GoodData Releases Agent Builder to Simplify AI Development
GoodData launched Agent Builder on April 22, a development environment for building and managing AI agents that can autonomously generate insights and execute business processes. The platform joins similar offerings from Databricks, Snowflake, AWS, Google Cloud, Microsoft, Domo and Qlik.
The tool addresses a persistent problem: most AI initiatives fail because relevant data is hard to find and data quality is poor. Agent Builder connects agents to enterprise data through GoodData's semantic layer-a system that defines metrics, enforces access rules, and has run in production for nearly two decades.
How It Works
Users can build agents through code-first or no-code interfaces, including natural language options and prebuilt templates. Agents automatically access contextually relevant data through GoodData's governed semantic layer, an AI Memory layer for organizational rules and terminology, and an AI Knowledge layer for analytics context.
The platform includes a structured reasoning framework that plans tasks, selects tools, executes steps and adapts. Administrators get centralized control over agent skills, personality, knowledge and permissions, plus built-in tracing and performance monitoring.
Speed and Reuse for Existing Users
GoodData users who have already defined metrics and data relationships in the semantic layer can turn that work into functioning agents in minutes rather than starting from scratch. This matters particularly for independent software vendors (ISVs) who embed GoodData analytics in their own products-they can now build agents once and deploy them across multiple customer environments without rebuilding for each one.
Analysts cited this reuse potential as a key advantage. One noted that GoodData's two-decade track record with its semantic layer means metric definitions and access rules are already enforced before agents touch data, a maturity advantage over vendors whose semantic layers were built recently to capitalize on AI demand.
Governance and Trust
Agent Builder is designed for analytics workflows rather than simple question-and-answer interactions. Ownership of agents goes to administrators rather than engineers, reducing the risk of ungoverned changes. The context layer provides governance, predictability, auditability and security across structured and unstructured data.
One analyst said the tight coupling between agent behavior and the semantic layer reduces the risk of inconsistent or ungoverned outputs. Agents are grounded in governed, decision-ready data rather than raw inputs.
Pricing and Transparency Gaps
GoodData prices Agent Builder per workspace with unlimited users and multi-tenancy built in, a structure that differs from competitors who charge per seat. This model could help customers control AI development costs more effectively.
However, analysts identified a gap: business users who consume agent-generated answers lack visibility into how agents reached their conclusions. Currently, they would need to open a developer console to understand what context an agent pulled and why it answered a particular way. Addressing this transparency gap could broaden GoodData's appeal beyond traditional builder audiences.
Product Roadmap
GoodData's near-term focus includes adding depth to prebuilt agents, providing more customization tools, enabling agents to extend beyond GoodData's environment to other applications, and deepening governance and observability. The company said this reflects a consistent bet on enterprise-grade, governed agents as the default experience.
For product teams, Agent Builder represents a way to protect core analytics platforms as the industry shifts toward agent-assisted and automated decisions. Existing GoodData users can extend analytics into agent-assisted workflows without introducing another platform.
Learn more about AI Agents & Automation and how they fit into product strategy, or explore an AI Learning Path for Product Managers to understand AI-powered product innovation.
Your membership also unlocks:
