---
title: "How Data 360 Builds Trusted Context: The Enduring Layer for Enterprise AI"
description: "Enterprise AI needs a way to assemble the smallest sufficient set of current, relevant, and authorized evidence for each agent turn instead of sending an entire data estate to a model. Salesforce presents Data 360 as a shared Trusted Context runtime spanning Salesforce, external platforms such as Databricks and Snowflake, operational systems, documents, knowledge bases, and live signals. Its progressive pipeline connects data through ingestion and Zero Copy, then harmonizes, unifies, models, retrieves, reconciles, governs, and compiles evidence into a token-fit Context Pack with citations and action contracts. The Agent Context Engine also returns typed interaction traces, promoting only approved facts into durable memory and validated outcomes into learnings. Q2 FY27 production usage is reported at 104.3 trillion connected records, including 81.5 trillion Zero Copy rows, while proposed benchmarking covers correctness, authorization, provenance, freshness, latency, cost, and memory continuity."
---

# How Data 360 Builds Trusted Context: The Enduring Layer for Enterprise AI

[Salesforce](https://yomu.fyi/company/salesforce) · Scott Nyberg · Sep 16, 2026

**Type:** Explainer

## Summary

Enterprise AI needs a way to assemble the smallest sufficient set of current, relevant, and authorized evidence for each agent turn instead of sending an entire data estate to a model. Salesforce presents Data 360 as a shared Trusted Context runtime spanning Salesforce, external platforms such as Databricks and Snowflake, operational systems, documents, knowledge bases, and live signals. Its progressive pipeline connects data through ingestion and Zero Copy, then harmonizes, unifies, models, retrieves, reconciles, governs, and compiles evidence into a token-fit Context Pack with citations and action contracts. The Agent Context Engine also returns typed interaction traces, promoting only approved facts into durable memory and validated outcomes into learnings. Q2 FY27 production usage is reported at 104.3 trillion connected records, including 81.5 trillion Zero Copy rows, while proposed benchmarking covers correctness, authorization, provenance, freshness, latency, cost, and memory continuity.

## Context

Trusted Context is the governed runtime layer between enterprise data and AI execution. It supplies current, relevant, and authorized evidence while preserving approved memory, applying access and privacy controls, and connecting models and agents to business systems without requiring every interaction to load the full enterprise data estate.

## Approach / What changed

Data 360 uses ingestion and Zero Copy to connect enterprise data, then applies a progressive context pipeline that harmonizes, unifies, infers, and models information as knowledge. The Agent Context Engine resolves identity and purpose, plans retrieval paths, reconciles evidence, applies governance, compiles a token-fit Context Pack, and selectively promotes typed interaction traces into durable memory or validated learnings.

## Takeaways

- The Agent Context Engine divides runtime assembly into Resolve, Plan, Reconcile, Govern, Compile, and Learn selectively stages, applying authority, freshness, provenance, and policy before evidence reaches an agent.
- Data 360 distinguishes working context, short-term memory, long-term memory, and learnings; transcripts do not automatically become durable enterprise memory, and only approved facts or validated outcomes are promoted.
- Q2 FY27 production usage is reported at 104.3 trillion connected records, including 81.5 trillion Zero Copy rows. The source confines performance comparisons to their published workloads, configurations, and test methods.

## FAQ

### How does the Agent Context Engine keep each prompt within runtime limits?

It retrieves, reconciles, authorizes, and assembles approved evidence, working state, memory, and citations into a token-fit Context Pack. Context assembly is measured against explicit token, latency, freshness, and cost budgets.

### What prevents every conversation from becoming permanent enterprise memory?

The system separates short-term memory from long-term memory and learnings. Only approved facts may be promoted into durable memory, while validated outcomes can become learnings; a transcript alone does not automatically become enterprise memory.

### Why does the post propose an Agentic Context Benchmark?

Traditional data benchmarks do not measure the full workload across structured and unstructured data, relationships, memory, fresh events, and source-specific permissions. The proposed benchmark would measure correctness, access enforcement, provenance, freshness, latency, cost, useful evidence per token, memory continuity, and improvement from validated outcomes.

**Tags:** [AI](https://yomu.fyi/topic/ai), [Data 360](https://yomu.fyi/topic/data-360), [Data Governance](https://yomu.fyi/topic/data-governance)

- Source: [Salesforce](https://engineering.salesforce.com/how-data-360-builds-trusted-context-the-enduring-layer-for-enterprise-ai)
- Source URL: https://engineering.salesforce.com/how-data-360-builds-trusted-context-the-enduring-layer-for-enterprise-ai
- Ingested by Yomu: 2026-09-16T20:00:48.315Z

[Read original post](https://engineering.salesforce.com/how-data-360-builds-trusted-context-the-enduring-layer-for-enterprise-ai)