Content Copyright © 2023 Bloor. All Rights Reserved.
Also posted on: Bloor blogs
Alation was founded in 2012, headquartered in Redwood City on the outskirts of San Francisco, with around 800 employees currently. A recent European user group held in London in October 2023 saw over 400 attendees. Alation arguably pioneered the first modern data catalog; one interesting feature is its ability to extract useful metadata from data sources such as SQL database logs, for example to determine data popularity and usage patterns.
There were three technology roadmap themes presented earlier this month at revAlation London, Alation’s annual global conference series. Generative AI has been embedded within the product using a model based on Meta’s LLaMA foundation model. Given the almost inevitable hallucinations that even a well-trained AI will produce, producing confidence levels as seen in a product demonstration, of generated AI content seems to be a sensible approach. This feature will be generally available in 2024.
Alation’s Open Connector Framework allows third-party partners to build connectivity between Alation and a wide range of other technologies. Existing connectors to Snowflake, Google Big Query and Databricks are examples, as is a connector to the Profisee master data management tool. There are currently over one hundred supported connectors, including links to Oracle, Postgres and SQL Server.
Alation has extensive data lineage and impact analysis capabilities, and this now includes column-level lineage and integration with data quality. Alation partners with a number of data quality vendors including Experian and Monte Carlo, with APIs to enable smooth integration of data quality functions within Alation.
One quote from a UK retailer at the conference seemed quite appropriate. “If an AI model affects millions of customers and it goes wrong, we can’t just say ‘oh well it is AI’. Products like Alation Data Intelligence Platform set a foundation for documenting data relationships and data quality in an organization, and as companies seek to implement generative AI they will need to ensure that data quality is as high as possible, since the quality of output of an AI is highly dependent on the quality of the data that it was trained on.