I Built a Game in 25 Days with AI
40sThe promise of creating a game in less than a month with AI is provocative and taps into the creator economy's fascination with AI tools.
▶ Play Clip
"Delivers exactly what the title promises — a complete breakdown of a 25-day AI build process with tangible results."
A passionate indie developer details the creation of their first complete game, a hardcore RTS, entirely with the help of AI tools over just 25 days. The video walks through everything from the initial concept and tool selection to asset creation, animation, and final gameplay testing, offering a candid look at both the successes and challenges of an AI-driven workflow.
The developer initially chose Unity for its code-first approach but switched to Unreal Engine 5 for its superior visual capabilities. This decision was crucial for achieving the desired look and feel.
The AI tools used were identified, including Claude for coding and development guidance, and multiple APIs for a range of tasks. The approach was to leverage AI wherever possible to speed up the process.
AI was used to generate 3D models, with tools like Tripo Smart UV being highlighted as a significant time-saver. The developer also emphasized that manual tweaking and supervision were still necessary for quality.
The developer discussed using AI for animation, including retargeting and mixing animations, and noted that while some attempts failed, the final results were robust. The process involved testing and refining.
The game includes 18 units, 11 castles, and 59 generated models. The developer also detailed their workflow, which involved creating concept art, documenting extensively, and managing a large number of commits.
The financial aspect of the project was estimated at roughly $500, accounting for AI tool subscriptions. The developer also detailed the number of iterations and experiments that were performed throughout the project.
The video concludes with a playable demo, featuring an in-depth look at the gameplay mechanics, unit strategies, and the developer's thought process during a live playthrough. The developer also encourages viewers to build their own dream games.
The game is called 'Skeletons', featuring a hardcore RTS experience where players have a limited time to build armies and capture enemy castles before the game ends, adding a unique strategic challenge.
Unity to Unreal: The Engine Switch
Illustrates the critical decision to prioritize visual quality over familiarity, a key trade-off in game dev.
00:03:52Project Scale in Numbers
Quantifies the massive output of a solo AI-assisted project: 18 units, 11 castles, 59 models.
00:33:16Tripo Smart UV Saves Time
A specific tool is identified as a major time-saver in the 3D pipeline, a key bottleneck.
00:11:14Claude: A Gift from Heaven
Highlights the transformative impact of a specific AI on the coding process.
00:35:01AI as an Enabler, Not Just a Tool
Articulates the philosophy that AI empowers creators to focus on design, not manual labor.
00:12:27[00:00] my own dream game one month ago stock assets.
[00:13] made with my AI day of my. faced and how I.
[00:26] is ready to play. levels, 10 races. So let me tell
[00:38] Really want to thank my game every day On my plate but
[00:50] your support really gave Now since it's been working days without calculating.
[01:03] say months of dedicated where I started. All the way before
[01:17] decided I want to that Unity was more. code first and that
[01:29] I would be able but it potentially would a lot of tests
[01:42] sort of takes more was a crucial step right. Where I carefully put
[01:56] a database for my again like they all Like this First information
[02:10] For example you can There is a skill Later I even used.
[02:23] them here in HTML To basically put them tune it by myself.
[02:36] created my character or It was very. this game mechanics test
[02:49] like I'm gonna build. like almost first three Primitives where some figures
[03:02] but with more focus if I can do because units are the
[03:14] how they attack how I wanted hardcore RTS. that I started to
[03:26] 3D former 3D engineer to do anything like especially. can do.
[03:40] because to. stuff etc before this It's done.
[03:52] to Unreal Engine 5 So I spent whole tools to see how it.
[04:06] Claude was extremely helpful. Multiple APIs such as my agent has access to.
[04:20] right now also it's The terrain I was to be like this
[04:33] I figure out that and it works. level almost completely myself
[04:46] Created trees I put that I first want I can tell later
[05:00] and I definitely didn't. except putting details that's I wanted later to
[05:12] I always say you like I probably will can say I need
[05:24] is not just level play etc. and by the way
[05:37] I really recommend you. and it's still top And smart UV is
[05:50] quality and I know top 3d AI where so definitely try it
[06:02] support me and get and then I. that that exact time
[06:15] of the biggest fear CGI for almost 11. in animations especially if
[06:29] Etc and my game different units like a you have to hire
[06:42] fit into the game. astra and I see This test whenever new
[06:56] tests related to 3d to be honest like building my game.
[07:08] try to do this Capabilities to understand the legs are etc and
[07:22] able to do that The model and then course it was a
[07:35] thinking about that and AI and my thought references from sight that
[07:48] in the. like here also you Right because attack animation
[08:01] a bow animation you like on multiple different I decided.
[08:14] references and then ask because I tested it I think and it.
[08:28] something that you can astra and dang it. 5 .5 and probably
[08:40] Fable 5 .5 the on my channel I on my channel maybe
[08:52] past now that moment wolf a scorpion and That was really scaring
[09:07] bad in that and if you want I.
[09:19] the thing for me With claude with the music and voice.
[09:34] voice line Gemini 3 Well for sound effects effects decimate was one
[09:47] different I tried moreca I also tried google. ended up using a
[10:00] Was definitely not one vfxs with all sounds quite a while I
[10:13] like different one trying out and later it the way guys all these.
[10:25] like right now this right how I. my workspace which is
[10:37] Up this easy this develop games all that Mark workspace and I
[10:49] is that for characters of Tripo smart UV significant step because look in.
[11:02] have sight right so The hero side like red color so I
[11:14] Coach UV define that a lot of time Updated their smart mesh
[11:26] actually made a video game and that's absolutely saved me crazy amount
[11:38] I saw and again network of professional 3d artists. and developing and improving
[11:50] Also developed with my years of experience including date how you can
[12:03] with combination with minimum knowledge. right now you can full control.
[12:15] know like all depths Process it how you bit right now for
[12:27] With set of animations day with all these Than 200 happy students
[12:39] the craziest thing that craziest for me at like I used.
[12:51] took me like literally I had in mind right I never did
[13:04] Come up from the now I create all and then design the
[13:17] them and. day I made all Was also ridiculously easy
[13:30] addressed vfx's right. first do some images Also for timings like
[13:44] simple html illustration but see like okay yeah then when you're gonna do.
[13:57] all these effects first Just have a right etc is just faster
[14:10] Vfx's I just gain Artist like they probably to describe all that link.
[14:26] alien language like I'm Knowing like all this just like went to
[14:39] this is what I send the name to this but mix it
[14:51] me after a day. in the game and There is an end
[15:04] wow and here again already mature and everything this crazy guy I cannot.
[15:17] things works I mean I mean six months longer and harder and
[15:30] I have a video say like it was like retarget animation like
[15:42] when from one skeleton. like different bones you It was failing like
[15:55] within like few hours Broken arms but now I also use like
[16:08] up like camo, reallusion part moves like. animation mix it up
[16:20] Now slow it a here and unable to becoming that you need
[16:34] exactly you want so AI can do anything As I also said
[16:47] a level also with Already understood everything the started designing it like
[16:59] helped me to put good but of. good and my design
[17:11] easily adjust. two I just went I want to fix
[17:24] it's totally fine so 18 units including two Available in tavern for
[17:38] coolest thing about this like a pile of I'm here actually right
[17:50] developed that in less helped of course is that. added like a tavern
[18:03] And one of them elf I also posted probably and day 22
[18:17] pretty much spent not building. own game looking around For example I added
[18:30] mind so they give yeah I actually added gold you can upgrade.
[18:43] in tavern well my So we're easy creatures That's why it's called
[18:58] game you start to Starts not to see some friends to play
[19:10] what's going on in clear what's going to happen. some time and this
[19:23] Realities making it playable to play it's completely demo on steam by
[19:36] I faced some issues like. all the documents but is that it only
[19:50] why I'm so unlucky I have to drop most likely it'll be.
[20:03] that happens when this the description that will be. put a lot of
[20:15] This game for years when I was like Was posting on some
[20:27] with this idea I you have like different was trying to find
[20:39] about game development but And I had to you some numbers all
[20:52] Play it. it guys here we
[21:07] go okay.
[21:21] that was basically the Next who knows they each castle produce some.
[21:35] can produce up to So we have one now with this unit
[21:48] because they're gonna come Strong okay so here's everything around like it
[22:02] want that's what I more units. hardcore rts and if
[22:15] Problems oh by the that will give me by you know like
[22:29] a world. but it was something Like Rome total war
[22:42] the middle but I old times you know Like another camp but
[22:56] to attack this camp okay I need to to aggro.
[23:09] go and I need Because skeletons oh they keep the distance between
[23:22] No they killed my guy. Oh like I I talked. Too long.
[23:38] I lost well I me try again the be harder and.
[23:52] go attacked right away To focus one unit otherwise it will be
[24:05] show you that there this one is a. I didn't attack you
[24:18] restart again. by design and even You let me show
[24:31] and since we are but if I attack that is not.
[24:44] there is a bribe Them and it actually is very valuable and.
[24:59] most likely they will Careful. It has a take
[25:13] Okay let's go now trees was one. they are very dense
[25:27] But there might be you and that will gonna need to look
[25:40] just killed like few wolves. this starts into regenerating Confidently now my plan
[25:53] aggro guys with my try to attack the out focus focus guys guys.
[26:06] to the castle they Guys please. Okay we are no
[26:23] Losing okay or maybe. win it's possible.
[26:36] to not lose time was too close to. having a buff okay
[26:48] Try to aggro like come here boys okay now this one okay
[27:02] What do we do strong I just win Attempt number five man
[27:18] this is the hardest thing. boys yes we got Oh why why it
[27:32] actually keep okay and House so the thing lose any moments they
[27:44] moving and you lose you see the guys he. it is balanced trust
[27:57] Sometimes hard to do have captured it these There why they attacking me.
[28:09] we're gonna need skeletons choose to upgrade some. no I think I'm
[28:21] Fight what are my here will be archer It out should they
[28:34] units cannot come to. chances guys they are With humans alone most
[28:47] you're gonna lose. Okay okay finally now Have skeleton archers and
[29:02] now let's collect them all. boys actually soon there because I built this
[29:16] a little bit well time is very precious army let me check
[29:28] I can find some In this location not wolves and nothing else okay.
[29:43] of them why okay of these. they will not aggro
[29:56] Then skeletons gonna go no they attack look yeah but I mean
[30:09] Like anyway crazy guys castle let's capture it. Really strong like look
[30:24] in his back he Nice okay should we which units you cannot hire.
[30:37] time like. go say hi to they're gonna go here
[30:51] army and I want is how cruel this Let's go this big
[31:05] now we're gonna go can win them by and it's making my
[31:17] guys by the way Look what he does like oh just one
[31:31] here's the thing about okay here. try to kill them.
[31:44] Can we okay. And it's gone what
[31:56] Do nothing what was a few seconds like. there is a way
[32:09] Camps cannot be right like you cannot go them because you have
[32:23] definitely have to figure. this army and how For those who are
[32:35] tester for my game Me in my discord finally let's do the
[32:47] nice page so what documented but again. I was making posts
[33:00] say like I. kind of units 11 castles. 59 generated models it
[33:16] in parts of course five or six I Tripo 247 animation clips
[33:30] Experiments with the video effects 1380 concept. was doing concepts like
[33:44] Right now who gonna 135 000 words of Very very important and
[33:57] because I work with this much commits here I created up to date.
[34:10] again each I spent The interesting charts that faster in fact and
[34:22] Wild wolf about a first I tried this. wolf with the fable
[34:35] Like sick maybe I'm GPT -6 Astra so wolf you know took
[34:48] faster when I actually started it was Claude Was gift from heaven
[35:01] issue and I also I wasn't impressed like 5 .5 when it
[35:13] 5 .5 to see can do any crazier it is for tokens
[35:26] crazy numbers like 30 billions. I think I figure Was not tracking precisely
[35:39] again like when you don't really spend a never reached weekly limits.
[35:52] busy like thinking like If you just expected like it will cook
[36:05] here and therefore I it somewhere like explaining rereading the project it
[36:17] little bit useless one now. have any subscription and Let's say so the
[36:31] of course it's much as like almost 10 so maybe like 20 000.
[36:44] max subscription Claude's and Anyway for my other etc so yeah let's
[36:57] most let's say only more rest precise I little bit inaccurate how
[37:10] by promotions and. lot of promotions and Direct use I would
[37:22] 500 bucks and here there is micro splats helps to work terrain
[37:35] I had to buy of should be a basic. I also bought one's
[37:49] actually a pack for my horses like built. like this before I
[38:02] Do that retargeting after changes to this animation motion from real illusion
[38:15] is the complete list. know in the comments Build your own dream
[38:28] subscribe and I see you in the next videos.
The creator immediately establishes credibility by showcasing the finished game and the impressive 25-day timeline, creating automatic interest and a 'how did they do it' curiosity gap. This direct display of results is an authority flex that promises valuable, actionable insights.
I built [PROJECT] in [NUMBER] days using [OPEN-SOURCE TOOLS]. Here's my complete workflow, including the [PROBLEM] that worked, the [OBSTACLE] I hit, and how you can do it too.
AI-generated clip ideas for Shorts based on the transcript
The promise of creating a game in less than a month with AI is provocative and taps into the creator economy's fascination with AI tools.
▶ Play ClipShowcasing the struggles and failures of AI tools in game development is relatable and provides a humorous behind-the-scenes look that engages viewers.
▶ Play ClipThe creator losing to their own AI-built game is both humbling and entertaining, creating a moment of vulnerability that viewers love to share.
▶ Play ClipRevealing the massive scale of AI-generated content in a short timeframe demonstrates the power of AI and sparks curiosity about the workflow.
▶ Play ClipBreaking down the actual costs of AI tools is useful and surprising, encouraging viewers to consider the economics of AI-assisted creation.
▶ Play ClipSnapshot 2026-11-18
Every observation TubeSum recorded, plotted as measured.
The vertical axis starts at 113K, not zero, so later observations stay readable.
Based on 13 observations · last observed Nov 18, 2026 · solid lines connect consecutive measurements (≤2 days apart); dashed lines span days TubeSum did not measure. · vertical axis focused on the observed range (starts at 113K)
Based on 13 observations · last observed Nov 18, 2026 · solid lines connect consecutive measurements (≤2 days apart); dashed lines span days TubeSum did not measure. · vertical axis from zero
Exact daily observations; gaps mean no observation was taken that day.
| Date | Views | Avg views / hour | Channel median |
|---|---|---|---|
| 2026-11-18 | 155,789 | 388.5 | — |
| 2026-11-17 | 154,059 | 408.6 | — |
| 2026-11-16 | 153,394 | 416.8 | — |
| 2026-11-15 | 150,870 | 446.4 | — |
| 2026-11-14 | 147,327 | 483.0 | — |
| 2026-11-13 | 145,559 | 514.3 | — |
| 2026-11-12 | 142,724 | 557.5 | — |
| 2026-11-10 | 136,785 | 657.6 | — |
| 2026-11-09 | 133,161 | 719.8 | — |
| 2026-11-08 | 130,566 | 768.0 | — |
| 2026-11-07 | 126,685 | 861.8 | — |
| 2026-11-06 | 120,542 | 1,076.3 | — |
| 2026-11-05 | 115,466 | 1,312.1 | — |
Audience reaction — separate from the performance metrics above.