AI Integration for New York CompaniesWall Street Meets Madison Avenue in 2026

Goldman Sachs: AI-First Investment Bank ($47.4B Revenue, 49,100 Employees)

Goldman Sachs announced its commitment to becoming an "AI-first" investment bank in 2022, investing $4 billion over 2022-2026 to modernize technology infrastructure, hire 2,000 AI engineers and data scientists, and integrate AI across every business line.

• Algorithmic trading: AI provides liquidity across 4,000+ securities, processing 2 million trades daily and generating roughly $8.5B annually — 18% of firm revenue. Execution algorithms (VWAP, TWAP, and proprietary ML strategies) handle $4 trillion in client trades annually.
• Investment research: NLP analyzes earnings calls from 8,000+ public companies for sentiment and topic signals; computer vision and NLP process alternative data such as satellite imagery and SEC filings to extract investment signals.
• Client services: Marcus, Goldman's consumer banking arm, uses AI-powered credit scoring and automated underwriting serving 8 million retail customers, with AI handling 70% of customer service inquiries. A robo-advisory platform manages $82 billion in assets.
• Regulatory compliance: AI trade surveillance monitors 100 million trades annually for market manipulation and insider trading, while KYC/AML systems screen clients against sanctions lists and flag suspicious transactions.

The firm reports trading revenue up $1.8B (2023-2025) attributed to improved algorithms, $840M in annual savings from automation, and a 42% reduction in Marcus customer acquisition cost — a total estimated AI ROI of roughly $3.2 billion annually on the $4 billion investment, about an 80% return over three years.

JPMorgan Chase: 400+ AI Use Cases ($162B Revenue, 293K Employees)

JPMorgan operates over 400 production AI applications across retail banking, corporate and investment banking, asset management, and operations, backed by a $15 billion technology budget and 62,000 technology employees — 21% of its workforce.

• COIN (Contract Intelligence): AI reviews commercial loan agreements — 360,000 pages annually that previously required 360,000 lawyer hours — with 95% accuracy matching human lawyers, saving $180M annually in legal review costs.
• Fraud detection: Real-time ML monitors 6 billion transactions annually across 86 million customers, achieving a 94% detection rate with a 76% false-positive reduction versus rule-based systems, preventing an estimated $2.1B in annual fraud losses.
• Treasury services: Cash-flow forecasting and fraud prevention for corporate payments protect $8 trillion in annual transaction volume across wire transfers, ACH, and checks.
• AI research: A 250-person research lab partners with MIT, Stanford, Carnegie Mellon, and NYU, publishing 40+ papers annually at top ML conferences including NeurIPS, ICML, and ICLR.

Omnicom Group: AI Creative & Media Optimization ($14.3B Revenue, 71,000 Employees)

Omnicom owns creative agencies BBDO, DDB, and TBWA plus media agencies OMD and PHD, serving 5,000+ clients including Apple, PepsiCo, and McDonald's, and managing $44 billion in annual media spend. Its Omni AI platform spans three areas:

• Creative intelligence: Creative testing AI evaluates thousands of variations before launch; dynamic creative optimization generates personalized ad variants per audience; generative tools draft visuals and copy, with human creative directors art-directing and approving.
• Media buying & optimization: Programmatic AI bids on digital impressions in real-time auctions, managing $18 billion in annual programmatic spend, with cross-channel attribution and incrementality testing measuring true campaign lift.
• Audience insights & targeting: Clustering algorithms build behavioral segments beyond basic demographics, lookalike modeling expands reach, and privacy-preserving approaches (contextual targeting, cohort-based methods) maintain performance as cookies deprecate.

The New York Times: AI-Augmented Journalism (10.8M Subscribers, $2.4B Revenue)

The Times transformed from a declining print newspaper (peak 1.1 million print circulation in 1993) into a thriving digital publisher with 10.8 million digital subscribers in 2026. AI played a crucial role: personalization driving subscriptions, content optimization increasing engagement, and operational efficiency enabling newsroom investment.

• Homepage & article personalization: Every subscriber sees a homepage optimized for their interests using collaborative filtering, content-based, and contextual signals. Engagement increased 47% since personalization launched in 2021.
• Dynamic paywall: Predictive models identify likely subscribers based on engagement patterns; casual readers hit the paywall after 5 articles while high-intent readers hit it after 2, lifting conversion 31%.
• Editorial AI tools: AI monitors 100,000+ sources for story discovery, assists fact-checking by comparing claims against trusted knowledge bases, and suggests SEO-optimized headlines without sacrificing editorial standards.
• Strict editorial limits: The Times maintains firm red lines: no AI-generated reporting, no AI bylines, and no AI replacing journalists — AI assists, it doesn't author.

The results: digital subscription growth of 32% CAGR (2021-2025), churn reduced 24% through AI-personalized retention, and total digital revenue of $1.9 billion in 2025, up from $800 million in 2020.

Compass: AI-Powered Real Estate Platform (32,000 Agents, $6.4B Revenue)

Compass, founded in New York in 2012, aims to modernize residential real estate through technology — recruiting top-producing agents, providing AI-powered tools, and taking market share from traditional brokerages like Corcoran, Douglas Elliman, and Coldwell Banker.

• Automated valuation model: ML models estimate property value from comparable sales, property characteristics, and location factors, with computer vision analyzing listing photos for features like renovated kitchens and hardwood floors. Accuracy: within 5% of eventual sale price for 78% of properties, versus roughly 12% error for traditional methods.
• Buyer-property matching: A recommendation system matches buyers to properties using explicit preferences and implicit behavior, understanding qualitative terms like "modern aesthetic," with proactive alerts when a matching property hits the market.
• Market insights: Automated comparative market analysis that previously took agents hours now generates in minutes, alongside investment analysis tools calculating cash-on-cash return, cap rate, and IRR.

Top Compass agents close 40% more transactions than non-Compass agents controlling for experience and market, saving roughly 12 hours weekly on administrative work. The platform has grown market share to 4.8% of US residential transactions in 2025 (up from 2.1% in 2020), handling $220 billion in transaction volume across 32,000 agents in 45 markets, backed by a $400 million annual R&D budget and 1,200-person engineering team.

Mount Sinai Health System: Enterprise AI Deployment (8 Hospitals, $8.6B Revenue)

Mount Sinai established its Clinical Intelligence Center in 2020, centralizing AI development, deployment, and governance across the health system with a 180-person team of data scientists, ML engineers, clinical informaticists, and physicians.

• Sepsis prediction: An ML model analyzing vital signs, lab results, and medications predicts sepsis onset 12 hours before clinical symptoms, enabling early intervention that reduces mortality 19%. Deployed across all 8 hospitals, evaluated on 420,000 patients in 2025.
• Radiology AI: Computer vision analyzes X-rays, CTs, and MRIs for lung nodules, fractures, intracranial hemorrhage, and breast cancer, reducing missed findings 42% and improving turnaround time 28%.
• Pathology AI: Digital pathology with AI detects cancer in tissue samples and grades tumor severity for prostate, breast, and colon cancers, matching expert pathologists' accuracy significantly faster.
• Operational AI: Patient-flow prediction enables proactive capacity management; no-show prediction reduces missed appointments 24%; supply chain forecasting prevents stockouts of medical supplies and pharmaceuticals.

Governance is central to Mount Sinai's approach: every model undergoes prospective clinical validation before deployment, continuous monitoring of accuracy and physician override rates, explainability so physicians understand which features drove a prediction, and bias testing to ensure performance holds across demographics — since healthcare AI historically trained predominantly on white populations must be validated for generalizability.