SetupAI
繁
SourceStalecollected in 21m

Goldman Sachs on data in the age of AI

Read original on Bloomberg Technology
#data-engineering#data-governance#enterprise-ai

Learn how a top-tier financial institution structures data pipelines to support enterprise-scale AI initiatives.

30-Second TL;DR

What Changed

Data engineering is the foundation for AI readiness

Why It Matters

Emphasizes that enterprise AI success is gated by data infrastructure rather than just model selection. Organizations must prioritize data governance to remain competitive.

What To Do Next

Audit your current data pipeline architecture to identify bottlenecks that could hinder large-scale model training or inference.

Who should care:Enterprise & Security Teams

Key Points

  • •Data engineering is the foundation for AI readiness
  • •Quality of data determines the efficacy of AI models
  • •Strategic data management is required for enterprise AI adoption

Deep Insight

Background and context from public sources — not the original article. 15 sources cited.

Enhanced Key Takeaways

  • •Goldman Sachs has adopted an engineering-led approach to data, treating it as a first-class asset with a focus on platform engineering, content curation, and robust governance to support firm-wide functions from client service to AI.
  • •The firm has open-sourced its data management system, Legend, to foster data interoperability across the financial industry, and integrates it with cloud platforms like Google Cloud's BigQuery and BigLake for enhanced data modeling and analytics.
  • •Goldman Sachs has deployed a secure, multi-model internal AI platform (GS AI Platform) that operates behind its corporate firewall, leveraging leading language models such as OpenAI's GPT-4, Google's Gemini, and Anthropic's Claude, and utilizing Retrieval Augmented Generation (RAG) for secure querying of private data.
  • •Neema Raphael, Goldman Sachs' Chief Data Officer, emphasizes that the world is facing a shortage of 'organic data' for AI training, underscoring the critical need for better utilization of existing enterprise data and the potential emergence of synthetic data markets.
  • •Goldman Sachs is actively piloting 'agentic AI' (AI 2.0) for autonomous tasks, including using AI software engineers like Devin for automated unit test generation, which has already shown a 180x speed increase in some coding tasks and a 20% boost in overall coding speed.

Technical Deep Dive

  • Goldman Sachs' core data architecture is being re-platformed onto AWS to achieve scalability and increased operational speed.
  • The firm is developing a modern Lakehouse and AI data platform to enable reliable, governed, and high-performing data use.
  • Legend, Goldman Sachs' open-source data management system, integrates with Google Cloud services such as BigQuery and BigLake for data modeling and analytics.
  • The internal GS AI Platform is a secure, multi-model ecosystem operating behind the corporate firewall, employing Retrieval Augmented Generation (RAG) to allow AI to search and use information from private databases like internal compliance policies and transaction records.
  • For software development, Goldman Sachs utilizes AI tools like Diffblue Cover for automated unit test generation for legacy Java code, significantly improving code coverage and reducing development time.
  • The Marquee platform provides programmatic access to proprietary and third-party data via APIs and offers GS Quant, a Python toolkit for quantitative finance, enabling advanced analytics and backtesting.
  • The firm is exploring agentic AI, which involves autonomous agents capable of executing complex, multi-step tasks independently, moving beyond passive assistance tools.

Future ImplicationsAI analysis grounded in cited sources

Agentic AI will fundamentally redefine enterprise workflows and job roles in finance.
Goldman Sachs is actively piloting autonomous AI agents for complex tasks, expecting them to act as 'virtual employees' and significantly boost productivity, shifting human roles towards planning and oversight.
The financial industry will see an acceleration in the development of data marketplaces and synthetic data generation.
Neema Raphael notes the world is 'running out of organic data' for AI training, suggesting a growing need for new data sources, including synthetic data and structured enterprise data, which could lead to emerging data markets.
Data governance and security will become even more critical and complex with the widespread adoption of multi-model and agentic AI systems.
Goldman Sachs' approach emphasizes strict governance and a secure internal environment for its multi-model AI platform, indicating that managing data privacy, provenance, and security for increasingly autonomous AI will be a paramount challenge and focus.

Timeline

2003
Neema Raphael joins Goldman Sachs in the Technology Division.
2013
Neema Raphael named Managing Director at Goldman Sachs.
2017-11
Harvard Business School case study highlights Goldman Sachs' digital transformation efforts, including the Marquee platform.
2020
Goldman Sachs open-sources its Legend data management system and Neema Raphael becomes Chief Data Officer and Head of Data Engineering.
2025-06
Goldman Sachs rolls out its proprietary generative AI tool, the GS AI Assistant, firmwide to all 46,000+ employees.
2026-03
Goldman Sachs details its strategic infrastructure for integrating generative AI, including piloting autonomous AI software engineers and reporting a 20% boost in coding speed.
Read original article on Bloomberg Technology

Weekly AI Recap

Read this week's curated digest of top AI events →

PostLinkedIn

AI-curated news aggregator. All content rights belong to original publishers.
Original source: Bloomberg Technology ↗

This is a summary, not the original. Read the source, or get the weekly briefing.

The weekly digest

One email a week. Unsubscribe anytime.