Alation launches AI governance suite to meet rising need
With interest in generative AI increasing, the vendor's new suite aims to help enterprises both mitigate risks as well as confidently use new applications.
Alation on Wednesday unveiled an AI governance suite aimed at enabling enterprises to improve returns on their increasing investments in AI and machine learning projects by using secure trusted data.
The set of tools includes AI documentation and collaboration capabilities, auditable AI lineage and traceability, compliance and risk mitigation features and expert guidance -- including best practices -- as customers develop models and applications.
Enterprise interest in developing AI, fueled by the explosion of interest in generative AI over the past two years, is increasing. Just as rising interest in self-service analytics required enterprises to develop data governance frameworks over the past 10 to 20 years, widespread use of AI tools requires AI governance frameworks, according to Doug Henschen, an analyst at Constellation Research.
Data and AI governance frameworks are sets of guidelines that enable organizations to protect themselves from regulatory violations as they use data and AI tools to inform decisions and automate processes. Simultaneously, they enable employees to confidently use data and AI tools in their workflows.
Vendors such as Alation and Collibra have long provided data governance capabilities to assure organizations as they operationalize their data. As the use of AI capabilities increases, it makes sense for vendors to provide similar governance capabilities for operationalizing AI.
"Being a data-driven activity, the development of AI must be governed as rigorously as we govern our data, so it's a natural extension of data governance programs," Henschen said. "Organizations need help with these challenges, so it's good to see metadata management -- a.k.a. data intelligence -- vendors adding functionality to address AI-specific risks and emerging regulatory requirements."
Likewise, Stewart Bond, an analyst at IDC, said that there's a real need for AI governance as enterprises increasingly rely on AI tools.
AI governance is critical. Research IDC has done shows that many of the issues that drive the need for AI governance are data governance-related. Without governance, laws may be broken, sensitive information leaked, or incorrect outcomes will happen, such as hallucinations.
Stewart BondAnalyst, IDC
Much of the need is related to the underlying data used to train and maintain models and applications. Enterprises must protect sensitive information and ensure that enterprises use only quality data to reduce the likelihood of incorrect and potentially toxic outcomes.
"AI governance is critical," Bond said. "Research IDC has done shows that many of the issues that drive the need for AI governance are data governance-related. Without governance, laws may be broken, sensitive information leaked or incorrect outcomes will happen, such as hallucinations."
Based in Redwood City, Calif., Alation is a data catalog specialist whose Data Intelligence Platform enables customers to integrate and organize data from disparate sources to train AI models and applications and inform data products.
In addition to Collibra, competitors include Atlan and Informatica, among others.
Governing AI
Enterprise interest in AI has exploded over the past two years, sparked by OpenAI's launch of ChatGPT in November 2022. ChatGPT represented a significant increase in generative AI capabilities, which has only been improved upon since ChatGPT's introduction with myriad other tech vendors also launching generative AI models and OpenAI continually adding its own updates.
Key capabilities powered by generative AI models used with an enterprise's proprietary data include true natural language processing that enables users of all skill levels to work with data and process automation.
Given those benefits, enterprises have responded by developing generative AI capabilities such as AI assistants that enable users to converse with their data and tools that automate coding, documentation and other repetitive tasks. They have also responded by developing more traditional AI and machine learning models and applications.
Meanwhile, vendors such as Databricks, Snowflake, AWS, Google Cloud and many others have built environments to make it easier for customers to develop AI tools.
With AI becoming more ubiquitous -- and potentially becoming the primary means of consuming data and informing decisions -- enterprises need to ensure that their AI tools are properly trained and used.
Enter AI governance.
Enterprises can develop their own AI governance tools, just as they can build their own analytics and data management systems. But just as it's often easier to use analytics and data management tools provided by a vendor, it's also simpler to use AI governance capabilities provided by a specialist.
Collibra responded to rising demand for AI governance by launching AI Governance in April. Alation is now taking a similar approach with a suite Bond called "a significant addition for Alation customers and a logical step in Alation's evolution."
Intelligence about data has historically been Alation's focus. The vendor's AI governance capabilities extend that to intelligence about models.
"Aligning data intelligence with model intelligence can help assure the right data is being used with the right model, at the right time, for the right reason, and within regulatory constraints," Bond said. "This is what Alation is enabling."
Alation's AI governance suite includes the following capabilities:
Intelligent search so that data scientists, engineers, machine learning specialists and other AI model and application developers can quickly locate and tag relevant data sets that meet compliance regulations so they can be used to build a trusted foundation for AI tools.
Centralized collaboration including model card templates so enterprises can create a primary data source for documenting and managing their AI assets that is easily visible to all potential users.
AI model lineage to provide users with full visibility of a model's lifecycle, including cataloging the data sets used to inform AI systems, the ways generative AI models were trained to understand an organization and the model's outputs.
Automated capabilities that flag noncompliant data sets to ensure that models are developed using high quality data that does not run afoul of regulations.
Expert Services to provide customers with guidance and best practices for developing model cards that can be catalogued.
Of particular note are the model lineage capabilities, according to Henschen. While other vendors are also addressing AI governance, visibility into specific training data and prompt engineering is a potential differentiator for Alation.
"I've seen AI governance-related announcements from rivals including Collibra, but Alation is unique in announcing training-data-specific, prompt-specific and model-specific capabilities," Henschen said.
Motivation for developing an AI governance suite came from both customer feedback as well as Alation's own observation of market conditions, according to Satyen Sangani, the vendor's co-founder and CEO.
In terms of market conditions, the combination of rising adoption in AI and new regulations provided a push.
"The urgency for AI governance has escalated due to new regulatory pressures and increased AI adoption," Sangani said. "Legislation … underscores a growing mandate for companies to track, manage and understand their AI deployments."
Customers, meanwhile, are increasingly aware of the costs related to ungoverned AI models and applications, he continued.
"Customers consistently highlight the need for reliable, AI-ready data to mitigate risks, support compliance and remain competitive amid tightening regulations and growing public scrutiny," Sangani said.
Looking ahead
With Alation's AI governance suite now available, Alation's roadmap is focused on adding AI capabilities that make data curation and data discovery easier and more efficient, according to Sangani.
In addition, knowing that data management can be challenging, the vendor aims to provide new tools to help customers use data to inform their most valuable projects, he said.
Henschen, meanwhile, said that while Alation's AI governance capabilities address some aspects of the model development process, others are not covered by the suite.
As a result, the vendor would be wise to update its AI governance toolkit with new features through product development and integrations with partners.
"There's a lot to the model development lifecycle, so … there are more capabilities that could be delivered and integrations yet to be created for market-leading model and service development products and platforms," Henschen said.
Bond similarly noted that Alation has room to add more features to its AI governance suite. For example, government policies for AI are likely to become more stringent. As those new policies emerge, Alation will have an opportunity to add new capabilities.
"We are just starting to scratch the surface of what is needed, and can be done, in AI governance," Bond said.
Eric Avidon is a senior news writer for TechTarget Editorial and a journalist with more than 25 years of experience. He covers analytics and data management.
