Innovative Features of Ghost
Tiger Data has unveiled Ghost, a managed PostgreSQL service tailored specifically for AI agents. This new offering aims to provide the necessary infrastructure for coding and workflow agents to execute large-scale experiments effectively. As noted in a recent report by SiliconANGLE, the service is positioned to support the evolving landscape of software development.
Ajay Kulkarni, co-founder and CEO of Tiger Data, emphasized the growing significance of coding agents in modern application development, stating, “Many new applications are being built with coding agents. Agents are the future packaging model for software.”
At the core of Ghost is Fluid Storage, an innovative storage technology developed by Tiger Data. This system employs a copy-on-write methodology, allowing multiple database instances to share the same data blocks. New blocks are only created when changes occur, ensuring that users are charged solely for the data that is modified. This approach not only optimizes storage efficiency but also reduces costs for users.
Another standout feature of Ghost is its “fast forking” capability. This function enables users to swiftly duplicate an existing dataset to create a new database in mere seconds to minutes. Such rapid replication allows agents to establish and dismantle experimental environments without the delays typically associated with traditional data copying processes.
Ghost maintains compatibility with major PostgreSQL extensions, including TimescaleDB and PostGIS, further enhancing its utility for developers. Kulkarni remarked, “It is hard to imagine that a database for agents would not be PostgreSQL.”
In terms of pricing, Tiger Data adopts a usage-based model, charging users based on actual computing consumption rather than the number of databases created. Kulkarni explained, “The database is free. You can create a database per agent or per task.” This flexible pricing structure is designed to accommodate the diverse needs of developers and their projects.
