Nikita Shamgunov, CEO of Neon, a company specializing in serverless Postgres, recently emphasized the transformative shift required in documentation for what he terms the "agentic era." In a tweet, Shamgunov stated, > "Agents grew up, so did our docs. This is what docs should look like in the agentic era." The statement, accompanied by a link, underscores a growing industry focus on how artificial intelligence (AI) agents are reshaping software development and the tools that support it.
Shamgunov is the co-founder and CEO of Neon, a venture-backed company that provides a fully managed, serverless PostgreSQL database. Neon's architecture is built on the principle of separating compute and storage, allowing for independent scaling and advanced features like instant branching of databases. This capability, akin to Git for code, enables developers to create isolated copies of their databases for testing and development, a feature that aligns with the needs of agent-driven workflows.
The concept of the "agentic era" refers to the increasing role of AI agents in automating and assisting various tasks in software development, from code generation to testing and deployment. As these agents become more sophisticated, the way developers interact with documentation and other resources needs to evolve. Traditional, static documentation may not suffice for AI agents that require dynamic, context-aware, and programmatically accessible information.
Neon's database branching, which allows for instant cloning of data and schema, is a critical enabler for agentic development. Databricks, which acquired Neon, has highlighted how this technology supports "agentic coding" by allowing thousands of AI agents to experiment with codebases and databases at virtually no cost. This provides AI agents with their own isolated environments, mirroring a developer's workflow of creating branches for new features.
The call for "agentic era" documentation suggests a move towards more intelligent, interactive, and machine-readable documentation that AI agents can effectively parse and utilize. This could involve structured data within documentation, APIs for querying information, or even AI-generated documentation tailored for agent consumption. As AI agents continue to integrate into the development lifecycle, the demand for documentation that caters to both human and artificial intelligence will only grow.
