AI in Education, Legal, Retail, and Real Estate

Learning Objectives

• Identify the leading AI applications in education, legal, retail, and real estate
• Explain how AI personalizes learning at scale and what the evidence says about its effectiveness
• Understand how AI is changing legal research and why it raises distinctive professional responsibility questions

Education: AI Tutors and Adaptive Learning

Education is one of the highest-potential domains for AI — and one where the gap between potential and evidence is worth understanding carefully. The promise: every student gets a personal tutor with infinite patience, perfectly calibrated to their current understanding. The reality is more nuanced, but genuinely promising.

Khan Academy: Khanmigo

Khan Academy is the most widely used free educational platform in the world, with over 150 million registered users. Its AI tutoring system, Khanmigo, is built on top of GPT-4 and provides a conversational tutor available across Khan Academy's entire curriculum. Khanmigo for Teachers has expanded to 180+ countries through a Microsoft partnership (December 2025), with LMS integrations for Canvas, Google Classroom, and Schoology.

Khanmigo's design philosophy is deliberately Socratic — it doesn't give students answers, it asks guiding questions to help students reach answers themselves. The research rationale: passive consumption of answers doesn't produce learning; active problem-solving does.

For teachers, Khanmigo provides lesson planning assistance, parent communication drafts, rubric creation, and administrative tasks — reducing the non-teaching workload that contributes to teacher burnout.

Sal Khan's perspective: The founder of Khan Academy has argued that Khanmigo represents the arrival of the "two-sigma benefit" — the well-documented finding that one-on-one tutoring produces students two standard deviations above the average of classroom-taught students — now made available to every student with internet access.

Duolingo: AI for Language Learning at Scale

Duolingo (52.7 million daily active users as of Q4 2025, the most downloaded education app globally) has integrated AI throughout its language learning experience.

Duolingo Max features:

• Explain My Answer: After getting a question wrong, learners can ask Duolingo's AI to explain exactly why their answer was incorrect and how the grammar rule works — instant personalized explanation at scale
• Roleplay: AI-powered conversation practice — learners can have real-time conversations in their target language with an AI character in contextual scenarios (ordering at a café, discussing a job application)
• Video Call with Lily: AI-powered video conversations with Duolingo characters, backed by research showing measurable speaking improvement

Duolingo has also used AI significantly to expand beyond languages into chess, K-12 math, and music — reducing the cost of creating new courses from years of expert labor to months. The company exceeded $1 billion in bookings for the first time in FY2025.

Synthesis: Accelerated Learning

Synthesis (founded by SpaceX educators, originally an accelerated learning program for children of SpaceX employees) uses AI to create collaborative, game-based problem-solving challenges that develop reasoning and mathematical thinking.

Synthesis claims significantly accelerated math learning outcomes — students working through Synthesis's AI-adaptive curriculum progress faster than the national average on standardized assessments. The company has made its platform widely available following its origins as an exclusive internal program.

Legal: AI Research and Document Intelligence

The legal industry is one of the most data-intensive professions — lawyers spend enormous amounts of time researching case law, reviewing documents, drafting contracts, and analyzing large volumes of legal text. AI is transforming each of these workflows.

Harvey AI: The Legal LLM

Harvey AI (founded 2022, backed by OpenAI and Sequoia) is the most prominent AI-native legal technology company — valued at $8 billion as of December 2025 (with talks at $11 billion in early 2026), with estimated ARR of $195 million. Harvey provides LLM-powered tools specifically designed for legal workflows:

• Legal research: Answering legal questions with citations to relevant case law and statutes — faster and more comprehensive than traditional Boolean search
• Contract analysis: Reviewing contracts to identify provisions, risks, non-standard terms, and deviations from standard positions
• Document drafting: Generating first drafts of legal documents based on natural language specifications
• Due diligence: Reviewing large volumes of documents in M&A transactions to identify issues that require attorney attention
• Shared Spaces: Collaborative AI platform for attorneys working on the same matter

Harvey is deployed at 50 of the top AmLaw 100 firms, reaching approximately 100,000 lawyers across 700+ firms and enterprises — including A&O Shearman, which made Harvey available to all of its attorneys globally.

Thomson Reuters Westlaw and Lexis+ AI

Thomson Reuters Westlaw and LexisNexis are the two dominant legal research platforms, both used by virtually all US law firms and in-house legal departments.

Both platforms have significantly upgraded their AI capabilities, moving toward agentic workflows:

• Westlaw Advantage (launched August 2025, described as the "final" version of Westlaw) with CoCounsel Legal — featuring agentic AI and Deep Research capabilities where AI agents autonomously employ Westlaw's proprietary tools (Key Numbers, KeyCite) to conduct comprehensive legal research
• Protege General AI (LexisNexis, launched August 2025) — featuring 4 specialized agents (orchestrator, legal research, web search, customer document), with access to multiple general-purpose AI models alongside legal-specific AI, plus voice AI assistant capabilities

The integration of AI into Westlaw and Lexis is significant because it brings AI research to the platforms where lawyers already work — no change in workflow is required.

AI in Contract Management

Ironclad and ContractPodAi apply AI to contract lifecycle management — the process of creating, negotiating, executing, and tracking contracts. These platforms use AI to:

• Extract key terms and dates from executed contracts automatically
• Flag contract provisions that deviate from standard positions
• Identify renewal deadlines and obligations across large contract portfolios
• Assist in drafting and negotiation with AI-powered redlining suggestions

For large organizations with thousands of vendor, customer, and partner agreements, AI contract management provides visibility that was previously impossible with manual review.

Retail: Personalization, Forecasting, and Last-Mile AI

Retail was among the earliest industries to adopt machine learning at scale — Amazon's recommendation engine has been influential since the early 2000s. Today, AI in retail spans the full value chain from demand forecasting to last-mile delivery.

Amazon: The AI-Native Retailer

Amazon built AI into the foundation of its retail business before "AI" became a mainstream term. Key AI systems:

Recommendation engine: Amazon's "customers who bought this also bought" recommendations are estimated to drive 35% of total revenue — one of the most commercially successful AI applications in history.

Dynamic pricing: Amazon adjusts prices millions of times per day based on competitor prices, inventory levels, demand patterns, and customer behavior — algorithmic pricing at a scale impossible for human pricing teams.

Demand forecasting and inventory: AI predicts what inventory will be needed in each fulfillment center, when to reorder from suppliers, and how to allocate inventory across the network to minimize delivery time and cost.

Amazon Go (Just Walk Out): Computer vision and sensor fusion allow customers to pick up items and walk out without checking out — AI tracks what each customer takes and charges them automatically. In January 2026, Amazon announced the closure of its remaining Amazon Go and Amazon Fresh stores, citing no scalable economic model for company-owned retail. However, the underlying Just Walk Out technology has expanded dramatically — now licensed to 360+ third-party locations across 5 countries (airports, stadiums, convenience stores), proving that the AI technology is commercially viable even when Amazon's own retail format was not.

Alexa AI: The voice AI assistant integrated into Amazon Echo devices handles shopping, smart home control, and information queries — with ongoing integration of large language model capabilities.

Stitch Fix: AI-Powered Personal Styling

Stitch Fix applies AI to personal styling — using machine learning to predict which clothing items a specific customer will like and keep, based on their style profile, purchase history, and feedback on previous selections.

Stitch Fix's algorithm selects a curated box of clothing for each customer; human stylists review and adjust the selection. The AI handles the initial filtering from a catalog of thousands of items to a shortlist; the human stylist adds the judgment and personal touch the AI cannot replicate. In October 2025, Stitch Fix launched Stitch Fix Vision — a generative AI virtual try-on feature where 75% of users return monthly and Freestyle purchases increased over 100%.

This human-AI collaboration model has produced a retail business with remarkably low return rates compared to traditional e-commerce — because the selection is personalized to each customer's actual preferences. Stitch Fix's revenue grew 9.4% year-over-year in Q2 FY2026, marking its fourth consecutive period of growth.

Real Estate: Predictive Analytics and AI Valuation

Real estate has historically been a data-intensive but information-inefficient industry — valuations driven by appraiser judgment, market data difficult to access, and transactions slow and opaque.

Zillow and AI Property Valuation

Zillow's Zestimate is one of the most widely known AI models in consumer applications — a machine learning model that estimates the value of virtually every home in the United States.

Zestimate processes hundreds of data points: property characteristics, recent comparable sales, local market trends, tax assessment history, school district quality, and satellite imagery analysis of property condition. The model achieves national median error rates under 3% for on-market homes — more accurate than many professional appraisals.

Zillow also uses AI for its iBuyer program (when active) — making instant cash offers to homeowners based on algorithmic valuations, then reselling. This program demonstrated the commercial power and limits of AI valuation at scale: Zillow's iBuyer lost hundreds of millions of dollars when its pricing models failed to anticipate market downturns.

Procore: AI for Construction

Procore (the dominant construction project management platform) integrates AI through its Procore Helix platform:

• Safety analysis: Computer vision analysis of jobsite photos to identify safety hazards — workers not wearing PPE, unsafe scaffolding configurations, blocked emergency exits
• Cost and schedule prediction: AI analysis of project data to identify risks of cost overruns or schedule delays before they materialize
• Document processing: Extracting information from construction documents (plans, submittals, RFIs) at a speed and volume impossible for human review
• Agent Builder (launched October 2025): A platform for creating custom AI agents via natural language, with pre-built agents including an RFI Creation Agent

Procore acquired Datagrid (a vertical AI firm) in January 2026 to accelerate its AI strategy. Construction is one of the least digitized major industries — which means both that AI has significant potential impact and that adoption is slower than in more tech-native sectors.

Key Takeaways

• Education AI (Khan Academy's Khanmigo, Duolingo Max, Synthesis) is making personalized, adaptive learning accessible at scale — with Khanmigo specifically aiming to deliver the well-documented "two-sigma" tutoring benefit to all students
• Legal AI (Harvey at $8 billion+ valuation with 100K lawyers, Westlaw Advantage with CoCounsel Legal, LexisNexis Protege General AI) is automating research, contract analysis, and document drafting with agentic workflows — but AI hallucinations in legal contexts carry professional responsibility consequences that require verification
• Retail AI is most advanced at Amazon — its recommendation engine, dynamic pricing, and demand forecasting remain some of the most commercially successful AI deployments; Amazon Go stores closed but Just Walk Out technology expanded to 360+ third-party locations
• Real estate AI (Zillow Zestimate) has made property valuations more accessible and transparent, but Zillow's iBuyer loss demonstrated the limits of AI models in volatile real estate markets
• Construction AI (Procore) targets one of the least digitized major industries — computer vision for safety, cost and schedule prediction, and document processing