Layer 1 blockchain Chromia, prized for its relational databases that allow for highly efficient data indexing and retrieval, has just launched a major upgrade. Mimir introduces a host of improvements with the greatest one being support for onchain vector databases. This enhancement should prove particularly useful for developers looking to create scalable AI applications.
Most blockchains record information in a list format, which is fine until data sets reach a certain level of complexity, at which point this system shows its shortcomings. Chromia’s relational blockchain allows data to be queried directly onchain. This ensures that dapps can perform hundreds of read-and-write operations in a single transaction, and this capability has been further bolstered thanks to the Mimir upgrade which is now live on mainnet.
Chromia Gets an Upgrade
In a web3 world with no shortage of Layer 1 chains, Chromia has sought to distinguish itself on account of its superior data handling. Given the anticipated data demands of the next generation of dapps, many of which will be powered by AI, it’s a smart strategy that essentially future-proofs Chromia, ensuring it’s capable of satisfying even the thirstiest of applications. And when it comes to AI, that thirst can run to exabytes of data.
Vector databases are a core component of AI workflows, supporting essential functions such as similarity search and long-term memory as well as contextual text retrieval. Up until now, most dapps have cheated when it comes to this part, relying on centralized servers to store and manage their vector databases. Chromia, thanks to its Mimir mainnet upgrade, solves this, allowing vector data to be stored directly onchain.
Mimir Makes AI Easier
When talking about abstract concepts such as vector databases, it can be hard to visualize exactly what the underlying technology offers, and why it’s such a game-changer compared to what’s gone before. To add some color to this, Chromia recently published a case study on Shadows of Harrow City as a proof of concept for its vector database solution. The game takes the form of a whodunnit in which players take on the role of the detective tasked with working out who the killer is.
Behind the scenes, Chromia’s onchain vector database is used to store packets of information associated with core concepts and keywords. Whenever the player engages with an NPC in Shadows, the relevant knowledge “bricks” are retrieved from the database based on similarity searches. Formerly, game developer Aarda AI relied on AWS for this. Now, thanks to Chromia, it can all be done onchain.
As well as making Chromia the first L1 with fully onchain vector database capability, the Mimir upgrade introduces the AI Inference Extension. This allows open-source models such as DeepSeek to run on Chromia provider nodes, again moving developers away from centralized services such as AWS to fully decentralized infra. As a result, developers are now able to create AI-powered apps that operate 100% onchain without compromising on performance.
