Tilki, the AI game creation platform, has raised $2.2 million in preseed funding to democratize the making of video games.
As the boundaries between traditional storytelling and interactive entertainment continue to blur, Tilki believes it can enable consumers to build video games without coding.
AI fund Twin Path Ventures led the rong, with participation from Atlas AI^VB, XTX Ventures, and SFC Capital.
Tilki empowers all creators—from imaginative children to accomplished novel writers and
professional game designers—to craft immersive and evolving game worlds without the
constraints of traditional coding. Tilki-created games are constructed in real-time and
designed around players’ choices.
Because every interaction and decision meaningfully impacts a players’ environment, Tilki’s AI-powered model can expand far beyond the potential of traditional pre-made game experiences.
Tilki’s approach represents a fundamental paradigm shift: moving from retrofitting AI into
existing game engines to building an AI-first creation and distribution platform from the ground up.
“AI is not just going to tweak the end of the development pipeline with NPC and asset
generation. AI is going to start the pipeline and run throughout, creating new video game
experiences that evolve dynamically around the player” Barn Cleave, CEO of Tilki.
Tikli’s team is concentrating on how to come up with a system to modify the creations that the AI comes up with, making it possible for devs to get the AI to execute on the vision just right.
In a Tilki game, non-player characters (NPCs) move freely throughout the world, make
autonomous decisions that carry meaningful consequences, and have persistent effects on the environment. This creates dynamic storytelling experiences rather than the static, pre-scripted interactions that have long characterised traditional gaming. For example, an
innkeeper NPC might gladly take payment for a night’s stay if that were all a player wanted.
However, (with the right persuasion) they could also ignite a rebellion, whispering the call to arms to nearby NPCs, who spread it from villager to villager until the entire town rises in fury.
The TILKI team of 17 people is a combination of video game developers, AI researchers and machine learning engineers who’ve built games for Atari, Sony, Microsoft and Nvidia.
“The strength of Tilki’s team was a key factor in our investment. The blend of experience
they bring – from gaming and AI to mathematics and product development – struck us as
unique, and we’re confident it will be instrumental in realising their vision,” Nick Slater,
partner at Twin Path Ventures, in a statement.
Members of Tilki’s senior team were also previously involved in creating a user-generated
content platform for 14 million players, in addition to founding a number of other AI startups.
The team are all avid D&D players and many are also published creative writers.
The funding will be used to build the company’s closed beta (which will launch in six months) and a subsequent open beta set to launch within one year. As part of this, Tilki plans to create an accessible marketplace where players can share strategies and
components, and collaborate on game elements.
“We see a future where anyone with a game idea can bring it to life, creating experiences that grow and evolve alongside the players,” said Cleave. “By democratising game creation,
we’re not just changing how games are made; we’re opening the door for entirely new voices to shape the medium. Through TILKI, we’re empowering the next generation of creators and helping transform the $176 billion global games market.”
As a company focused on empowering creators from all backgrounds, Tilki aligns current
and future generations of game makers behind the latest AI innovations. Tilki’s goal is to
transform a $176 billion global games market by opening game creation to entirely new
voices.
While Roblox is using a lot of AI to enable players to build their own games more easily, Cleave believes there are still big challenges there, as the game makers still have to learn the Lua programming language. That means perhaps the top 1% of game developers on Roblox really know what they’re doing.
“What you really want to do is have the game designer have full more control so they can build more interesting things,” Cleave said.
Origins
Cleave got into learning about AI and games about 18 months ago, when he was approached by someone who was building agents to have difficult conversations with people. Those applications were initially built for HR purposes. Cleave also took note of the smart non-player characters that AI companies were trying to create for games, and he noted how it wasn’t really going that well just integrating ChatGPT into games.
Cleave got more excited by the idea of having AI create a world and also create agents within the world who could do more, and have the free will to create more interesting and complex stories. He thought that giving agents back stories, goals and expectations was a better way to give them the freedom to do more interesting things.
Cleave said it is depressing to see much of the motivation for AI in game development to be how to replace humans in the development pipeline.
“I think, personally, that’s the wrong question. It’s what happens if you have a human building a game with AI? You start there. How does that go through the pipeline to build more interesting games? And I think that’s a far more interesting story because you get into this idea of Dungeons and Dragons, where you have someone who you’re with having the game conversation, and then they can understand what you’re trying to do and what your motivation is.”
The team has been building the AI tools for about 18 months. The focus has been more narrative based content. Cleave said he doesn’t like it when the player doesn’t have much control. Expedition 33 has a beautiful story, but then you only get two choices at the end.
For today’s gamers, that’s not enough, Cleave believes.
“It’s like a puzzle in a video game, where the developer creates a sandbox and there are multiple solutions to the puzzle,” he said. “Someone has a good, clever way of solving it but that you can’t manipulate the world in that way. And I think that’s the exciting thing of seeing if you have an AI that’s working with the player. It’s the same way with Dungeons & Dragons. You find an acceptable way of dealing with the situation.”
“That’s where I come from, and it’s maybe from this background of having having built user generated content in the past, specifically, more like puzzle games, where one of the things we found was, when you were a puzzle designer, there are just a limited number ways of ways that someone can solve it. You desperately try to think of three or four other ways, and you’re always boxed into that corner.
“There’s a freedom in always allowing the player to build something. And that’s something actually, I’ve been doing for a while,” Cleave said.
“The kind of narrative you want to be able to sit down and create is like going back to Westworld, the first good season of Westworld, right? I would love to be able to sit down and interact with an AI and create a narrative in real time and have it unfold in front of me, and then be able to deploy it to users, where it will be able to appropriately expand and and fulfill their needs,” Cleave said.
Supplementing the pro narrative with additional content
I can see where Tilki could get involved with triple-A game companies and get them to allow it to use their intellectual property in the service of satisfying hardcore fans. As an example, Hangar 13 worked a lot of years to come up with a brand new Mafia game. Yet the title only had about 10 hours of gameplay, and so the company had to sell it for $50, less than a premium price. But what would the value be if the fans were able to create their own content and make the gameplay hours unlimited?
“It makes it far more interesting. And that’s why narrative games have gotten into this trap where, as you say, they only do 10 hours, and everyone will go, ‘It’s not really worth the money, because I can play a shoot-em up over and over again.”
Cleave foresees that someone who has an IP will come to Tilki to make a game based on their IP.
“Then we’ll do the model where they’ll bring on some of the marketing and say, right, here is our IP in this game, and you will use that partner to help to market that product, and we’ll end up building the game because their core competency is not making video games
interesting,” he said.
“If you think of a novelist who has decades of work, but that author has been told that it’s far too expensive to make a video game based on any of their IP, then we can take that in and have the team actually build that out,” Cleave said.