I'm Harry, as you know. Thank you, Reggie. You've definitely described MindStone better than I can do, so thank you very much for your intro.
I want to talk to you a little bit about how I work now and how we write software at MindStone. So hopefully, kind of, I don't know, we like to say at MindStone we like to live in the future in terms of using AI so that we can help our customers do the same thing for them and their teams. So hopefully, I'm going to give you a glimpse of that and my personal story at MindStone.
So, like Reggie said, we offer AI programs and we like to describe ourselves as an AI transformation company. We have training programs. We teach you how to use the tools and AI tools.
And we also have an enablement product called Rebel, which is, I'll demo it, because Rebel is an AI agent application, which sits on your desktop and it has access to all of your file systems and has shared memories in your team so you can really use it in your team and be productive with AI. I will demo that afterwards.
So what do I want to show you today? Yeah, a bit about my personal story and then I'll demo you how I write code now to give you an idea.
So I started at MindStone over three years ago now, and I joined as an iOS engineer, just working on our iOS application. And I wrote code kind of traditionally by hand, like carving it out of stone, like line by line. And it's all changed now.
So, yeah, my thing, I trained for a long time to focus on one task at a time and to do one thing, get it finished. Sometimes that would take hours, sometimes it would take days or weeks, depending on the size of the task. But now that's all changed and it's gone on its head.
I'll show you how I write code with a live demo. This is my coding agent. At the moment it's Codex, but we changed it a little bit. I'll just give a live demo of some of the tasks I would do.
so can you change the send button on rebel to not use text but actually use like an icon which explains what it does so I would send off some tasks like this and that would take a few minutes do its thing so now I'm in a weird position of
like what do I do this time like do I switch focus do I go get a cup of tea and if I go get a cup of tea my caffeine levels would be through the roof so that's not a good thing. Normally I would open up a new tab, start on a new piece of work.
That first one was a really simple thing, but I want to show you a bit of a deeper prompt what I could do to start a deeper piece of work. I want to add a new tab to the top navigation bar in Rebel.
This tab would show me a heads up display of what I have on today, how to to prioritize my day and a really simple calendar display, use some sub -agents, research best practice when it comes to UIs like this, ask me questions if needed, use our existing machinery and models, and use the chief engineer as well to do this task.
So I said a lot of things there, but I think the thing that I want to communicate is this This chief engineer, we've built a few different skills in our code base, and not going into too much detail, but this chief engineer allows us to write code in a robust way without having to do lots of human checks.
So it goes through a planning stage, and then it asks multiple different models to review the planning stage from different perspectives, and then it would re -implement the plan and then checkpoint with us to say, okay, does this plan look good, have we got any more questions?
and then we'd really spend time to shape the intent, what we want to build a given feature. And that's where I think the big change from my work has really come.
I'm no longer an engineer. I guess I've been promoted or I'm higher, which is a weird thing to say. I'm at higher level.
I'm just giving these agents intent, like spending time to really ensure that they've got a good understanding of what I want. And then I just let them free and hopefully come back when they finish the task.
So that being said, this has presented some new challenges in ways of working. Like sometimes I would have 10 of these tabs open and I'll be kind of, well, it presents two new challenges.
One is what I like to call agent anxiety and my teammates share this anxiety. anxiety, it's when I do not have an AI agent doing something in the background for me.
When nothing's running in the background, I'm kind of like, oh, am I wasting time? I don't feel productive. So, like, you'd see me that frantically before a meeting at the company, like, kicking off an AI agent to do something.
So that's a completely new feeling, and my colleagues are sharing that, specifically my product team colleagues, but also, like, Like people outside of a product team at MindStone, where, for example, people in the finance team or people in the sales team, when they need to use agents for long -running tasks, they are running into the same anxiety that I'm having.
And then there's another problem on the opposite side, which is agent overload, which is when I have 10 tabs open, I'm constantly switching between these tabs. I'm like where am I I've lost track of all of this like I'm just my brain
explodes so I feel like I'm still trying to find a sweet spot in terms of the number of agents open at one time to be optimal for me I think it's like six or seven and after that yeah my my head just falls off so um that gives you a
very quick kind of glimpse into like how how I'm working now and this thing
thing sitting at the top of my screen is an application we built to manage this context overload or cognitive overload.
It's just a little open source app for Macs which works for terminal agents and it gives you a heads up display of your agents what stage they're at.
If they come back with a question it will come up with a little question mark so I can really use that to manage the context overload what I'm having.
So that's open source? Yes. Yeah, that's open source.
And open source is kind of a new thing, but we're in our strategy at MindStone because we feel that software, well, we see that software has become a lot like less of a moat in terms of it used to be a long process where it cost a lot of money to hire lots of engineers and it took a long time. Now it's becoming a lot cheaper to write software, so we feel like open source has become more of a part of our strategy because of that, and it also opens new opportunities, like building a bespoke tool exactly for our needs wasn't really an option before, but now we did that in a few hours, and it's really enabled us to improve how we work.
work. We feel like, as I said, we want to be like the canaries in the coal mine in terms of using AI at work.
We feel that software is now, but other knowledge work is next in terms of AI and how we can really use AI to be more productive at work. That's about it.
I realized I didn't show the other platforms we offer in my demo.
This is our learning platform. It enables you to go through a structured learning program about AI.
And this is what you can, if you've checked into the event, you can win a seat on this AI competency program. It costs like, I think, 300 euros or something. think.
This is the community platform, which manages all of the infrastructure around these meetups. And finding the speakers getting their talks confirmed.
And yeah, it's actually quite a complicated problem. But that's one of our platforms as well.
And like I just showed you that is rebel, the AI agent desktop app that we built.
And you can see here here, that change did work, that send button changed into a little send icon down here.
So that is us.
I think that's everything I wanted to communicate.
We would love to, if you're interested in our open source endeavors or getting access to Rebel or we've also open open -sourced all of the MCPs that we've written for Rebel.
MCPs, I'm not sure if it's an acronym. So it's the connectors that we've built in Rebel to connect to your email and Slack, and we've built them custom so that they're secure and robust. We've open -sourced all of that code as well.
So yeah, feel free to talk to me afterwards or scan my QR code if you want to keep updated with what we're up to in the AI space. So, yeah, really happy to answer any questions that you have. Thank you.