Developer launches Strata semantic layer tool with LLM access controls

By Splitfeed EditorialLast updated Sep 30, 2026, 18:00 UTC

On September 30, 2026, a developer named Ajo posted a Show HN submission on Hacker News introducing Strata, a semantic layer product built to govern what queries a large language model can run against business data. Ajo describes four years of work at Netflix on self-service analytics for non-technical business users as the foundation for the project. Strata is presented as a full-stack tool, covering a semantic layer, dashboards, subscriptions, and Google Sheets exports, with the stated design goal of balancing expressiveness against ease of use for people who are not data professionals.

The significance of Strata sits at a practical intersection that has become increasingly relevant as companies connect LLMs to internal databases. The core problem it addresses is one of access control: when a natural-language interface can freely query business data, the range of possible outputs becomes difficult to predict or restrict. Ajo's claim is that by designing a semantic layer that can explicitly say no to certain queries, the tool gives organizations a defined boundary between what an LLM is permitted to surface and what it is not. Whether that kind of guardrail is sufficient for enterprise-grade data governance, and whether a single-developer product can meet the security expectations of large organizations, are the questions most likely to be contested as Strata attracts scrutiny.

None of the three camps, Pro-AI, Anti-AI, or Middle Ground, have published reactions to this announcement. Pro-AI voices would typically welcome a tool like Strata as evidence that LLM integration with business systems can be made safer and more accessible, pointing to controlled query execution as a mature engineering response to earlier concerns. Anti-AI commentators would likely question whether semantic-layer restrictions are robust enough to prevent data leakage or misuse, and might raise concerns about placing sensitive business data within reach of any LLM at all. Middle Ground observers would probably focus on implementation details, asking how Strata's controls are audited, who sets the permission boundaries, and how the tool performs against established data governance frameworks.

The immediate things to watch are whether Ajo releases technical documentation on how the query restrictions are enforced and whether early adopters, particularly those from enterprise environments, report on how Strata performs against real access-control requirements. Community feedback in the Hacker News thread itself may surface the first concrete objections or endorsements within days of the launch.