A
Copy into YouTube. Yeah, yeah.
- [24:21]
B
And then I have this other scale called the video Post skills. It's kind of similar to yours, but I'm sure it's not as sophisticated. But like, it basically takes the YouTube video for this thing.
- [24:30]
A
Cool.
- [24:30]
B
And then it extracts it and then it gives me some ideas for clips to.
- [24:36]
A
Yeah, yeah. Does it do the clips as well?
- [24:38]
B
Yes, it actually does. So I. I said do two, and then it uses. I don't actually know what it uses. Uses. It uses a bunch of random stuff too.
- [24:45]
A
Too FFMPEG and stuff, right?
- [24:47]
B
Yeah, probably fmpeg. And then it makes a. It makes a video. Yeah. And it has captchas. Yeah.
- [24:52]
A
Yeah. That's awesome.
- [24:53]
B
So. So, yeah, it's not perfect though, but,
- [24:55]
A
but, but, yeah, yeah. I mean, this is great. How, like, how well do you feel like it matches what you want, you know, like. Or do, you know, like, what it would look like to be better?
- [25:11]
B
I wish it could, like, add some, like, you know, B rolls and the stuff that you showed me, like, overlays and stuff like that. And like, I wish it was smart enough to know how to pull in, like, for the video clip. Like, you know, if I want the cloud logo or something, you can just find it online.
- [25:27]
A
Yeah, yeah.
- [25:27]
B
So should I just tell cloud code to do all this stuff or.
- [25:30]
A
I think you can. Yeah. I mean, I think that, like, are you in one persistent repo or like, I think one of the things I think about is, like, sometimes you want a skill, sometimes you want a repo with a lot of scripts and things that you are, like, you know, like, it's more of like a workspace that you're working in. Right. And like, I think of a skill.
- [25:49]
B
Yeah.
- [25:49]
A
Like, I think sometimes the skill can be instructions on how to create that workspace. Right. And so that's maybe something to think about because, like, the more, you know, scripts and sort of like things that you build up, the more the agent can. The less it has to do from scratch, you know? But yeah, I mean, I folder each time.
- [26:09]
B
No, I have all my skills at the user level. Okay. Yeah. And there's like this personal or whatever personal OS folder that has like all the output and stuff.
- [26:19]
A
I see, I see. Okay, cool. Yeah. But, but, but you're saying, like. Yeah, I mean, I think you can sort of build like a video editing harness here. I. I'm Curious about the thumbnails. Do you ever use, like, an image gen API like Gemini or even an OpenAI or something like that?
- [26:34]
B
Yeah, I found it really bad at changing my face. So, like, for example, if I have a smiling thing and then ask me, make it like a shocked face. It makes me look super ugly. Dude,
- [26:46]
A
that's funny. That's funny.
- [26:47]
B
But. But I think it's pretty good at, like.
- [26:50]
A
Like, I don't know.
- [26:51]
B
If I give it my face and I tell you to change the background of the text, it's pretty. Pretty good. Yes.
- [26:55]
A
Yeah, I think Claude is quite good at. One of the things that Claude is good at is using other tools. So you could give it the Gemini or OpenAI image gen APIs, and then you can ask it to look at your faces that it generates as well and sort of tweak it there. So it can do this interactively and progressively as well, which can be helpful.
- [27:18]
B
Okay, interesting.
- [27:19]
A
Yeah.
- [27:20]
B
So basically what I do is I basically just chain a bunch of skills together, like, you know, prepare this thing and then use a thumbnail and so on and so forth. I feel like I know how to use skills now, but, like, some of the other stuff you talked about online, like dynamic workflows and stuff, like, I have no clue. Right? I have no clue.
- [27:35]
A
Yeah. How would we use a workflow for something like this? Right. I think, like, one thing you can imagine is, like, when you're using. I think your thumbnails example was a good one where. Or, sorry, your shorts example was a good one where maybe you want to generate 10 different shorts or 5 different charts or something. Workflows are a great example for that. Right? So you can like, say, hey, like, the main agent decides which five areas of the clip you know you want to, like, create a short for. Right? And then the, like, workflow spins off a sub agent for each one. You also give it maybe a rubric for, like, what does a good clip look like? Right? Like, and then each one is verifying against it and then you're getting back at the end this, like, each clip has a maximum amount of compute put into it to make sure that it's matching what you want versus like, if you're doing two or three simultaneously. Sometimes I think Claude might verify or put less work into any individual clip.
- [28:41]
B
I see. And to create a workflow, just how to make a workflow because basically just
- [28:47]
A
be like, hey, create 10 clips here. Use a workflow. This is my rubric to verify what a good clip looks like. And this can also be A skill. So you can package a skill with a workflow and then the workflow is just a JS file and you can ask it to. You save the JS file into the skill and now you have this reusable skill that you can use.
- [29:15]
B
And this is like a dumb question, but the main advantage over doing this workflow thing versus just using a skill is it can spin up sub agents and keep the context clean.
- [29:25]
A
Yeah, I think it's like, yeah, one of it is context. I think the other one is sort of like laziness and verification. Right. So if you have a. Know like a short is one example of something where you don't have a deterministic like, hey, is this a good short or not? Right. And so if you have a rubric and you have a like, you know, verification agent that reads the rubric and then it looks like, oh, like, hey, let me review the short to make sure it's good and give feedback. Right. That's something that the workflow enables as well. And we found that like, you know, we call it self preferential bias. Like when a model prefers its own outputs, it's going to be like more lenient at like, it's going to be more lenient at like, you know, verifying it.
- [30:18]
B
Okay, so basically you want to have a separate cloud doing the work versus verifying the work, right?
- [30:23]
A
Yeah. And in this case like a separate cloud coordinating the work. So your main agent is coordinating it. Coordinating it. Each one doing the work and then verifying. Yeah.
- [30:33]
B
Okay. And this way like the three separate clause have like different context windows and it's not just like less bias.
- [30:38]
A
Yeah, exactly. And they will all use more compute. I think like they'll put more like, you know, less likely for it to like stop early. Yeah, think more.
- [30:46]
B
Yeah, okay, got it. Awesome, dude. Well, let me ask you some like high level questions, dude, since it looks like you. Do you mostly use Claude in the terminal or do you actually use the app?
- [30:58]
A
A mix of terminal and desktop? Yeah, depending. Obviously we do a lot of like what we call ant footing. So it's sort of like, you know, whatever. I feel like I need to test the most. Yeah, got it.
- [31:10]
B
Do you have any tips for like. I feel like sometimes I get exhausted, man. Like there's like five threads going on at once for five different things and then like, I'm just like, they're constantly paying me. It's actually worse than having back to back me things in some way. There are tips for how to keep your own cut window clean or you Know.
- [31:27]
A
Yeah, I mean, this is a good question. You know, I think that this is like my goal for this year is to be more productive, but work less, you know, and so I think this is like, I think something we should all sort of be able to push ourselves on. I think that like, you know, like, I, I think one thing we try and do is sort of like I try to have like one project I'm focusing on on, you know, and, and so even if there are other things that I just need to get unblocked, like I need to have this build and merge and you know, things like, and explore something. Like, I think one project that I'm really focusing on is very helpful just because I find that the thing that cost me the most time is when I'm like, yeah, maybe a little bit lazy. And, and now I've like done a bunch of. I'm like sort of multitasking these things. I do a lazy prompt and then I'm like, oh, like I've waste this time now, you know. And so yeah, I think that there's like, just. Because like, I think there's a optimal amount of multi clotting depending on, you know, who you are, what you're doing. But for me at least it's something where like I tend to have like one task where I'm really focusing on the most. Yeah.
- [32:37]
B
Okay, so maybe like the agents take care of the rest or something, but like.
- [32:41]
A
Yeah, or even prioritization for yourself, you know what I mean? I think can be like a good like angle, you know?
- [32:46]
B
Yeah, makes sense. And like, do you have concerns? Like, I worry that like, you know, when. Because the agents like work pretty hard and keep going and like, yeah, maybe this is my problem. Like, I don't actually, like I said, I don't actually read everything that it pre produces carefully. So then I worry that like the whole report just turned to slop at some point. Like if I don't like. Do you have any routines or jobs that clean this stuff up periodically or.
- [33:12]
A
Yeah, so I mean, obviously we have things like Simplify, which Boris put out that like, you know, sort of simplifies the repo. I think it depends on what you're using it for as well. You know, like I think someone like when you're using it for the outputs, like you're using it for the, like the video outputs, those are pretty like, you know, maybe the quality of the code matters less, you know, and the agents are very like, you know, persistent and they'll figure it out. But I find Organizations often more for me than the agent to sort of like, you know, make me feel better about the workspace if I'm only caring about the outputs. But yeah, you can always just ask it to simplify or organize.
- [33:56]
B
Yeah, okay. Okay, got it. And you try to keep this. Do you try to, like, minimize and try to keep the cons window clean? Like, you try not to, like, have a bunch of, you know, MCPS on or like, you know, cloud MD has a super long. Like, did you try to optimize that or.
- [34:11]
A
Yeah, I mean, I think that, like. Yeah. Yes. I think that one of the things we've noticed, especially as the models have gotten smarter, is that like, you know, we'll be talking more about this. We cut down the cloud code system prompt by 80%. And the reason is that as the models have gotten smarter, they need less direction, they need fewer constraints, and they need fewer examples. And so a lot of our system prompt would be like, okay, here's the bash tool. Here are like five examples of using the Bash tool. Never do it in these cases, right? And the models are now kind of aligned enough that they know, like, hey, we don't need to. Like, we don't. The examples are almost constraining it because now it's like, oh, like you want things like this example. And so if you remove the examples, it actually can be more free format. And like, the constraints can also constrain it too, because like, you oftentimes when you say never, you don't really mean never. You just mean like, most of the time, don't do this. And if you give it the reason you don't want to do it instead, that can be more effective than the like, don't do this constraint. So I think that all this to say is that you want to trim your context. I feel like cloud MBs are probably too long right now and you probably want to shorten them more and more. Probably a lot of skills are too long. I think MCPS depends on the mcp. I think obviously some of them can take up a lot of context. But I think the MCP team has put a lot of work into this to make it better with tool search and things like that. But yeah, definitely Cloud nds like any sort of instructions. The models just need more room to run.
- [35:59]
B
Oftentimes then do you like, instead of saying, for example, let's say writing a Twitter post instead of saying, hey, make sure it's 280 characters or less and do not do this, do you give it more principles to follow? How do you prompt it then? Is it more like.
- [36:16]
A
Yeah, I think a Twitter post is a good example where you might give it context on yourself. So you're like, hey, I am. I work on cloud code. Anthropic. This is sort of like some of. Yeah. Some of the principles we follow. Obviously 280 characters are still like a good, like, sort of constraint, because that's like, it does need to happen. Right. But you can even just say it's a tweet and it will, like, sort of know, like. Or this is a good example where you might say, like, hey, keep it under 280 characters. But let's say that maybe there's a version of it that's better as a tweet thread with two tweets. You know what I mean? And so giving it the sense of like, hey, I want to write a tweet thread. I'd prefer it to be one tweet. You know, gives it more freedom and flexibility. Yeah. To find something good.
- [37:02]
B
Okay, got it. Okay. And then let me ask this. So Boris has been saying, like, coding is like a solved problem. Right. And I feel like just like becoming more technical these days is like a little bit. I guess my question is, for someone like me who wants to actually learn how to work with agents more, to actually learn what they're doing, how do you actually become more technical? It's not about the syntax or anything, right. Is it just using it more or how do you.
- [37:30]
A
Yeah, I mean, I think for the first step is motivating yourself to learn things. You know what I mean? I think that is honestly really hard. And I find myself doing this as well, where if you don't need to learn something to get the job done, maybe you won't. But I think that it is really important. I think the goal of learning to be more technical is to know my unknown unknowns. There are some things that being technical and knowing the syntax of TypeScript is not really very helpful. But I think being technical and knowing, okay, hey, what are the trade offs of different backend services? And, and what are the different video encryption libraries and how do they work? And what's the difference between a local or remote video encryption or transcription library? These are all, I think, pretty helpful. And so I think oftentimes I'm trying to learn the constraints of the system and so what is possible? How is it doing it right now? How good could it be? What if we did something else? And Claude can often brainstorm and teach you this if you push it. Right. But you really do have to push it. I think like this is like, you know, just the hardest thing about education and what everyone says is sort of like it feels good to like try and learn something, but like actually learning something is more work, you know, and like it should feel like work. I think Kaparthi says this a lot, right? Like education should feel like work more than fun.
- [39:01]
B
So. Yeah, yeah, that's actually because actually easier to just like for example, the video thing. Right. Like it's actually easier to just keep prompting cloud to just do it and then look at the output and then like, like not actually learn anything, dude. Like just like, like go figure it out and then it figures out. But the fact that you actually generate these like pretty detailed HTML reports and read them, I think it's probably the exception to. I, I think most people don't do. Do this.
- [39:25]
A
Yeah, I mean, I think it's something I'm trying to push. Yeah. But like, I think that like this is how you can learn to make. Just like, I think the question for you is like, okay, like, how do you go from making good shorts to. To making like really like the best shorts, like highest production quality shorts out there? You know what I mean? And I think if that's the goal, right, like, I think you need to like, then sort of probably learn more about like both about video production and editing and, and like the technical concepts. Right. And then push yourself that way. And I think that like, I think we all want to push ourselves towards like being better and faster and not just faster. Yeah.
- [40:01]
B
So, okay, so maybe I'll add some custom instructions to my cloud, be like, hey, make sure you generate HTML reports for everything so I can read them.
- [40:09]
A
Yeah, yeah, yeah. Well, I mean, you know, yeah, I think it's something you want to push yourself to as well, you know, to like figure out when. When do you want to understand how something is working?
- [40:18]
B
Got it. Awesome, man. Thank you for giving the gift of cloud code to all of us. And yeah, really excited for what's next. Dude, I don't know if you probably can't share anything, but yeah, really excited for what's next.
- [40:29]
A
A lot of what's next is CloudTag and it'll keep getting better, but I think it's just something that it's hard for us to understate how much it's sort of changed work at Anthropic here and so excited for everyone else to get on board or to try it out.
- [40:43]
B
Yeah, if you have a really capable employee, you're not going to micromanage them. Right. You just, hey, tag them on Slack and be like, I can't do this, then hopefully.
- [40:51]
A
Or you want to talk with it iteratively with other people and collaborate. Yeah.
- [40:55]
B
You should build it to a point where I can walk by Claude's desk and just, like, ask us some questions.
- [41:00]
A
Yeah, Like a robot. Yeah. I mean, that could be a hack project for you. You know, I feel like you could already do that. Yeah.
- [41:06]
B
Cool. All right, Tariq, well, really great chatting. And, yeah, I think people know where to find you online, so I don't think we have to talk about that.
- [41:13]
A
Yeah, sounds good. Sounds good. Yeah, Amazing. Thanks, Peter.
- [41:16]
B
Appreciate it.