GoodData on Wednesday launched a Model Context Protocol server, enabling agents and other AI tools to access customers' proprietary data to develop analytics tools and continuously analyze data.
Released in November 2024 by AI provider Anthropic, the Model Context Protocol (MCP) is a set of open-source code that standardizes connecting models that inform agents and other AI applications with databases, data lakehouses and other data sources.
Because MCP is a repeatable standard for feeding AI tools the data they require to properly perform -- saving developers from having to configure complex pipelines each time they build a new AI application -- many data management vendors provide MCP support within development platforms. AWS and Microsoft were among the first to add support for MCP. Throughout 2025, vendors such as Alation, Confluent -- now under agreement to be acquired by IBM -- Databricks, Oracle, SnapLogic, Snowflake, Starburst, StarTree and Teradata also unveiled MCP support.
Now, analytics vendors are adding MCP servers as well to provide AI-powered insight generation capabilities. For example, SAS, Sisense, Tableau and ThoughtSpot all provide MCP servers.
GoodData is following suit in a move that is important for customers because it expands AI-powered analysis beyond chat interfaces by enabling AI tools to take on aspects of the analytics workflow, according to Mike Leone, an analyst at Omdia, a division of Informa TechTarget.
"The significance here is that it moves AI from a 'read-only' capability to a 'read-write' role within the analytics stack," he said. "Right now, most AI tools in this space are just chat interfaces that can answer a single question but cannot tell the AI to actually go in and update a metric or reconfigure a semantic model. GoodData's approach allows users to offload the actual engineering work."
Based in San Francisco, GoodData is a longtime analytics vendor that now offers a platform including a semantic layer to ease data access and discovery, and a data lake to store data.
Powering BI with AI
AI has altered business intelligence since OpenAI's November 2022 launch of ChatGPT represented significant improvement in generative AI (GenAI) technology.
The significance here is that it moves AI from a 'read-only' capability to a 'read-write' role within the analytics stack. Right now, most AI tools in this space are just chat interfaces that can answer a single question but cannot tell the AI to actually go in and update a metric or reconfigure a semantic model.
Mike LeoneAnalyst, Omdia, a division of Informa TechTarget
BI had historically been limited to a small percentage of experts within organizations. Data analysis necessitated expertise in statistics, coding knowledge and data literacy. Even self-service platforms that reduced some of the barriers to more widespread use required some coding and users still needed to be data literate.
As a result, despite attempts to simplify using complex platforms, on average less than one-third of employees within enterprises used analytics as part of their jobs as recently as the spring of 2022, according to a study by BARC and Eckerson Group.
GenAI changed that, enabling workers to explore and analyze data using true natural language.
Now, with agents capable of autonomously performing tasks such as modeling data, building dashboards and reports and analyzing data, AI is not only making the use of BI to inform decisions more widespread but exponentially speeding up the process of generating business insights.
With its MCP server, GoodData is enabling users to automate certain processes in a move motivated by a combination of customer feedback and what the vendor viewed as the natural evolution of its platform, according to Peter Fedorocko, GoodData's field CTO.
"We see strong demand and interest from our customers who want to effortlessly give their AI and agents analytical capabilities," he said. "At the same time, it was also a low-hanging fruit for us, because … all of our functionality was already programmatically available. Encapsulating it with MCP simply exposes it to AI and agents."
GoodData's new MCP server is designed to facilitate connections between customers' AI applications and the vendor's platform so the AI tools can take on work that previously took humans significant time and effort.
For example, by connecting an agent with the vendor's analytics-as-code approach to development -- a method that standardizes and automatically generates code for building analytics workflows -- customers can automate developing and updating the analytics process from data preparation through insight generation and recommended actions.
In addition, connecting MCP-compatible agents with semantic models, business metrics, dashboards and other data products enables customers to create agents that provide continuous AI-powered analysis.
Because GoodData is doing more than just facilitating access to data with its MCP server, its addition is significant, according to Michael Ni, an analyst at Constellation Research.
"With MCP servers rapidly becoming table stakes for analytic platform providers, GoodData's approach stands out by exposing more than just data or tools," he said. "GoodData exposes governed analytics logic as an executable infrastructure for AI [and offers] a reference pattern for the broader market to more easily operationalize their analytics and insights for machines and AI."
While significant for GoodData customers, GoodData's launch of an MCP server could help the vendor stand apart from some competitors, according to Leone.
MCP support, though now so standard among data management vendors that those not providing such capabilities risk losing customers, is not yet widespread among analytics providers. However, even among analytics vendors offering MCP servers, connectivity between agents and semantic layers distinguishes GoodData given that not all BI vendors offer semantic modeling capabilities.
"The differentiation here comes from applying that connectivity to the semantic layer," Leone said. "This effectively lets AI interact with the governance and business rules instead of just the data pipelines. It shows a shift where the BI platform becomes the brain that other AI agents can consult for logic rather than just a visualization tool for humans."
Ni similarly noted that GoodData's opportunity to stand apart from other analytics vendors now providing MCP servers is by enabling AI tools to work with not only data but semantic layers and other data products.
"Most MCP servers today live in data platforms like Databricks and Snowflake because that is where automation has lived for years," he said. "I'm look forward to seeing how GoodData exposes their analytics as infrastructure and enables AI to work … directly with semantic models, metrics and dashboards."
Looking ahead
With GoodData's MCP server now generally available, one of the vendor's next initiatives is to launch its own agents, according to Fedorocko.
The MCP server enables customers to build agents aimed at automating the analytics workflow. By providing its own agents for tasks such as data modeling and building dashboards, GoodData would remove some of the development efforts customers must still perform on their own.
ThoughtSpot and Tableau are two analytics specialists already making their platforms more agentic. ThoughtSpot, in fact, is building agents for tasks such as building semantic models and embedding BI in user workflows in an effort to automate its entire platform.
"Our main focus in [the first half of 2026] is launching our own agents and agentic workflows that automate much of the analytical process, as well as continuing to be the … provider of analytical capabilities for our customers' agents," Fedorocko said. "This is all supported by investments into a robust context layer that grants trusted and governed multimodal data and context to every agent."
Adding tools such as agents that simplify using GoodData's platform would be wise, according to Leone.
The MCP server will aid data scientists, engineers and other experts that spend significant time developing AI and analytics tools. However, there is still work to be done to improve GoodData's user experience.
"The next hurdle is translating all the engineering power into simplicity for the business user," Leone said. "They've nailed the back-end infrastructure with this announcement. Now, they need to make sure the front-end experience feels just as intuitive for someone who never writes code. If they can mask that complexity while delivering the speed, they'll attract a much wider audience."
Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.
