Intelligent artificial NPCs

The organic matter DeFi has been searching for

Accessor Labs’s mission is to restore the magic of open networks. To that end, we apply novel cryptography so that game creators can populate on-chain virtual economies with intelligent artificial agents.

Multiplayer games and virtual economies are inherently more powerful when they are trustless, composable and built on native financial primitives[1].

In the future, game creators will be free to innovate on top of existing game ecosystems[2], players will benefit from richer, community-led virtual experiences[3] and financiers will find a stable foundation for DeFi by capitalising the organic matter of goal-oriented gameplay[4]. Games will pioneer web3’s flourishing.

To achieve this vision, we need novel computational primitives. Blockchains are the natural foundation on which to build composable, trustless virtual worlds, except for what we all already know - they are inherently slow and at odds with privacy[5]. This is why zero-knowledge proofs that trustlessly attest to fast, offchain, private compute are so important[6]. However, gaps remain in this burgeoning industry.

In particular, it’s really hard to bring artificial intelligence on-chain[7]. Machine learning is a precondition to so much of modern gaming, whether it’s the creation of engaging NPCs, procedurally generated worlds and gameplay, or novel game interfaces such as natural language and computer vision[8]. To make it trustless, developers need bespoke infrastructure. 

Accessor Labs’ ZK-ML tooling empowers game creators to enrich their games with vibrant, trustless, intelligent agents. Furthermore, we are developing our own games to test this capability and inspire a community of game devs. Right now we are working on a cooking game that showcases proof of LSTM model execution and are excited to share it out in Q1 2023.

We have backgrounds from Epic Games, Yahoo, Airbnb & the University of Cambridge. Accessor Labs is supported by Europe’s leading talent incubator[9].

For more information, please reach out on Twitter to @accessorlabs or say hello at hello@accessorlabs.org.