Presentation: Powering the Future: Building Your GenAI Infrastructure Stack

2026-05-19 · Source: InfoQ · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Cloud Computing & IT Infrastructure · Depth: Advanced, extended

Summary

Intuit's Distinguished Engineer Merrin Kurian detailed the company's GenAI infrastructure stack, GenOS, and its role in scaling AI agent development across 8,000 developers, enabling 3,500+ production experiments. The presentation highlighted Intuit's "fixed, flexible, free" framework for technology adoption and its strategy to unify existing capabilities while enhancing them for AI. Key aspects covered include the evolution of AI agents from conversational assistants to "done-for-you" experiences, addressing common agent failure modes, and implementing an "LLM-as-a-judge" evaluation strategy. Intuit's GenOS platform provides an AI Workbench for prompt management, RAG pipelines, and evaluation frameworks, supporting 15+ models across 70+ versions. The company also emphasizes preparing for future agents by designing "tool-ready" APIs, enhancing data metadata, and supporting multimodal user experiences.

Key takeaway

For AI Architects and MLOps Engineers building enterprise GenAI, you should adopt a structured platform approach like Intuit's GenOS. Prioritize standardizing core infrastructure while offering flexible options for developers. Implement continuous evaluation and robust governance from the outset to manage agent failure modes and ensure compliance. Focus on designing "tool-ready" APIs and enriching data with metadata to prepare for future agent autonomy.

Key insights

Intuit's GenOS platform and "fixed, flexible, free" framework accelerate enterprise-scale AI agent development through structured governance and continuous evaluation.

Principles

• Standardize core platform concerns.
• Offer flexible, opinionated options.
• Continuously evaluate and adapt.

Method

Intuit's GenOS provides an AI Workbench for prompt management, RAG pipelines, and evaluation. It includes GenRuntime for multi-agent orchestration and GenUX for user experience, all supported by a CI/CD-enabled Agent Starter Kit.

In practice

• Design APIs to be "tool-ready" for agents.
• Enhance data with metadata for agent context.
• Implement multimodal native user experiences.

Topics

• GenAI Infrastructure
• AI Agent Development
• LLM Evaluation
• GenOS Platform
• API Design
• MLOps

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

Related on AIssential

• The Agent Stack Bet — Current agent deployments suffer from a "stack ceiling" requiring fundamental architectural shifts for enterprise-grade reliability and governance.
• Generative AI for Business: A Complete Strategy and Implementation Guide — Generative AI drives significant economic value by enabling broad automation and content creation across diverse business functions.
• How Addepar Scales Investment Workflows with Databricks AI Agents — Unified data, governance, and agent orchestration are critical for secure, scalable GenAI in regulated financial services.
• Managing AI Agents at Scale with Google Cloud's Riyaz Habibbhai — Google Cloud's AI strategy focuses on an integrated, open, and simplified full-stack platform, evolving Vertex AI into the Gemini enterprise agent platform.
• Creating a new AI platform — Generative AI demands new platform capabilities beyond traditional ML infrastructure, even for established systems.
• The Tech Stack Powering Wise — Standardized internal platforms enable autonomous teams to scale complex, high-reliability financial services.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by InfoQ.