SetupAI
EN
來源較早收集於 40m

SAS 推出受治理 Copilots 和代理框架

閱讀原文： Computerworld
#ai-governance#agent-frameworks#copilot-tools

SAS 治理 Copilots + MCP 標準化代理存取企業資料/工具（28 字）

30 秒速覽

有什麼變化

SAS Viya Copilot 嵌入人治理 AI，用於自然語言分析、程式碼生成和模型指導。

為什麼重要

讓企業安全擴展代理式 AI，解決碎片化環境中的可見性和信任缺口。将 SAS 定位為金融和醫療等產業治理 AI 運營領導者。

下一步行動

整合 SAS Viya MCP 伺服器，將您的 LLM 代理安全連接至 SAS 分析。

誰應關注：Enterprise & Security Teams

關鍵要點

  • •SAS Viya Copilot 嵌入人治理 AI，用於自然語言分析、程式碼生成和模型指導。
  • •初始 Copilots：資產負債管理用於財務風險，以及 Health Clinical Data Discovery。
  • •MCP 伺服器標準化外部代理存取 SAS 工具，無需自訂整合。
  • •Agentic AI Accelerator 提供程式碼和元件用於治理多代理系統。
  • •聚焦從 AI 形式轉向行動的治理。

深度解析

本篇為 AI 生成分析，非原文內容。

增強重點摘要

  • •SAS has integrated these new agentic capabilities directly into the SAS Viya platform's existing 'SAS Viya Workbench' environment, allowing developers to maintain existing CI/CD pipelines while deploying agentic workflows.
  • •The Agentic AI Accelerator utilizes a 'human-in-the-loop' orchestration layer that enforces SAS's proprietary 'Model Risk Management' (MRM) framework, ensuring that autonomous agents cannot bypass regulatory compliance checks.
  • •SAS is positioning its MCP implementation as a direct response to the fragmentation of enterprise data silos, specifically targeting the interoperability challenges between SAS Viya and third-party LLM providers like OpenAI and Anthropic.

競品分析

Primary Focus
SAS Viya Copilot
Regulated Analytics/Risk
Databricks AI Agents
Data Engineering/MLOps
Microsoft Fabric Copilot
General Business/BI
Governance
SAS Viya Copilot
Built-in MRM/Compliance
Databricks AI Agents
Unity Catalog/Governance
Microsoft Fabric Copilot
Purview/Responsible AI
Interoperability
SAS Viya Copilot
MCP Standard
Databricks AI Agents
Mosaic AI/Open Source
Microsoft Fabric Copilot
Native Azure Ecosystem

技術深入

  • •The SAS Viya Copilot architecture leverages a RAG-based (Retrieval-Augmented Generation) approach that connects directly to the SAS Viya data fabric, ensuring data never leaves the secure perimeter during inference.
  • •The MCP (Model Context Protocol) implementation acts as a standardized bridge, allowing the SAS Viya Copilot to query external data sources or trigger actions in third-party applications without requiring custom API wrappers for every integration.
  • •The Agentic AI Accelerator provides a library of pre-built 'Agent Templates' that utilize LangChain-compatible patterns, specifically optimized for SAS's internal execution engine to handle high-concurrency analytical tasks.
  • •Governance is enforced at the 'Agentic Orchestration' layer, where every action taken by an agent is logged in the SAS Viya audit trail, providing a full lineage of decision-making for regulatory reporting.

前景展望基於引用來源的 AI 分析

SAS will shift its primary revenue model toward consumption-based pricing for agentic task execution.
The transition from static analytics to autonomous agentic workflows necessitates a move away from traditional seat-based licensing to reflect the compute-intensive nature of agentic reasoning.
SAS will release an open-source SDK for the Agentic AI Accelerator by Q4 2026.
To compete with the rapid adoption of open-source agent frameworks, SAS must lower the barrier to entry for developers to build custom agents on top of the Viya platform.

時間線

2023-05
SAS announces $1 billion investment in AI-driven industry solutions.
2024-03
SAS Viya Workbench is launched to provide a self-service environment for data scientists.
2025-02
SAS integrates generative AI capabilities into the Viya platform for natural language querying.
2026-04
SAS launches Governed Copilots and Agent Frameworks.
閱讀原文： Computerworld

AI 週報

閱讀本週精選 AI 大事摘要 →

PostLinkedIn

AI 策展新聞聚合。所有內容版權歸原始發布者所有。
原始來源： Computerworld ↗

這是摘要，不是原文。去看原站，或訂閱每週簡報。

每週電子報

每週一封，可隨時退訂。