Other AI Case Studies: Legal, Education, HR, Logistics, and More

AI adoption is not confined to the sectors covered in earlier lessons. Across legal, education, HR, real estate, logistics, agriculture, cybersecurity, and customer service, documented deployments with measurable outcomes are accumulating rapidly. This lesson surveys each area, anchored to specific named platforms and statistics from 2024–2025.

**Legal Technology.** Contract review and e-discovery have been transformed by AI. Luminance, a UK-based legal AI, is deployed at over 700 law firms and legal teams worldwide and processes contracts in dozens of languages, flagging non-standard clauses and anomalies in minutes rather than hours. Harvey AI, which raised $100 million at a $1.5 billion valuation in 2024 and is backed by OpenAI, is used by firms including PwC Legal, A&O Shearman, and Macfarlanes for drafting, research, and matter summarisation. Casetext's CoCounsel (acquired by Thomson Reuters for $650 million in 2023) automates legal research, deposition preparation, and document review. A 2024 Stanford Law School study found that lawyers using AI research tools completed research tasks 34% faster with no statistically significant difference in accuracy compared to unassisted research — a rare apples-to-apples comparison.

**Education.** Khan Academy's Khanmigo, powered by GPT-4, provides Socratic tutoring across math, science, and humanities — guiding students toward answers rather than providing them directly, consistent with learning science research on active recall. Duolingo's AI features, including AI-generated conversation practice and personalised difficulty adjustment, contributed to a 42% year-over-year increase in daily active users in 2024. Carnegie Learning's AI math tutoring platform, used in over 3,000 US school districts, shows consistent evidence of accelerating student progression through curriculum — particularly for students who were previously performing below grade level. The major challenge: AI tutors perform best for motivated students with reliable internet access, which means they risk widening rather than closing equity gaps if deployed without intentional access strategies.

**HR and Recruiting.** LinkedIn's AI recruiting tools — including AI-generated job descriptions, candidate matching, and InMail drafting — are used by over 90% of Fortune 500 companies. Workday's AI capabilities include skills inference (identifying employee skills from work history and performance data without requiring explicit self-reporting), career pathing recommendations, and attrition risk scoring. The HR AI sector carries significant regulatory scrutiny: the EEOC has issued guidance that AI hiring tools must comply with Title VII anti-discrimination requirements, and the city of New York requires bias audits for any automated employment decision tool. HireVue and other video interview AI vendors have faced public criticism over facial expression analysis features, with several quietly removing those capabilities in response.

**Real Estate.** Zillow's Zestimate AI valuation model covers over 100 million US homes and is within 2% of sale price for roughly 50% of listings. Opendoor uses AI pricing models to make instant cash offers on homes, then resell them — a model that demonstrated both the power and the risk of AI valuation when Opendoor reported a $928 million loss in Q3 2022 after its models underestimated post-pandemic price corrections. The company rebuilt its models with stronger macro-economic signal weighting and returned to profitability in 2024. AI listing description generation is now mainstream, with platforms like Compass and Realtor.com generating first-draft property descriptions from structured data. Commercial real estate is adopting AI for lease abstraction (extracting key terms from thousands of pages of lease documents) at significant scale through platforms like Kira and Leasecake.

**Logistics and Supply Chain.** UPS's ORION (On-Road Integrated Optimization and Navigation) system, deployed since 2012 and continuously improved with ML, saves the company an estimated 100 million miles of driving per year — roughly $400 million in fuel and time savings annually. DHL has deployed over 10,000 collaborative robots across its warehouses as of 2024, with AI-driven picking and sorting increasing throughput by 25% at automated facilities. Project44 and FourKites provide real-time supply chain visibility using ML to predict shipment arrival times, integrating signals from carriers, ports, weather, and traffic — helping shippers reduce safety stock requirements and improve customer communication.

**Agriculture.** John Deere's See and Spray system uses computer vision to identify weeds versus crops at planting speed, applying herbicide only to weeds — reducing herbicide use by up to 77% in field trials. John Deere's fully autonomous tractor, launched commercially in 2022 and expanded in 2024, uses AI to navigate fields, plant seeds, and manage soil compaction without a driver. Climate Corporation (owned by Bayer) provides field-level crop yield predictions that help farmers optimise planting decisions and insurance companies underwrite crop insurance more accurately. Precision agriculture AI is concentrated among large commercial farms; smallholder adoption remains low due to cost and connectivity barriers.

**Cybersecurity.** Darktrace's AI models learn normal network behaviour and flag anomalies in real time — the system detected the SolarWinds-style lateral movement patterns before signatures were available for the specific malware. CrowdStrike's Falcon AI platform processes over 2 trillion security events per week and uses ML to detect novel threats without relying on known-malware signatures. Microsoft Security Copilot, launched in 2024, uses GPT-4 to help analysts triage alerts, summarise incidents, and generate remediation guidance in natural language. The cybersecurity AI challenge: adversaries are also using AI — spear phishing emails generated by LLMs are now nearly indistinguishable from human-written ones, and AI is being used to discover software vulnerabilities at scale.

**Customer Service.** Intercom's Fin AI agent, launched in 2023, resolves over 50% of customer support queries without human intervention for typical SaaS companies — the company publishes resolution rates by industry, ranging from 28% for complex technical products to 65% for straightforward consumer services. Zendesk AI triage and routing features are deployed across thousands of support teams, automatically classifying tickets by issue type, intent, and sentiment — reducing average handle time by 15–20% in published case studies. Klarna's AI assistant, built on OpenAI, handles the equivalent of 700 full-time customer service agents and resolved 2.3 million conversations in its first month of full deployment, with customer satisfaction scores matching the human team.