Alation Aims to Automate Data Management Drudgery with AI

Companies that struggle to effectively use modern data management tools such as data catalogs may be interested in new software unveiled by Alation today. The company unveiled an Agentic Platform that uses AI agents to automate time-consuming, manual data management processes. Alation’s first autonomous AI agent tackles data quality, but more are on the way, according to CEO Satyen Sangani.

Alation gets credit with inventing the data catalog more than a decade ago, and over the ensuing years, it’s expanded its metadata-driven ethos of connecting people and data into adjacent areas of data management, as well as data governance.

At the same time, customer data ecosystems have exploded in complexity. Data silos are everywhere, with complex pipelines connecting them, but serving up a seemingly endless supply of suspect data. The hopes and dreams of automated modern data management have not always panned out as expected, which has cast a shadow on companies’ push to become data-driven, Sangani says.

“The state of the art was, you have a business problem, you need data to solve it. Do all of these data things–catalog it, improve the quality, curate it, go build policies, hire a team to build data access control policies. You do all these things and then you can get to your outcome, right” Sangani tells BigDATAwire in an interview. “But the problem is, in that list of things, people get lost in the forest and they don’t know where they are and they just don’t know what to go do.”

Alation plans to have a handful of AI agents as part of its Agentic Platform

That’s the backdrop for the introduction of Alation’s Agentic Platform announcement. Instead of requiring data curators to understand their users’ data needs at a very detailed level, and then meticulously craft data policies that accurately reflect those needs within the context of the customers’ industry and the specific business goal, Alation is now using AI to automate much of that grunt work.

The Alation Agentic Platform brings a software development kit (SDK) that enables customers to connect Alation AI agents into their existing data environment. It built upon Anthropic’s Model Context Protocol (MCP), which is likely on its way to becoming an industry standard for AI agent management.

Alation also unveiled its first autonomous agent as part of the Agentic Platform. Alation Data Quality (DQ) uses AI to understand customers’ existing data usage patterns and then automatically create data quality rules, which are then executed on the actual platform storing (or accessing) the data. The Agentic Platform SDK and Alation DQ are slated to become available during the second quarter.

The idea with the Agentic Platform, which builds off Alation’s 2023 addition of generative AI-powered suggestions for data policy generation, is for AI to slowly take over time-consuming tasks, with the goal of delivering the right level of data management automation for each individual company, customized for their specific needs, Sangani says.

“What we’re doing is we’re leveraging agents in order to be able to guide people down the path of deciding what rules they have to apply in the first place,” he says. “What is amazing about these agents is it takes all of the knowledge of using software–the knowledge of best practices and data governance, the knowledge of the company and its context, what industry you’re in, how big you are, where you’re located, what is a best practice program in in data governance, what are the best policies that you can build–and it brings all of that to bear in guiding people through the appropriate level of actions.”

Alation Data Quality uses AI to suggest data quality rules for customers, and then automatically implements them

Alation sees a handful of AI agents starting off doing data policy work on a semi-autonomous basis, with supervision from human data curators. As time goes by, the agents will become better tuned and adjusted to the company’s specific workflows and needs, the AI agents will work more autonomously, Sangani says. Eventually, there will be hundreds of sub-agents handling a range of data management tasks, he says.

Once the DQ agents are in place, Alation plans to roll out additional AI agents. Sangani declined to provide specifics, but indicated they likely will be in the general context of the types of data management tasks that Alation is known for.

“These [agents] are not easy things to build. There’s a lot of context that you have to provide,” Sangani says. “There’s a lot of guessing and checking. You have to constrain them in the appropriate ways. And so the goal is not to build 15,000 things and have them all screw up all your data overnight. The goal is to build thoughtfully and progressively, so that as we build more of these, we will get more and more knowledge about how to do that better, and our customers will too.”

Alation also today announced the Data Products Marketplace, a new offering that allows customers to set up and host one or more repositories of data products that allow business users and data teams to browse and download trusted data products. These marketplaces can be used by either internal or external users, Alation says.

Alation is helping its customers adopt and use data products, which are reusable collections of pre-curated data that data analysts, data scientists, and others can use. They’re becoming a popular way for companies to share data with internal and external users, particularly for users who lack the technical skills to do their own data prep.

“The idea is to create flexibility and a stronger level of curation,” Sangani says of the Data Products Marketplace. “What we’ve found is that the marketplace helps solve a lot of problems for core data consumers. Those products also are discoverable not just by people, but also can be leveraged by applications and programs and LLMs, so they’re AI -ready as well.”

Alation Expands Into AI Governance

Alation Turns to GenAI to Automate Data Governance Tasks

The post Alation Aims to Automate Data Management Drudgery with AI appeared first on BigDATAwire.