Data 360

AI Asset and Model Lifecycle

Sabarish K S, Adrian Massey, Som Liengtiraphan, Sabarish K S

AI Asset Details

Why this work is important

Customers need a single place to understand, manage, and improve their AI models end to end. Clear visibility into model health and next steps helps organizations confidently move AI from training into production.

The problem this work will solve

Model information is fragmented across multiple pages, making it difficult to understand a model's current state. Users lack guidance on what to do after training or when a model requires attention.

Steps taken to get there

Unified the workspace into 'AI Models' to ensure consistency and added real-time tracking for model health.

Current status

GA in the 264 release (October 2026)

Figma designs
Visual from AI Asset Details, source slide 29.
1 / 2

AI Asset Tasks

Why this work is important

Customers expect AI to solve an expanding range of business problems Investing in shared patterns reduces design and engineering effort as the AI portfolio grows.

The problem this work will solve

Scaling the AI portfolio risked increasing product complexity for customers. Established shared UX patterns that create a consistent experience across AI asset types.

Current status

Shared UX patterns adopted across new AI experiences 3 new asset types released in Q2 3 additional types for Q3

AI Models Figma Project
Visual from AI Asset Tasks, source slide 30.

AI Models Northstar

Why this work is important

AI is rapidly evolving toward agent-first experiences, requiring a shared vision that aligns teams around a common future. A unified North Star reduces fragmented product decisions and consistently with the broader Data 360 experience.

The problem this work will solve

Independent product decisions risk creating disconnected experiences Teams have differing assumptions about the role of agents, traditional workflows, and AI asset management..

Current status

Planning cross-functional workshops to capture product constraints, assumptions, and end-to-end lifecycle requirements.