The Legal AI Vendor Landscape
Eight archetypes are fighting for the same lawyer's attention: elites, incumbents, nichers, foundation models, agentic builders, OS plays, benchmarkers, and PM veterans.
- وقت القراءة: 12 min
ما يغطيه هذا الفصل
- The 8 Archetypes
- Why It Matters
- Where HAQQ Sits
فصول الدورة مكتوبة بالإنجليزية. أما بقية الأكاديمية فمترجمة.
TL;DR
The legal AI market is not one market. It's eight overlapping archetypes fighting for the same lawyer's attention. Knowing which bucket a vendor sits in is the first step to evaluating them. Most vendors think they're #1. Most are #5. Winners get scored by #7 - and don't flinch.
The 8 archetypes
The Elite
Legora, Harvey. Multi-billion valuations. Big-firm associate replacement plays. Optimized for white-shoe workflows.
The Incumbents
Thomson Reuters, LexisNexis, Bloomberg Law. Distribution moats. Bolting AI onto existing data and research products.
The Nichers
Specialized vertical tools - litigation, M&A, IP, immigration. Deep workflow obsession over one slice.
The Foundation Models
OpenAI, Anthropic, Google. Not legal-specific, but increasingly used directly by lawyers - a horizontal threat.
The Agentic Builders
Open-source / GitHub-first projects. Composable agents, prompt libraries, transparent stacks. Lawyer-developer hybrids.
The OS Plays
Platforms touching multiple layers - model, data, workflow, billing - and not boxed into a single category. HAQQ sits here.
The Benchmarkers
legalbenchmarks.ai, Vals AI / VLAIR. No vendor money. Practitioner-led scorecards. Becoming the gate between vendors and buy-side counsel.
The Practice Management Veterans
Clio, MyCase, PracticePanther. Old PM stacks racing to add AI before AI-native eFirm products eat them.
Why this taxonomy matters to you
When a vendor demos to you, ask which bucket they're in. They will dodge the question. Their answer (or non-answer) tells you the truth.
Elites sell prestige. Incumbents sell distribution. Nichers sell depth. Foundation models sell raw capability. OS plays sell unification. Benchmarkers don't sell - they grade.
HAQQ doesn't fit one bucket cleanly. We're an OS - touching the model layer, the incumbents' data, the nichers' workflows, built in public. The course is written from that vantage point.