@Nathan Campbell you are right to be nervous, I’ve seen it do some pretty crazy stuff although nothing destructive at this stage and the amount of work it can do on my behalf is staggering. But I have all of that stuff running in a stand alone PC separate from my other proper work stuff. And I just set up a Tech and AI Consultancy a few months back with an old colleague so it’s important we push some of these things hard to get a proper handle on various aspects of AI. That may sound like I’ve just jumped on the AI bandwagon but I actually had an IT support consultancy in London for about 15 years before moving to Canada and setting up a Geodesic Dome Luxury Glamping site which I’ve been running for the last 8 years! The advent of AI and the unexpected arrival just down the road of my former right hand man from my IT consultancy days led us to create a new company. Fun times 😁

@Nathan Campbell yep I would have gone the cloud VM route if I didn’t have a beast of a PC just “lying around” !!

@Nathan Campbell thanks for the inspiration to push a little harder on lander today, so I ended up deciding I wanted to get it as a standalone installer and a couple of hours later it’s done and now installs like a regular PC game. Yay! I love using Claud Cowork and dispatch for stuff like that. Most of the heavy lifting was done from a few prompts ob my phone and the odd Manual intervention. And now I have a standalone machine executable. Great stuff. Which in turn is inspiring me to finish the slightly souped up side version I’ve got that was never published at all.

@Nathan Campbell yep, I’d been putting off turning any of my web based games or apps into packaged apps ready for iOS or Google App Store submission as the general consensus was that it was pretty much a nightmare to do, would take days if not weeks and with a LOT of headaches. Using CoWork and Compute I was able to get my first app into Test Flight (compacted and ready for App Store submission if I wanted to do that) in a day of messing about on and off. By the time I then had the system learn from its mistakes and document the process I was able by the fourth attempt to get the entire process (starting with the GitHub of my web app and ending up with a packaged app ready for store submission) in under 30 minutes. And it’s repeatable and just needs a couple of small prods by me during the process. Which is great, because after 30+ years in IT I’m fully aware that I can apply myself to stuff like that and would get there in the end with some mocking and screaming but I really have no desire to be spending my time Learning command prompt stuff to achieve the end result. So Claude does it for me but I always have it document everything so if one day I do need to do it the old fashioned way I have the documentation to refer to. 😁 (Claude is so so good at documenting its own processes)

@Nathan Campbell hearing you say you had to force yourself to stop playing Neon Lander has actually made my day. My main aim with that game is to try and get it out to more people who would like it. As I’m well aware it’s very niche, but for people into that kind of old school arcade game I think they could get a ton of fun out of it. I’ve put in countless hours of play testing by myself and I still have fun as it’s one of those games where you can keep getting better and better in terms of mastering the controls (then you start chaining 360, 720 and 1080 bonus spins with near misses for extra points or challenging your own ghosts or the global ghosts in levels, or both at the same time!). It’s really designed to go on an arcade cabinet so now I’m beginning to get the hang of converting web apps to downloadable apps my next step is to make it a stand alone package that people can download and install rather than play from a website. I feel that’s what puts a lot of my target audience off, the fact it’s just running on a website, but if I can turn it into something that can be installed and run offline then I feel arcade cabinet owners (who already have PCs attached to their cabs) could get a lot of fun out of it. But I kind of lost steam with it when I release my beta as I couldn’t find anyone to play it, even in Facebook groups that I’m a member of such that cater to Arcade and retro gaming. It was like my posts on the whole were invisible. So I have a lot to learn about marketing but my current plan is to use Claude Cowork to essentially do a lot of the daily grind with regards to getting my stuff out there in an automated way so that I can focus on making the games while it does the marketing push for me. So when I’m ready to go again that’s going to be my approach. But next step is to make my stuff work standalone as a package. I’ve already managed to turn them into iOS packages running successfully in test flight ready for the Apple Store (not that I’m submitting these games as they are designed for PC), and I have that done in an automated way so hopefully it won’t be too tricky to do the same in terms of making it a setup file for PC installations. Thanks again for giving the game a try, it is VERY much appreciated. 🙏☺️

@Nathan Campbell indeed. I think a tutorial mode would help for sure but I have so many other things I’m working on right now. But I specifically put the abort button in to help people who are struggling with the controls but then there’s not much onboarding happening to show people how that works. I went hardcore old school 1980 arcade feel, but perhaps that’s a little harsh for most of today’s gamers lol

@Morgan Page I know, the detail in things like cassettes blew my mind. ChatGPT image generation in the latest model is nuts. It already has the entire context of my game so my prompt (which gave 4 amazing posters and 4 magazine covers) was literally “make me a video game magazine cover featuring my game and a poster advertising it, both from the 80s”. And those were some of the results with that simple prompt. Context is everything! What I’m trying to achieve with almost no time is a bit silly really and is really stretching my “speed coding” abilities. I have 27 different modes, each with their own engine and controls. Some of the modes can overlap into several other modes. And I can’t test anything on my phone as the game is designed for PC play so would take forever to test properly on my phone as it’s too tricky. So basically during the day whenever I have a minute, waiting for kettle to boil etc, I send over a ton of pretty complex and detailed requests into lovable. I might do somewhere between 15 and 20 during the day with absolutely ZERO testing. Then every evening I have about an hour of testing everything that I did that day and make notes as I go. Then in bed I try and think about what areas need improvement / any new ideas / and what would ultimately make this fun. Then the next day the cycle repeats. I make a ton of requests based on my notes and test that evening. It’s a crazy approach but literally all I have time for right now. And while this approach is unlikely to get me the winning game I think it’s powerful to show what is possible and it should give me a strong enough foundation to then give it the extra time and polish it needs after the contest has ended to really make it shine (when I’m not in such a rush that I can’t test and refine as I go)

@Morgan Page essentially for this I’m going 90% of the actual “coding” on my mobile as that’s the device I tend to have with me when I have idle time. And is one of the major reasons I have stayed with Lovable vs looking at Claude Code with cursor. So it’s just mostly Lovable in Safari and then the testing I can do on my laptop late at night once kids and dinner and other work stuff is out of the way. With my first big project, Lander, one criteria I gave myself from the off was to see if I could create the whole thing on mobile (even though they get platform to play it on was PC) and I probably managed to do about 95% of the creating just from my phone. What has proved incredibly helpful is being able to get Lovable to report and document the code. So for instance I now have 27 different modes that need sound effects, with some sounds shared between all modes but then lots of unique per mode sounds etc. At this point I have absolutely no idea of all of the potential sound triggers in the game, what is currently shared over what isn’t etc as it’s jus a huge task to begin to keep track of that. So was able to get Lovable to create new sound design document detailing the handful of sounds currently in the game as well as full descriptions of all of the possible triggers and events within all of the modes that would need sound adding. I was able to give that document to ChatGPT, have it analyze it and then come up with an implementation plan and draft descriptions (and in some cases raw values) for all the sound to then feed that all back into Lovable. It’s just taking in what would insanely massive manual tasks and completing it in a few prompts. Now is this going to create the perfect soundscape for the game? Absolutely not, not even close. But it is going to give me sound, which I need and don’t have time to work on too much ore deadline, without taking up too much of my time. And then when I do have the time later (probably after gamejam submission is in) I can go back to tweaking and replacing sounds, adding music and samples etc. but for now I just want something that doesn’t make it feel empty from an audio perspective. So as always my method normally uses a couple of tools but to do lots of sweeping changes in one hit and continue down my path of speed coding.

@Greg Dunn Star Trek I spent a fair amount of time with, Omega Race I’m going to have to look into tonight…thanks for sending me down that route (the name does ring a bell so we will see )