Orion at Gravity: Trustworthy AI Analysts for the Enterprise

2026-03-08 · Source: Data Engineering Podcast · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Cloud Computing & IT Infrastructure · Depth: Advanced, extended

Summary

Lucas Thelosen and Drew Gilson, co-founders of Gravity, discussed their agentic analytics platform, Orion, designed for enterprise use. Orion functions as an AI analyst, integrating data semantics with business context to provide accurate, actionable insights. It utilizes governed, role-specific "custom agents" for analysis, recommendations, and meeting preparation, emphasizing accuracy, lineage transparency, and human-in-the-loop feedback. The discussion covered the evolution of semantic layers, agent memory, retrieval, and operating across diverse data sources, including multiple warehouses and external context like documents and weather data. They highlighted Orion's ability to uncover board-level issues, accelerate executive preparation, and reveal BI investment utilization in public companies, advocating for accessible models and a focus on business actions over mere metrics to achieve ROI.

Key takeaway

For data leaders aiming to transition from static dashboards to dynamic, action-oriented decision-making, you should explore agentic analytics platforms like Orion. This approach allows your teams to move beyond basic metrics, uncover deeper "why" questions, and receive proactive, context-rich recommendations, ultimately driving measurable ROI by focusing on business actions rather than just data visualization. Prioritize accessible data models and robust context engineering to maximize AI's impact.

Key insights

Agentic analytics combines semantic layers and broad context engineering to deliver trustworthy, actionable business insights.

Principles

• Trust and 100% accuracy are paramount in enterprise AI.
• Context engineering extends beyond data to business operations.
• Accessible, non-proprietary data models enhance AI utility.

Method

Orion builds a knowledge base from dbt, Looker, and database metadata, then integrates business context. Data leaders validate and refine this understanding, creating guardrailed "custom agents" for specific roles or projects.

In practice

• Use AI to automate data model updates and metadata management.
• Integrate external context like weather data for richer analysis.
• Prioritize data infrastructure speed for effective AI deployment.

Topics

• Agentic Analytics
• Semantic Layer
• Context Engineering
• AI Analysts
• Data Governance

Best for: CTO, VP of Engineering/Data, Director of AI/ML, Machine Learning Engineer, Data Engineer, AI Architect

Related on AIssential

• Data Products: The Essential Context for Enterprise AI — Enterprise AI agent success hinges on robust, architected context layers, not just advanced models.
• Presentation: AI Agents to Make Sense of Data at OpenAI — OpenAI's Kepler agent uses advanced AI techniques to automate complex data analysis across vast datasets, improving data accessibility and accuracy.
• Why Foundation Models Haven’t Replaced Classical Machine Learning — Classical ML remains vital for proprietary, structured data, requiring context-aware agentic systems to navigate fragmented enterprise ecosystems.
• Podcast: Context is the Key to the Agentic Architecture Revolution: A Conversation with Baruch Sadogursky — Context engineering, not prompt engineering, is key to leveraging AI agents for software development by making specifications the source of truth.
• What Is a Context Graph — and Why Is Everyone Talking About It? — Context graphs provide machine-readable institutional memory, enabling AI to accurately interpret complex organizational knowledge and decision logic.
• Unlocking the Data Layer for Agentic AI with Simba Khadder — Dynamic context engines, built on materialized views and semantic layers, are crucial for scaling autonomous AI agents.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Data Engineering Podcast.