AI in Legal Tech: Contract Review, Legal Research, and the Automation of a $1 Trillion Industry

The $1 Trillion Industry Meets AI

The global legal services market exceeds $1 trillion in annual revenue — approximately $1.03 trillion in 2025, according to Precedence Research — dominated by law firms, corporate legal departments, and a vast ecosystem of courts, regulators, and legal aid organizations. It is an industry built on language — contracts, statutes, case law, briefs, opinions — which makes it theoretically ideal for large language models. Yet until recently, legal tech adoption was notoriously slow. Mid-sized law firms dedicated roughly 2% of their expenses to technology, and even expert recommendations topped out at 7% of revenue, well below the 7-10% typical in financial services or 5-6% in healthcare.

That resistance is crumbling — fast. Legal tech funding in 2025 reached $5.99 billion, featuring fourteen rounds of $100 million or more. Harvey, the legal AI startup co-founded by Winston Weinberg (a former O’Melveny & Myers litigation associate) and Gabriel Pereyra (a former DeepMind and Google Brain research scientist), has become the sector’s most prominent company. Backed by Sequoia, OpenAI, Kleiner Perkins, and Andreessen Horowitz, Harvey closed its Series F at an $8 billion valuation in December 2025 after raising over $800 million across four funding rounds in a single year. By February 2026, it was reportedly in talks for an $11 billion valuation. Its clients include A&O Shearman (where 4,000 staff across 43 jurisdictions use the platform), PwC, and several Magic Circle firms. In June 2025, Harvey and LexisNexis announced a strategic partnership — called the most important legal tech move in a decade by industry analyst Richard Tromans — integrating Harvey’s AI with LexisNexis’s primary law content and Shepard’s Citations.

Luminance, a UK-based legal AI company founded by machine learning researchers from the University of Cambridge, has deployed its contract analysis platform across more than 1,000 organizations in 70 countries, including all Big Four consultancy firms and over a quarter of the Global Top 100 law firms. Thomson Reuters integrated AI across its Westlaw and Practical Law platforms, launching CoCounsel Legal with agentic AI and deep research capabilities in August 2025. LexisNexis launched its Protege AI assistant the same month, with a major next-generation update in December 2025 supporting multiple AI models including Claude and GPT-5. The message from the market is clear: AI legal tech has moved from experimental to essential.

The driver is economics. A junior associate at a major US law firm now bills $600-$1,000 per hour — at elite firms like Paul Weiss and Sullivan & Cromwell, first-year associates approach $900-$1,000. A significant portion of that associate’s time — estimates range from 30% to 60% — is spent on tasks that AI can now perform: document review, contract analysis, legal research, due diligence, and drafting routine documents. If AI can perform these tasks at a fraction of the cost with comparable accuracy, the business case is overwhelming. The question is not whether AI will transform legal work but which tasks it will handle, and what happens to the humans who currently perform them.

What AI Can Reliably Automate

Contract review is the most mature legal AI application. A commercial contract — a vendor agreement, a lease, an NDA — contains standard clauses (indemnification, limitation of liability, termination, governing law) that AI can identify, extract, and compare against a company’s preferred positions in seconds. Kira Systems (acquired by Litera in 2021, with expanded hybrid Gen AI capabilities launched in January 2026), Luminance, and Ironclad use machine learning to read contracts, flag deviations from standard terms, identify missing clauses, and generate redline suggestions. Ironclad surpassed $200 million in annual recurring revenue in early 2026, with its AI assistant Jurist delivering six-fold year-over-year revenue growth. In due diligence scenarios — an M&A transaction involving thousands of contracts — AI review that would take a team of associates weeks can be completed in hours.

Legal research, traditionally performed using keyword searches on Westlaw or LexisNexis, has been transformed by AI that understands semantic meaning. Harvey and CoCounsel Legal (Thomson Reuters’ AI tool, launched in August 2025 with agentic workflows) can answer complex legal questions — “What is the standard for piercing the corporate veil in Delaware?” — by retrieving and synthesizing relevant case law, statutes, and secondary sources. The AI does not just find documents; it reads them, extracts the relevant holdings, and presents a synthesized answer with citations. CoCounsel Legal’s Deep Research feature can now conduct comprehensive multi-step legal analysis autonomously, while its bulk document review capability handles up to 10,000 documents. LexisNexis’s next-generation Protege deploys four specialized agents — an orchestrator, legal research agent, web search agent, and customer document agent — collaborating on complex workflows.

Document drafting is a growing application. AI can generate first drafts of routine legal documents — employment agreements, corporate bylaws, privacy policies, demand letters — from templates and instructions. Spellbook, which raised a $50 million Series B in October 2025 at a $350 million valuation, has surpassed 10 million contracts reviewed across nearly 4,000 law firms and legal teams in 80 countries. Its Library feature, launched in July 2025, lets lawyers power AI with their own precedents and past work. Spellbook Associate, the company’s AI agent for multi-document transactional drafting, represents the next frontier: AI that handles not just single clauses but entire deal workflows.

The shift from individual AI tools to agentic AI systems is the defining trend of 2025-2026 legal tech. A&O Shearman and Harvey are jointly launching agentic AI agents for antitrust filing analysis, cybersecurity, fund formation, and loan review — tools that will be sold to other firms and corporate clients. Ironclad introduced five specialized AI agents (Manager, Review, Drafting, Editing, and Research) that work in concert across the contract lifecycle. These systems do not just answer questions; they execute multi-step legal tasks with minimal human intervention, and the industry is rapidly moving from AI as an assistant to AI as a capable junior team member.

What AI Cannot Do (Yet)

Courtroom advocacy remains a distinctly human domain. Trial work requires reading a jury, adjusting strategy in real time based on witness testimony, making split-second objection decisions, and projecting credibility and authority. These are embodied, social, and improvisational skills that current AI cannot approximate. Even the most advanced AI systems lack the ability to stand before a judge, build rapport with jurors, or navigate the unpredictable dynamics of a courtroom.

Client counseling — advising a CEO on the legal implications of a strategic decision, guiding a family through a divorce, counseling a criminal defendant on plea options — requires empathy, judgment, and an understanding of the client’s broader life context that goes far beyond legal analysis. A contract AI can identify that a non-compete clause is enforceable in Texas but not California. It cannot advise a client on whether to sign the contract given their career goals, family situation, and risk tolerance. The human lawyer integrates legal analysis with personal judgment in ways that AI does not replicate.

Legal strategy — deciding which arguments to make, which precedents to rely on, how to structure a transaction, when to settle versus litigate — requires creative reasoning that combines legal knowledge with business acumen, negotiation psychology, and an understanding of how specific judges or opposing counsel are likely to behave. While AI can provide data to inform strategy (case outcome prediction models, judicial analytics), the synthesis of this information into strategic decisions remains human.

The highest-profile AI failure in legal tech reinforced these limits. In 2023, in Mata v. Avianca, attorney Steven Schwartz submitted a brief to the Southern District of New York containing citations to cases that did not exist — they were hallucinated by ChatGPT, which even confirmed to Schwartz that the fabricated cases could be found on Westlaw and LexisNexis. Judge P. Kevin Castel imposed a $5,000 sanction under Rule 11 for subjective bad faith. Since then, the legal profession has responded with guardrails. In July 2024, the American Bar Association issued Formal Opinion 512 — its first formal ethics guidance on lawyers’ use of generative AI — addressing duties of competence, confidentiality, communication with clients, candor toward tribunals, and supervisory responsibilities. Every major legal AI platform has implemented citation verification, and the ABA’s 2025 AI Task Force report noted rapid increases in state-level guidance across jurisdictions. But the Mata case underscored an enduring principle: AI in law requires human oversight, and the lawyer remains responsible for every statement in every filing.

Access to Justice: AI’s Democratic Promise

Perhaps the most compelling argument for AI legal tech is not about corporate efficiency but about access to justice. In the United States, 92% of the civil legal problems of low-income Americans receive inadequate or no legal help, according to the Legal Services Corporation’s 2022 Justice Gap study — up from the already dire 80% figure in earlier reports. Nearly three-quarters of low-income households experienced at least one civil legal problem in the previous year, and LSC-funded organizations must turn away one out of every two requests due to limited resources. Globally, 5.1 billion people lack meaningful access to justice, according to the World Justice Project — including 1.5 billion who cannot obtain justice for specific civil, administrative, or criminal problems, and 4.5 billion who lack fundamental legal protections like identity documents or formal land tenure.

AI-powered legal tools could democratize access to legal services, though not without growing pains. DoNotPay, the self-described “robot lawyer” app, attracted millions of users for consumer legal actions like parking ticket appeals and subscription cancellations. But in early 2025, the Federal Trade Commission finalized a 5-0 order against the company for deceptive claims, finding that DoNotPay had never tested whether its AI operated at the level of a human lawyer and had not retained attorneys to verify the quality of its legal output. The company was fined $193,000 and barred from advertising lawyer-equivalent capabilities without evidence. The episode illustrated both the demand for affordable legal help and the risks of overpromising what current AI can deliver.

More measured approaches are showing results. Chatbot-based legal aid services — including pilots by the Legal Services Corporation and the American Bar Association — provide guidance on housing disputes, immigration applications, and family law matters. In the UK, the Judiciary’s online court initiatives integrate AI to help litigants in person understand procedures and complete forms correctly. The Association of Corporate Counsel reported in 2025 that corporate legal AI adoption more than doubled in a single year, jumping from 23% to 52%, with 64% of in-house teams expecting to depend less on outside counsel as a result.

The potential in developing countries is enormous. In Algeria, for example, the ratio of lawyers to population is significantly lower than in Europe, legal aid is limited, and many citizens navigate bureaucratic and legal processes without professional assistance. An AI-powered legal assistant that could explain administrative procedures, generate complaint letters, review lease agreements, or guide users through court filing requirements — in Arabic and French — could meaningfully improve access to justice. The technology exists; the challenge is localization, trust, and integration with existing legal frameworks.

Corporate legal departments are also benefiting from democratized AI tools. Companies that previously could not afford outside counsel for routine legal matters — contract review, compliance checks, employment law questions — can now use AI to bring some legal capabilities in-house. This is particularly relevant for small and medium enterprises, which constitute the vast majority of businesses globally and in Algeria, and which have historically operated without legal review of their contracts and compliance obligations.

Sources & Further Reading

• Harvey Raises $300M Series D Led by Sequoia — Harvey
• Legal AI Startup Harvey Confirms $8B Valuation — TechCrunch
• Thomson Reuters Launches CoCounsel Legal with Agentic AI — PR Newswire
• LexisNexis Unveils Next-Generation Protege General AI — LawSites
• Luminance About Us — Luminance
• Legal Tech Raised $6Bn in 2025 as AI Boom Shows Divisions — Artificial Lawyer
• FTC Finalizes Order Prohibiting Deceptive AI Lawyer Claims by DoNotPay — FTC
• ABA Formal Opinion 512: Generative AI in Legal Practice — ABA
• The Justice Gap Report — Legal Services Corporation
• Measuring the Justice Gap — World Justice Project
• Ironclad Surpasses $200M in ARR — PR Newswire
• ACC GenAI Report: Corporate Law Department AI Use More Than Doubled — ACC