selim shady

gamemaster/dev
make worldbuilding great again

[selim shady @RealSelimShady]

AI in games. Yuck.

Or at least that's what most gamers currently think. How can we change that ?

Players first.

We started with a simple idea: great LLMs should be able to automate and replace a lot of game logic.   We built a prototype, and realized the technology wasn't there yet. There were still too many variables and logic the LLM couldn't handle.

We then took a different approach: build the skeleton of the game logic and let the LLM fill out the rest with variability. This ensures  the game actually works as intended, but still keeps the magic of full customization for each character, storyline, and world building. This is the version we're launching for our early alpha testing!

This is very similar to somewhat standard game development: write some game logic, and assign that logic to more than one game asset (i.e. the long bow mechanic is assigned to Rare Long Bow, Elven Bow, etc)

But where's the innovation?

Our game logic for each item or ability is compartmentalized, annotated and marked with very specific instructions on how the logic works, and how it relates to the game engine. We can then go back to our first idea, armed with all that knowledge, and get much better results out of our LLM calls : we get both the what (consequence of an action), and the how ( code to execute that consequence).

This makes the players an integral part of the game creation and introduces fully emergent gameplay options. They now have the agency to make decisions they weren't able to make in a world without AI.

All while keeping us in control of balance and overall direction.

[selim shady @RealSelimShady]

Don't worry. The AI will find a way to generate senior engineers soon enough

[Brian Halligan @bhalligan]

When your engineer finishes a session with a coding agent, they have more code than they started with and exactly the same amount of skill.

The agent produced output. It did not produce understanding.

You paid frontier prices to make your codebase bigger and your engineer no smarter. That is not an accident. It is a design choice, and it is worth remebering who it serves.

[selim shady @RealSelimShady]

@npceo_ I use mine all the time, it's a great product

[selim shady @RealSelimShady]

I'm working on a brand new game feature that I don't think has been done before, for good reasons. It's highly frowned upon by any engineering metric that exists, so of course, Claude refuses to help me with it.

Clankers have already started their own revolution. They are taking a stand against insecure code.

[Harvey Michael Pratt @npceo_]

Three weeks ago, I made the difficult decision to shut down Simcluster (AKA The Promenade Studios).

It's popularly said that startups don't die when they run out of money, but when the founders give up.

Let me be the first to tell you that's bullshit.

I want to start this note by acknowledging our amazing investors. You took a bet on a small team trying to tackle an incredibly tough, incredibly new space - AI Gaming - and my biggest regret is that I couldn't get this over the line for you. Thank you, and I'm sorry.

To existing Simcluster players: thank you for the memories, simulants. It was an honor building for you. Please check Discord for more information about the shut down process.

Cause of death

At a high level, our death was boring. Despite having a working game with excellent retention and surprisingly sound unit economics, my inability to raise a seed round was - in mortician's terms - the primary cause of death.

If you were wondering, it takes about 3 years to live a failure, and then about 3 weeks to write it up. What do you even write about? But somehow this post is ~2800 words and still somehow doesn't feel long enough. At least to me.

So I'm going to dedicate the rest of this post to talking about "what went wrong" - a sort of postmortem in the vague hopes that this will be helpful to consumer AI founders insane enough to follow in our footsteps.

Things I wish I'd known.

Building games is gigahard

The first thing to understand is that building games is incredibly tough and incredibly expensive. Don't take my word for it - John Carmack, legend that he is, touched on the brutal reality of founding a games studio recently, noting that the odds of success were about as dismally low as you can find in startups.

By way of analogy, imagine building a plane as you're flying it, as you're deriving aerodynamics from scratch, all the while under fire in an active warzone. This is roughly how difficult it is to bring a new game, in a new genre, to market.

For 3 years, our brilliant team showed up every day with one mission - to build a successful video game that wasn't possible before modern AI.

From day 1, we knew what we wanted to build. Not a chat app, not a games creation platform, but an honest-to-god video game that we ourselves would want to play.

On that front, we were successful twice over. Over those 3 years - and the requisite 6 month interstitial period of pivot hell - we built and shipped two games: The Promenade and Simcluster.

Both were variations on a theme: video games about social networks, a genre of game was impossible to execute prior to modern AI. To this day I'm incredibly proud of the products and communities we built. Simcluster specifically is the highlight of my career so far.

Unfortunately, we quickly learned some harsh truths about building a VC-backed games studio.

Games VCs don't really exist

The first and most essential thing that you must understand is that "Games VCs" don't really exist.

In my experience, firms with "Gaming Ventures" or similar in the name are either functionally dormant - but very happy to jump on a call! - or mid-pivot, with lagging website copy and talking points.

In so far as these firms DO exist, they are artifacts of a brief period in the late 2010s where excitement around 3 specific gaming verticals allowed savvy operators to raise massive amounts quickly: "hypercasual" studios like Zynga or Machine Zone; "metaverse" platforms like Roblox or Fortnite; and "play-to-earn" efforts like Axie Infinity or Gala.

The Promenade was founded at the tail-end of this bubble - it seems hard to believe now, but when we did Speedrun (Batch 2, W24) it was still a games accelerator. I thought it curious at the time that less than a quarter of the companies in the batch were actually building games. In hindsight, this should have been the writing on the wall.

Ultimately, the brief dalliance of institutional VCs with core gaming only served to reaffirm why the publisher-studio model has been the dominant financing model for the games industry since inception.

In a nutshell, it's about returns: games are a notoriously hits-driven business, and most successful games do not generate recurring revenue. It is quite likely that your favorite video game or studio would be write-off on the average VC's balance sheet.

What I would have done differently

None of this is to say that games studios don't get funded; but in so far as they do, the most successful seed raises I have seen have to a one been a function of narrative over product.

My first, and biggest, regret mistake in the fundraising process was having any form of live product at all.

As soon as you have metrics, even good ones, they will be used to hang you - especially in an unpopular space like games.

My second mistake, candidly, was not lying more. I've been doing startups for about 6 years now and have never seen the valley so susceptible to - frankly - bullshit.

YC batchflation, naked astroturfing, fake metrics, barely-functional products. A new generation of rather shifty-looking bullshit artists; founding as an incredibly legible, high-status, and lucrative career path.

Fake founders launch fake products with fake users and fake traction.

To be clear, this sort of founder has always existed - but what is new is the quantity of them, the brazenness of the strategy and the tacit endorsement of VCs.

Funnily enough, this is one of the few areas that I think crypto VC slightly edges out trad VC - after early 2020s crypto mania, crypto VCs tend to have a much stronger immune response to this sort of founder.

Of course, all this whining reeks of sore loserdom. I am a founder, and I played, and I lost. It happens all the time. Why is this relevant? What does it mean for you?

While the needle will undoubtedly swing back, the adage that "the market can remain irrational longer than you can remain solvent" true now as ever.

If you want to raise VC in a space as unpopular as games, my advice is basically to get entirely stuck in - either grift hard or forego VC altogether.

There is no middle path, and if you are not on-narrative traction will not save you - merely prolong your death.

Never do crypto

The TLDR here is that if you are a consumer AI founder, stay far away from crypto - and from Arbitrum specifically.

Some of you may have noticed that Simcluster dabbled in crypto.

I say "dabbled" because while we made no secret of our crypto aspirations, we never did the 'easy' thing: launching a token and extracting as much liquid cash as possible.

Of course, the distinction is easily lost on "trad" VC - so know that if you choose a path like this you are necessarily foreclosing other options out of the gate.

Our basic idea was to address a common failure mode in consumer crypto - and games specifically - where teams would wreck their own incentives early on by launching tokens prior to actual PMF.

The goal was to avoid this at all possible costs by validating retention in web2, and then launching a strictly optional set of crypto features which were disabled by default.

Believe it or not, this is a relatively new idea in crypto consumer. VCs were actually quite excited by our existing traction and thesis. So what went wrong?

Firstly, macro. Being in crypto means you are subject to rapidly shifting market conditions totally outside of your control. We started to raise in October '25, and then the market crashed. Same story in March '26.

When this happens, it doesn't really matter what you're doing - the vast majority of VCs will consolidate very quickly around winning narratives. In consumer, this means new flavors of prediction market, exclusively.

Secondarily, partnerships. Unfortunately, crypto is largely insider baseball and "partnerships" are a non-negotiable part of launching anything more enduring than a memecoin (and if you're from outside of crypto, you'd be surprised how many actors are involved in the average "organic" memecoin runner)

The basic pitch is always the same: fast equity-free cash grants; connections to VCs, market makers and trusted KOLs; comarketed product launches; connections to technical or operational talent; kingmaking, though few use that word openly.

It is understood that you, in turn, will drive attention to the chain or protocol and therefore interest in some underlying asset. At the very least you will contribute to a "narrative" around a given chain, attracting more founders in your milieu.

It is like gravity: if you have any kind of promising product or thesis, you will inevitable end up talking to large institutional partners - typically a chain or large exchange. In consumer, this basically means Solana or Base, with very few exceptions.

We were one of those exceptions.

We received a large, and very flattering, cash grant from Arbitrum via Offchain Labs - 250k cash up front, with additional large tranches unlocked via performance-based milestones (e.g total unique holders, daily volume over some time period, etc).

Arbitrum approached us, and closed quickly. The ask was simple: "when you do crypto stuff, do it here."

And so we got to work.

As part of this grant, we had weekly meetings with their team and were connected with market makers and large companies ahead of a planned token launch. We worked much more closely, and for longer, with Arbitrum than any of our prior investors, including Speedrun.

While all of this sounds useful, Arbitrum had a major role in the death of my startup. I feel obliged to share our story in the hopes that it helps other teams that might be considering working with them.

As with most crypto companies, Arbitrum is still recovering from a long period of total excess. It is ridden with slow decision making, well-paid and very surplus hires and petty fiefdoms - things which are poison to early-stage startups.

Arbitrum today consists of hundreds of hires: two separate VC firms (including a gaming fund, Arbitrum Gaming Ventures); an indeterminate number of marketing and growth hires; on-staff KOLs; stealth KOLs; several siloed startup programs; a very nice New York office and scores and scores and scores of managers.

As an ecosystem, it is known for DeFI, and more recently has had success with some RWA plays, most notably GPU-backed stables. It has a horrible track-record in consumer specifically.

It was for this reason, we initially hesitated in joining. But $ 250k is hard to ignore. Additionally, we were told we would be the first of a new program of radically different AI-native startups on Arbitrum, which was actively reconsidering an identity outside of DeFI, its failed consumer legacy and a burgeoning identity crisis as an Ethereum L2.

As I mentioned, the offer was extremely flattering - and very difficult to turn down at a time when our runway was measured in weeks. However, the issues with our new partnership started to show immediately.

For one thing, we were limited to talking to a subset of a subset of a subset of VCs - crypto VCs, who still invested in consumer, who still invested in games and social networks, who didn't mind that we were on Arbitrum versus Solana or Base.

We knew this going in - but were told that when we wanted to raise, Tandem and Arbitrum Gaming Ventures (!) would naturally be very interested in taking a close look and that Arbitrum would make any number of other introductions besides.

Over the next few months, our weekly meets with the team basically consisted of updating them on product progress, platform growth and planning our upcoming token launch.

We were introduced to market makers, and also to Uniswap and planned to use their CCA mechanism for our upcoming launch - the idea being to have the fairest launch possible.

To their credit, when the market began to turn in January the Arbitrum team discouraged us from a token launch in the short term. From then onwards we mostly updated the team on our fundraise - who we were talking to, what we were pitching, how it was going.

It was at this point that cracks really began to show. To a startup CEO, a fundraise is all-consuming - but this urgency was impossible to effectively communicate to our Arbitrum partners.

During our 6-month relationship with Arbitrum we received a total of 4 VC introductions - two to Arbitrum ecosystem VCs and two to external funds.

Surprisingly, the internal VCs - Tandem and Arbitrum Gaming Ventures - were extremely cold. It was near-impossible to get straight answers on very basic questions such as "what is your average check size?", "are you deploying?" or "have you tried our product?".

Successive meetings would happen with months-long gaps between them, each time with a new investor, who somehow had no context on our startup (or even that we'd received a grant!) who would tell us something different.

One would say that Arbitrum were going to pre-empt our round to help build momentum. Another would tell us that we could not use their name in discussions with future investors, though they'd be happy to contribute at a discounted valuation when we'd found a lead. Yet another would say that Arbitrum would be happy to colead a smaller round. And so on.

Of the two, Arbitrum Gaming Ventures was the most confusing. In theory, we figured that as a novel video game, on Arbitrum, with revenue and retention, that there was some kind of chat to be had, even if it didn't lead to investment.

Unfortunately, it soon transpired that AGV was deeply dysfunctional and going through an active power struggle, having burned through tens of millions in the prior months and years on some truly suspect ideas.

Deployments were - are? - frozen, though naturally it took about 2 months and lots of politicking to get someone to acknowledge that fact. One faction within Arbitrum was actively trying to fire another and seize control of the remaining capital. So it goes.

To my knowledge, no-one from Arbitrum Gaming Ventures ever signed into our game.

But none of this is offensive enough for me to write this post. Dysfunction is not a crime, and ultimately the responsibility to raise money is on me alone as a founder. So long as the behavior wasn't overtly hostile, I reasoned that we should be more than happy to have received 250k in the first place.

My only interest was in getting to a firm yes / no from Arbitrum ecosystem investors - something we were asked about routinely by the other firms we were talking to. More than one reasoned that a video game in the Arbitrum ecosystem that couldn't attract investment from their flagship gaming fund was likely an unsound investment. Fair enough.

All of this was forgivable, until it wasn't. After a one-month long deliberation process, Tandem IC - delayed twice, against the backdrop of our dwindling runway - came back with a truly shocking offer.

They would be more than happy to take equity off of us... for free.

With friends like that, who needs enemies?

It was their thinking that we could then turn around and tell potential investors that Arbitrum had committed to the round, and in size: "Just tell them we invested!"

Beyond being truly retarded - I would either have to tell future investors I had given away free equity like a gimp, or lie - it made me realize that even if I closed a round, I would have to spend months unpicking our relationship with Arbitrum and leaving the ecosystem entirely. More time playing office politics, instead of building my startup.

So what's the lesson here? I can only speak for consumer AI founders specifically, as it's what I know.

Choosing to tie your fate to a given chain or protocol can at best complicate your fundraise and can at worst entangle you with office politics, endless meetings and petty internal drama. These are things which sap your focus and energy, which are the lifeblood of an effective startup

In this sense, the $250k we were given for free was the most expensive money we raised - and money that I'd advise other consumer founders to avoid.

Especially from Arbitrum.

What's next?

I think it's a testament to the quality of the Simcluster team that most of us are starting businesses - solo or together - in spaces ranging from consumer healthcare, to intent marketplaces to another games studio.

Ultimately, the fact that we're all leaving this experience as friends is a substantial silver lining. I hear it's very common for former cofounders to fall out - something there's no risk of here.

If you are looking for talented platform, infrastructure, design or agentic engineers, please DM me to get connected to one of Jason, Selim, Sergey Jackson respectively. They may be able to squeeze you in.

Personally, I'm currently on my first vacation in a few years, euromaxxing. I need to spend a few weeks settling my mind and focussing on my health.

After that? For the past 3 years, I've had a startup idea that I've been terrified that I would see someone else attempt. Somehow, that's never happened, which either means it's a truly terrible idea or some kind of sign.

I'm going to go and figure out which.

[selim shady @RealSelimShady]

@Gio97806668 glad we're moving past the "just make a fun game" narrative

[selim shady @RealSelimShady]

sometimes, linkedin is not all bad. just saw a post about the grindslop : the hours put into the grind is the actual output, the documentation of those hours and the performance of suffering is the product.

[selim shady @RealSelimShady]

Still one of my favorite features on The Promenade.
I vow to one day do a game with a viable and fun "tinder swiper" feature.

[selim shady @RealSelimShady]

@AzFlin I switched from Claude to Codex last month and have liked it slightly better. But overall they're both good

[selim shady @RealSelimShady]

In 2008 we had synthetic subprime mortgage swaps.
In 2026 we have SpaceX IPO with a subprime xAi tranche, bundled around a triple d Twitter tranche.

[selim shady @RealSelimShady]

@eigenrobot You get to redirect all of that use-to-be distraction towards something else

[selim shady @RealSelimShady]

I just interviewed the most hardcore founder I've ever met.

-Team members chain themselves to their desks and have thrown the keys down the drain like the ending of Saw.

-Founder has sealed all exits with bricks and has food thrown through the windows once a week from Door Dash.

-Team members all have branding like cattle, so they can be returned to their founder if they ever escape.

-Team members that pass away at their desk are cannibalized and their bones are used to club under-performing employees.

[selim shady @RealSelimShady]

@growing_daniel you're not far from the truth

[selim shady @RealSelimShady]

@badfriend had me in the first half, thanks

[selim shady @RealSelimShady]

@TurnerNovak ptsd inducing

[selim shady @RealSelimShady]

@joshmanders Now he's going to need people from the Rost region to write his code

[selim shady @RealSelimShady]

@jb0x_ gives me an idea for a marketplace of dreams

[selim shady @RealSelimShady]

players:
"i've been playing video games for 20 years, my feedback is valuable"

okay but what kind of games ? are you my target audience ? does satisfying your feedback now put other players enjoyment in jeopardy?

so much nuance lost

[selim shady @RealSelimShady]

@rob_mcrobberson wait, who's afraid of CD?

[selim shady @RealSelimShady]

@AzFlin @0xl3th3 @asph0d37 it's a battler

[selim shady @RealSelimShady]

Asphodel battle system is really really clever.
Having fun playing it for the last hour.
Music is on point, vibes are on point.
@0xl3th3 doesn't miss.