Huwise launches agentic AI framework for data marketplaces

Huwise has launched an Agentic AI Framework for its data product marketplace platform, and says multiple customers are already using it.

The framework is built on what Huwise calls a usage context layer. It draws on how data is used, what users search for and which roles access information within a marketplace. It also incorporates company-specific semantics, such as a business glossary, to give AI agents more relevant context as they operate across an organisation.

Customers can use the framework to build and deploy AI agents across self-service data marketplaces, as well as in wider enterprise AI agent deployments. Huwise has also introduced a set of pre-built agents focused on marketplace management, data product improvement and automating routine tasks.

The launch comes as companies look for ways to make AI agents work with internal data systems rather than in isolation. A recurring challenge is connecting agents to data that reflects how an organisation actually works, including the terms, rules and usage patterns behind formal datasets.

Turnkey agents

The agents now available include Opportunity Spotter, which identifies unmet demand in a marketplace, such as searches that return no results and abandoned user journeys. Marketplace Curator recommends actions to improve the presentation and visibility of data products. Data Product Builder helps generate assets tied to a data product, including a product page, dashboard and metadata. Metadata Filler suggests additions where information is missing.

The framework also supports customer-built agents. These can be configured around a company's business rules, governance structure, operating model and data sources, rather than relying only on standard vendor tools.

That approach reflects a broader split in the market between out-of-the-box AI tools and systems that need tailoring to fit existing corporate processes. Huwise is positioning its framework as a way to do both, with pre-built agents for common marketplace tasks while leaving room for companies to develop their own.

The product also underpins Huwy, a data mining agent the company launched previously. According to Huwise, the broader framework builds on technical elements developed over many years, including APIs and an embedded Model Context Protocol server.

Context focus

Huwise argues that usage context is central to whether AI agents produce useful results. Rather than relying on a static description of datasets, the usage context layer is updated through signals generated by activity on the platform, including search behaviour and access patterns.

In practice, that means an agent can be informed not only by the existence of data, but by evidence of how that data is consumed across an organisation. Huwise says this creates a continuously enriched layer that becomes more reliable over time as use of the marketplace grows.

Jean-Marc Lazard, President and Co-Founder of Huwise, set out the company's view of the market in comments released alongside the launch.

"AI agent deployments are accelerating rapidly, driven by the promise of greater automation and increased productivity. However, context is essential to their success. An agent, no matter how efficient, is only as good as the data it relies on: it needs rich, reliable and relevant context to produce truly usable results. That's why we have developed our own agentic framework, which we are implementing to deploy AI agents based on our usage context layer. Our customers are already seeing benefits, both through our turnkey agents and those they are creating themselves. The framework will accelerate AI agent deployments and the adoption of data within organizations, so that everyone can maximize the value of their data," Lazard said.

Huwise sells a software-as-a-service data product marketplace aimed at business teams and non-specialist users. Its platform is designed to provide secure self-service access to data products for both people and AI systems, and can be deployed for internal or external marketplaces.

It serves more than 350 clients across 25 countries and has deployed more than 3,000 platforms worldwide. That installed base gives it a pool of marketplace activity from which to draw the usage signals now at the centre of the new framework.

David Thoumas, Chief Technology Officer and Co-Founder of Huwise, said the launch builds on technology the company has been developing over an extended period.

"At Huwise, our mission is to enable businesses to increase data consumption and value - both by humans and AI. Our Agentic AI framework accelerates this vision, building on the key capabilities we have been developing in the platform over the last 15 years, including robust and documented APIs, an embedded MCP server, and proven hosting and scaling capabilities. We are committed to helping our customers embrace agentic AI and are working closely with them to create new uses, based on their feedback. Our ambition has not changed: to accelerate the virtuous circle of data usage, by making the platform more relevant to attract more visitors, generate new signals and enrich usage context for agents," Thoumas said.