Create README.md
Browse files# Agentic Data Access Benchmark (ADA benchmark)
ADA benchmark is a set of questions over few closed-domains which can have tremendous benefit from AI/LLMs. If you are evaluating an assistant product or building your own, you can use the ADA benchmark to qualitatively measure the capabilities of your assistant/agent.
Although, there are ever increasing number of AI assistants over various domains, their performance is often limited to simple questions. The most important element in building useful closed-domain AI assistants/agents is their ability to access and compute on data on demand as per the task at hand.
## Test
If you want to test the performance of your own assistant over these questions, you can find the sample schemas and data in this repo: https://github.com/hasura/agentic-data-access-benchmark