Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Making opinionated AI tooling decisions with Nimbalyst's Greg Hinkle
0 Plays2 seconds ago
Greg Hinkle, co-founder of Nimbalyst and former VP of Software Engineering at Salesforce, joins Scott to discuss the future of AI-assisted development. They explore the challenges of managing multiple AI coding agents, finding flow state in an agentic world, and why visual workspaces matter. Greg shares Nimbalyst's opinionated approach to integrating tools like Excalidraw, task management, and session organization directly into the developer workflow.
https://nimbalyst.com/
Recommended
Transcript
00:00:00
Speaker
Yeah, I've had some moments of absolute pure delight in the sort of interactive voice. You know, it's it's still in testing, but... it I basically have a voice agent that can control coding agent. I can control a whole set of coding agents and it's I'm talking to it and it's firing off prompts and it's just a much higher rate of speed and it can wake up and and notify me. And I can be walking around you know the yard having a conversation controlling multiple agents all through my my headphones. It's pretty it's pretty cool.
00:00:30
Speaker
Yeah, it is cool. And I think also it's worth noting that people oversimplify and they think it's, I talk to it, it talks to me. But the trick is you can talk faster than you can type and you can read faster than it can talk.
00:00:43
Speaker
That's right. Hey friends, I'm Scott Hanselman. This is another episode of Hansel Minutes. Today I'm chatting with Greg Hinkle. He's a chief technology officer and co-founder at Nimblist. was previously at Salesforce and was a manager of engineering at Red Hat. So you've been doing this for a minute, sir. How are you?
00:00:59
Speaker
I'm doing well, thanks, Scott. How are you? I'm okay You know, there's so much AI going on right now. People keep saying like, why are you talking about it? And I realized that it's not about AI for me. It's about software engineering.
00:01:10
Speaker
of which AI is the tool that is kind of like blowing up software engineering. And if it was 20 years ago, I'd be talking about, you know, build servers and continuous integration and, you know, agile and TDD and things like that. But trying to, you know, figure out this moment that we're in and how to build software correctly and still value the craft is something that I'm really interested in. So that's why I wanted to talk about, about Nimblous because it feels like you're trying to navigate this this space of multitasking. Multitasking seems to be the the skill and then expressing your taste, you know, expressing your, your judgment when the AI generated pull requests and diffs and stuff comes in and Nimblest takes a very opinionated approach.
00:02:00
Speaker
Yes, yeah, we're we're trying to, you know, thread that needle of changing the way the experience is is it's happening for a user by just giving them a different interface, focusing on what the user needs to see as they're managing many agents doing many jobs. So it's really about, you know, increasing the bandwidth for the user so that they can keep control of the chaos.
00:02:24
Speaker
The chaos is true. like I was talking with Mark Rusinovich, my buddy at work, about what's the right number of agents that I can manage comfortably. And he and I feel for us, it's three to five. But then I've got guys at work and gals at work that are doing 10, 20 things, and they have squads of squads of agents and things like that. And I'm finding that the context switching is just too much. The cognitive load for me running all these basically terminal tabs, that's as far as I've gone, is terminal tabs. And it's hitting the wall pretty quickly.
00:02:53
Speaker
Yeah, that's exactly right. And we we have this opinion that keeping track of everything goes beyond just having tabs. Tabs are just too short-lived and a lot of, you know, it's funny, it's sort of like a hurry up and wait where you come up with this great idea, but then you have to wait for the agent to go do the work or go do the research.
00:03:14
Speaker
And so you're not getting the best bandwidth if you're not multitasking, but you it really, you have to carefully multitask. You have to do prep work, have the agents go and do research, and then stop so that you can come back when you have time to work on that particular task. And that's what I find is the best way to improve the bandwidth.
00:03:35
Speaker
Yeah, that's a great point. increase the parallelization. Yeah, the throughput, the throughput for sure. It feels like, you know like i again, I'm waxing philosophic back 20 years ago, like it was waiting for the build. And now it's waiting for the agent to turn.
00:03:48
Speaker
So some people just kind of like sit there and cross your arms and wait for the thing to turn. But at the same time, I don't want to multitask all the way to a drastically different thing. I want to get in the flow. And Mark and I were trying to figure out do we Have we lost the flow and what have we done to our tooling that is preventing us from getting into that that that flow state, that hyper-focused state? and I wonder how much you've been thinking about, you want to multitask, but it shouldn't feel like multitasking, if that makes sense? shouldn't feel like you're ripped out of one task and thrown into another.
00:04:25
Speaker
Exactly. It's the context switching that's the pain. So what you're really doing is trying to do all the prep work so that when you do context switch into a particular task, you can really focus and spend, you know, that 20, 30 minutes doing nothing but thinking about, you know, what did the agent come up with? Is it the right thing?
00:04:44
Speaker
Testing it, using it, looking at the code, having another agent review the code and really talking it over with that agent. That's what I find the best you know results come from being able to focus while still multitasking.
00:04:57
Speaker
Yeah, indeed. The other thing that I think it does a well, Nimblest does well, is you're bringing some some skills and some work along that is already in our lives, like Excalibur and different mock-ups and tools like that. And you're you're bringing them out of the box because I usually end up telling the agent, oh, may make me some ASCII art mock-up. And I get kind of a weak sauce response.
00:05:20
Speaker
ASCII art with a non-aligned right margin and everything like that. But you've got skills and commands built in for for stuff like that. You're actually acknowledging that these tools exist and then baking them into the flow directly.
00:05:33
Speaker
That's right. We think if you have a single, you know, pane of glass, that is all of the tools that an agent that you need to work with an agent to be able to, you know, build, to build product, to build technology. If all those tools are in the same place, then it's not an agent making MCP calls all over the place and you're using different interfaces. You can you know, work on the mock-up and the design of something. You can work on the data model with the Prisma, you know, graphic design. You can design the architecture with Excalibur, and you can do that all in a single set of agents, agent sessions that don't lose context. But this is quite opinionated and I think this is the moment, right? None of us are experts. We're all trying to figure it out, but you have an advanced and extensive experience in software engineering. Like I said, that you know you were VP of software engineering at Salesforce, so you know what you're doing. When when you at Nimblest are making a decision about what is the opinion that we're going to make into the product, like Excalibur matters, PDF viewers matter, data model like you know model data models matter.
00:06:37
Speaker
How do you decide what's the right thing to plug in out of the box? The biggest thing I can say is that i use it and I see whether it helps me. and you know I'm using all of these tools to build Nibbalist and I've got my data model and i and my EMT and I'm trying to figure out is this the right storage model for multi-agent storage of all this you know session behavior.
00:07:02
Speaker
And if it works for me, then I figure it might work for some other people. Now the planning stuff, you've built in a a little SQL database, you've got plannings, decisions, bug bugs, tasks.
00:07:13
Speaker
I noticed that that is being checked in, you're versioning that, you're plugging that into the flow as opposed to like using GitHub or or Jira. Is that a conscious choice as well? it's It's one of those things we're experimenting with right now. Some of our files, we do end up checking into Git as just the mechanism to share, but we're also experimenting with you know basically collaborative document sharing.
00:07:35
Speaker
So real-time markdown editing, it's essentially you know a local Google Doc you know sitting inside of NimbleList. and And we're getting some pretty interesting results from that in sort of the coll collaborat team collaboration side.
00:07:50
Speaker
Yeah. Now, hopefully you'll, this is a compliment. Don't take this wrong, but like it doesn't, it doesn't feel like vibes. It feels like a software development lifecycle tool. And the reason I say that is that you put the file viewer upfront.
00:08:03
Speaker
You're not trying to hide files and diffs from me, which I'm finding to be somewhat unique in this space. So many AI tools are hiding my code from me. I think that's true and and nothing think against those tools. You know, they're they're fantastic for, you know, the jobs that they're trying to enable. But if you really, if the end result is great software and it's the entire product, including all the work that you put into designs, that's all of your, you know, work. And we just think that, you know, open formats that are on disk, not sitting in some, you know, database or tool or out in the cloud, or, you know, you get sort of just a thin view of it. That's like, here's the running view, but to get to the code is, you know, challenging. We figured just, you know, local first, open formats first, use the tools that are out there and make them accessible.
00:08:57
Speaker
Mm-hmm. Now you've got Claude, you've got OpenAI, you've got pluggable agent providers. Also, notably, you have my friends at LM Studio. I love those guys. So you can connect to local LLMs. Have you had good experience with that? I feel like I need a really beefy machine to get valuable and insights.
00:09:14
Speaker
I think that's true today. you know, it depends on the tasks. Some people are using Nimblest not to just software code, but to, you know, just do prototyping um or even other business tasks. And I think they're finding more success. The coding tasks with the just the size of the context needed is still a challenge, but you also see advancements every week. And I think the models are going to get smaller and more efficient. And we're going to do more local work in the future as well.
00:09:43
Speaker
I know I'm jumping around a little bit, but I've got like 40 tabs open and I've got Nimblest open doing two different things right now. And one of the other things that i think you do that's really cool, which is opinionated, but I haven't quite cracked it in my head yet is the Kanban stuff.
00:09:57
Speaker
How is that stored? Explain how you think about Kanban boards, because you can move around from an inbox item, like a thing that's happening in a session, move it over to the backlog, move it over from planning to implementing and move it you know all the way through the Kanban. Can the agents take action on that? And could I start?
00:10:14
Speaker
That's where the visual things start becoming interesting to me. Yeah, that's right. And we're experimenting with more of, you know, the kind of collaborative agent task management. So the the agent can create the tasks on the Kanban and manage the flow of the sessions from, you know, planning to implementing to user validation.
00:10:35
Speaker
And all of that's, you know, customizable by the user by updating their agent files and saying, hey, you know, when Nimblist, when you create a plan and we're going through these steps, here's what, here's the, you know, the columns I want to use and here's what I want you to move it to ready for development or, you know, validating.
00:10:52
Speaker
And, yeah you know, you can let the agent help you organize the the chaos. Yeah. And and it's one of the chaotic things that I find myself doing a lot is is markups where I'll have my agent on one side and then I'll have a playwright or a markdown file or some kind of mockup or some live HTML. But with Nimblest, and help me understand how you do this, you can circle the screen, say, put something there and then mention it.
00:11:18
Speaker
like put this here or put it in the circle area. So you're you're you're closing a loop that I didn't realize needed to be closed because I usually find myself trying to describe the the thing, like use this CSS ID when I really just want to point at it and go that one.
00:11:35
Speaker
That's right. And we have, you know, within our extension system, there's this API to do a deeper level of integration with the agents so that, you know, when you're in the ERD diagram, you can click a particular, you know, table and say, these are the changes I want to make to this. And you don't have to tell it what table it's getting that as a link or in a mock-up, like you said, drawing. And so we're just trying to improve that.
00:12:00
Speaker
that bandwidth between the user and the agent. And I think things like drawing or, you know, pointing, you can you can make a lot of, make it a lot simpler for for user to instruct an agent, because it's really about how much control can the user deliver to the agent and how much, you know, how much typing. So I think voice is also going to be really important.
00:12:22
Speaker
How do we improve that bandwidth? Yeah, when you think about voice, i I would like to be able to not sit down. I wanna stand up and steeple my fingers and walk around my office yeah talking to it.
00:12:35
Speaker
you know And if you think about something like a Prisma file or looking at a database schema, you know I can edit the Prisma file. I can talk to the chat and let it edit the Prisma file. But I think too many AI tools fall into the all input goes through here now. And with Nimblist, you seem to be walking a tightrope between talking to the chat and editing the actual files themselves.
00:13:00
Speaker
Yeah, that's right. And, you know, if there's a ah way that we can get to multiple inputs, you know, things like talking to AI while drawing, you know, here's what I want you to do here. I think there's just some really interesting things that we could do in the future.
00:13:17
Speaker
you know i was I was talking to a friend who is a certain age about ah Star Trek. I'm not sure you knowm not sure how old you are, but Star Trek 4, when Scotty goes back in time and he picks up the mouse and he's on the computer you know like,
00:13:30
Speaker
Of course you would talk to him. Why wouldn't you talk to it like that? And the guy's like, just use the keyboard. I feel like we're right there, right right on the edge. We're almost there. Yeah, I've had some moments of absolute pure delight in the sort of interactive voice. you know it's It's still in testing, but...
00:13:48
Speaker
it I basically have a voice agent that can control coding agent. I can control a whole set of coding agents and it's I'm talking to it and it's firing off prompts and it's just a much higher rate of speed and it can wake up and and notify me. And I can be walking around you know the yard having a conversation controlling multiple agents all through my my headphones. It's pretty it's pretty cool.
00:14:09
Speaker
Yeah, it is cool. And I think also it's worth noting that people oversimplify and they think it's, I talk to it, it talks to me. But the trick is you can talk faster than you can type and you can read faster than it can talk.
00:14:22
Speaker
That's right. It's asymmetrical, isn't it? Yeah. Yeah. Yeah. I mean, and not just, you know, you can read faster than it can, it can speak. You can just the higher bandwidth of like, let's see a layout um or a diagram. There's, you just can convey so much more information than trying to explain it.
00:14:39
Speaker
So yeah, absolutely. Yeah. Is it fair to say that this kind of visual stuff has been like a bigger part of your career? Cause you worked on interaction studio at Salesforce and you came in from your first company that you founded Evergauge.
00:14:53
Speaker
So like, has this been percolating in your brain since even before ai coding agents? Well, I always loved focusing on the user experience. You know, different jobs have different requirements. And so at Evergauge, we focused on personalization, helping to, you know, confine and narrow up any experience, any digital experience, so that it was, you know, tailored to what the user was trying to accomplish.
00:15:16
Speaker
And so, yeah, I think there's a there's an angle of personalizing the, you know, the, the painted glass for the job that you're trying to solve. I think that's always been something that interested interested me.
00:15:28
Speaker
um I love data visualization and user, you know, user experience elements. And, and I think we're just in an incredible age right now where, you know, the, the, what's possible is advancing so quickly that you just have to try things.
00:15:43
Speaker
i I feel like I'm pretty far along in my journey, but I'm learning like every day. But I'm also, I think, a little bit hamstrung by knowing how it absorbs the tokens. And I wonder in your experience, as you're creating data models, mermaid charts, Excalibur draws, mockups, it can't see stuff, even if you give it a screenshot.
00:16:08
Speaker
it can't really see stuff or can it? Like how much better or is your result because you're planning visually than my results, which is a lot of pros. Yeah, I mean, I think that's true that the, you know, the the agent doesn't get any benefit necessarily from looking at a ERD diagram instead of just the raw Prisma.
00:16:31
Speaker
but But the user does. And so a fair bit of what the agent does, you know, I'm perfectly fine with it spending tokens to improve how it communicates with me. And I kind of feel like that's that's what we're doing with this.
00:16:44
Speaker
Hmm. That inverted it the way I thought you were going to. That's interesting. So it's as much about me understanding what's going on as it is about the agent. If I'm going to control it, if I'm going to steer it and, you know, accomplish the task I'm trying to accomplish, I need that that support from the agent. I want the agent spending its time trying to help me control it.
00:17:05
Speaker
So, yeah. Now you mentioned that you're using Nimblist to write Nimblist. And i was always fascinated when I was talking to like Anders Heilsberg on, we made C sharp and it's like, now we've got the compiler building the language, the compiler builds this itself now. How, what was that like? When did you get to the point where like Nimblist could basically talk to itself and make its own self? Cause I assume you started, you didn't vibe the thing, but now you're in this line of loop where it's building its own, its own body.
00:17:33
Speaker
Yeah, we it was pretty quickly that we got to the point where the coding happened more Nimbolist than anywhere else. And... you know I still have other IDs that I use and whatnot, but the you know people talk about the harness. And a lot of what I've done with Nimblist over time is create more and more control for the agent with Nimblist. So you know as the agent is you know doing coding on Nimblist, it can read the log files live. It can you know query the database live. It can read the settings live. It can do that.
00:18:10
Speaker
actually navigate the UI and take screenshots and you know look at the DOM. And so now you know I'm trying to get it to you know control and optimize the you know the layout of something. And I can say, you know here's what I want it to look like. And now it can basically loop until it's done.
00:18:28
Speaker
And it's not a forced Ralph loop where it's just banging on the keyboard until it gets it right. like it's It's executing and goal seeking towards a plan. That's right. Yeah. You know, the the more complicated the the task, you know, the more time you want to spend on the harness on, you know, did the right results come out? The test cases are the test cases, you know, covering what we're trying to accomplish. It doesn't, it isn't necessarily going to do the best thing yeah or have the best taste when it's like, hey, design me a UI. And it's like, well, I designed it and then I took a screenshot. It's not usually going to fix, you know, what you
00:19:06
Speaker
things that a human would would find wrong with it, but it'll make it work. you know It'll make it work and exercise it, make sure it's functional, and then the human can come in and but the the taste and the touch on it.
00:19:18
Speaker
So you've got both Codex and Cloud Code today. It's all pluggable. So you're an AI-native workspace built on top of each of them. Do you bounce between them? Because I use GitHub Copilot, and I feel like Claude is fine. It's great. Opus is good.
00:19:33
Speaker
But if I only had Opus available, I would feel frustrated. And I feel the same about Codex. So I like Copilot because I can bounce around between models. I'm assuming you did the same thing and you're switching models as you do your job.
00:19:45
Speaker
That's right. Yeah. Every day I'm, I am having one agent review the work of the other agent. I find that extremely valuable. And so we have this concept in Nimblist called work streams where I can kind of group sessions. And I do that frequently. I make a Claude code session and then have Codex review the plan that Claude code came up with it. There should be usually find some errors that I'll have Claude code implement.
00:20:08
Speaker
I'll have Codex review the implementation. It's, you know, it's a cycle. Okay. but it But I love using multiple models. Yeah, yeah. And then we've seen things like i had Steve Yege on the show a while back and talking about things like beads and squads. And then there's a fellow at Microsoft named Brady Gaster who has this thing called squads. It's Gastown.
00:20:28
Speaker
you've You're building in kind of automations and tracking and things like that as well. It seems like you're kind of borrowing from things and building them into Nimblest. But i can I still use my favorite wacky squad thing or agent thing or Gastown or would Nimblest fight with me?
00:20:45
Speaker
and Yeah, no, I think you you absolutely can. And I think what we're trying to do is figure out the best way to lay that out within the user experience. Because we do we we built in teammates, the you know, the Claude version. And I find it to be, I don't use it as much as I thought I would use it.
00:21:04
Speaker
It's just... Isn't that funny? you know I feel the same way. Yeah, I think it's just, there's only so much you can do by instructing a model. And so when I'm sitting there going, okay, actually I'm finding more value for one model reviewing another model's work than I am from saying, pretend you're a designer, you know.
00:21:20
Speaker
um You know, that's a conversation that people need to have. And i've been I'm interested for folks who are who are listening right now or watching this on YouTube to let me know because they really want me to go and use these squads and then like, you're an expert database. It's like, you're not.
00:21:38
Speaker
And telling them that you are, does it they're they're like, no, it it fires synapses and it spins it. I'm pretty sure it's not how it works there, bud. The only argument that I can get for effectively anthropomorphizing my teams and naming them and making your the database and you're the is is context window savings. Like, okay, you are thinking about and focusing on React so you don't have to hold the whole thing in your head.
00:21:59
Speaker
That's right. I mean, but that is the one thing I do find that it actually valuable with is knowing the what conversation I'm going have with that particular ah agent, that teammate. you know We're focused on one area, so I can improve my you know conversation with that agent by having a focus for it.
00:22:17
Speaker
That's actually the more valuable part is you know just interacting with me yeah than necessarily that it does a better job. Well, you know strong opinions weekly held. Certainly, I don't want to say that this is...
00:22:29
Speaker
wrong, but it does feel like it's not right for me yet because it hasn't clicked. And if it was obviously better, we'd all be we'd all be doing it. But it's it's fascinating to watch people swarm up on a problem with with anthropomorphized specific headers effectively. Like you're you're the UI, you're the tester, you're that.
00:22:51
Speaker
But it certainly, it sounds like nimblest won't stand in my way if I choose to do it that way. That's right. yeah Yeah, that's pretty cool. You said you use other tools as well. I assume like what VS Code or cursor, or you're bouncing around seeing what different people are coming up with and different ideas?
00:23:06
Speaker
Yeah, I use them. I use at at least VS Code, Cursor, and you know the JetBrains series on a on a daily basis, um if not a weekly basis. But you know it's it's largely about making sure i still have contact with how other people are thinking about this because it's it's a leap to go from...
00:23:27
Speaker
you know the code first experience or seeing what some others are doing, which is you know really flipping it and saying, well, forget the code. Let's just do you know session management. And I think I'm just trying to find that balance that's probably somewhere in the middle.
00:23:42
Speaker
Yeah. that It is such a challenge to to manage so many. like it's it's We all multitask now. like i'm I'm almost... Like I can said, hamstrung before, but it's there's an analysis paralysis that happens and the paging back in, like, what was i what was I working on again in that other session? I want to give it as much time as I can to turn, to to do its job.
00:24:08
Speaker
I was actually, let me bring me back up. I was in an interesting little lowercase a argument with a coworker today. I tend to do what myself and Mark Persinovich call code code sculpting,
00:24:19
Speaker
where I give it a good-sized task and we rough it out. And then it's sculpt, sculpt, sculpt, sculpt. And you might do 40 to 50 turns. And my coworker is still, three hours later in the day here, working on the plan.md.
00:24:35
Speaker
which i I call it the prose compiler. You know, he's trying to make this giant 40 page spec. That's so clear that it would build in an autopilot. And i I keep coming back to like, it's not a compiler and you could run that plan five times through five different models, you get five different products because the plan is never going to be as specific as the code. So like,
00:24:59
Speaker
i'm I'm working analogies in my head, you know, like ice. We know how you take a chainsaw and they make like ice sculpture. They start with a giant block and they just hack at it. They just hack at it. And like kind of that kind of shaped like a horse.
00:25:10
Speaker
And then the work starts. are you Are you a big upfront planner or are you a code sculptor? Well, the models have really advanced in the past year. You know, a year ago, yes, you really need to spend time on a plan if you were going to get valuable work out of it. I think what has happened is I've shifted. i still do plans for just about everything. Really?
00:25:33
Speaker
What I've focused on is making sure that the plan isn't the instruction to the agent. It's the documentation of what I want. Because I like being able to come back to it. I like being able to reference a second session to that plan and say, you know, is this correct?
00:25:49
Speaker
You know, did you forget anything? And and specifically, it's not about... this is what i this is how I want you to solve this agent. It's about here's the motivation behind it. And here are the important you know aspects that you can't miss.
00:26:03
Speaker
And that's that gives me durable long-term mapping to what I'm trying to accomplish, not just one single session that I gave it instructions and maybe I spend time you know sculpting the code, which I do, you know especially naming and and organization and architecture.
00:26:18
Speaker
But that the why is is critical to document. And so that's what I end up doing in those plan files more and more. It's funny. Sometimes I feel like I will get the code done, review the code, and then I'll back up and I'll say, write me a markdown to talk about what we just did. And then I wonder if I'm being a phony doing that.
00:26:35
Speaker
Yeah, I do that a lot too. Those are not my plan files. Those are my design architectural, you know, they're they' doc historical documents. Exactly, my artifacts. So, um and and I do that a lot too, especially, you know, if I accomplish something that's big and important, I don't i don't want the model to, you know, veer off. And so it's really, it's an anchor.
00:26:56
Speaker
I do the same thing. What I'll do is I think about it almost on a subsystem by subsystem basis. So like I had a really interesting problem with dots per inch issues on windows and multi-monitor. And I felt like we, the collective royal we, had accomplished something. I said, all right, write down everything that we learned you know for future generations and myself tomorrow so that we can remember. because i don't because like why did we we were Because we weren't just turning. We were Googling and Stack Overflowing and all the classic programming and just thinking. And then it was like, oh that's how it works. And I know for a fact that tomorrow morning, the AI is going to have forgotten it all.
00:27:34
Speaker
So like if we ever get it again, yeah. But but you got to keep that sort of, you know, the creating that that knowledge management of just all of the work. And there's a lot of things that are not code that come out yeah that that are part of that knowledge management. but Another thing, and just as we get towards the end here, I wanted to call out is that you were quite early on the idea of managing agents on the go.
00:27:54
Speaker
Like everyone everyone everyone's moving fast, but you were thinking about the idea that you're going to have to connect to these because the agent's running on your machine and you've gone to lunch, you're gone for a walk, and you might just need to say yes, or you might need to steer it.
00:28:08
Speaker
ah Is that a mobile app that you pair or is that a tunnel through Ngrok? How do you do that? Yeah, we just wanted a you know, kind of seamless native experience. So we built um iOS first, Android coming. But, you know, it's it's really just nimblest in a phone experience that's just, you know, a glass int back to your machine.
00:28:29
Speaker
We're doing it in encryption, tunneling, just to make it, you know, seamless. You're using secure web sockets through nimblest servers, but it's ultimately exact tunneling. Yeah, exactly. And, you know, just giving you that that that high bandwidth ability to answer a question to the you know to the agent or to get notified and say, oh, it's done, you know, so I can be, you know, making lunch and you know, get notified. I find that really valuable.
00:28:59
Speaker
Sometimes I'll go for, you know, walks out in nature and I'll be able to have a little bit of a voice conversation, kick off a few researches. I'm not necessarily coding at those times, but I'll be like, here's an idea. Let's, what about this? Go do research on that. And I find that, you know, freeing.
00:29:15
Speaker
You know, some people think it's funny that you say that it's freeing because I find it freeing as well. Some people say, well, you're just tying me even more to the work, you know, but I think that they don't understand that the personality of the programmer is such that if if if the agent could let you know when you're in the shower that it had just had an important insight, most programmers' personalities would want to know that right away because we need that little dopamine hit that says, we solved it. Let's go. Let's go. don't want to lose 12 hours because I forgot to check a notification.
00:29:44
Speaker
That's exactly right. And you know programmers notoriously sit at our desks all day. And you know this feels it feels like this gives me the freedom to not feel like I have to be to be able to keep my agents busy. And i had that yeah that does sound like freedom to me.
00:30:00
Speaker
Yeah, it does sound like freedom. So today, people can go to Nimbalist, N-I-M-B-A-L-Y-S-T, Nimbalist. And it's it's free for individuals. you just check it out. i downloaded it. Didn't take any time at all to get it set up. I have Claude and Codex already. It picked it up.
00:30:16
Speaker
um Right now, it works on on Windows. It works on Mac. It works on Linux. Is it a Tower Eye app or an Electron app? Electron, yeah. Electron app. So it works everywhere, works cross-platform. How do we give you feedback? How do we you know get involved? Because you know if I bump into an an issue on on Windows or it doesn't work the way I expect, where would I talk about that?
00:30:38
Speaker
Well, we've got a great Discord server, so feel free to drop in there. And nice there's a feedback button right in Nibalus where you can drop us notes. So yeah, love to hear from you. Yeah. There's also a community on GitHub just at the bottom footer there of nimblelist.com. You can see the community section, both gith GitHub, Discord, and then all the socials. You're everywhere.
00:30:58
Speaker
That's right. Very cool. thanks so much, Greg Hinkle, for chatting with me today. Thank you very much, Scott. This has been another episode of Hansel Minutes, and we'll see you again next week.