Skip to main content

Legal Industry Platform

What does the legal industry operate on — and what is changing underneath it?

This page is the industry-level view of the tech, data, and capital assets that power legal services. The function-level twin — what a single business's GC needs to run their own loop — sits at Legal Operations Platform.

Three Layers — Industry Stack​

The legal industry runs on three stacked layers. Each is changing fast, but at different speeds.

  • System of record — Court filing systems (PACER + state equivalents), regulatory registries, IP databases (USPTO, EUIPO, WIPO), CLM platforms inside firms, conflict-clearance databases, counsel-marketplace registries. Mostly proprietary; mostly fragmented per jurisdiction; mostly slow to evolve.
  • Agent layer — AI workers that draft, redline, monitor, research, summarise. Wave 1 (closed proprietary) and Wave 2 (open infrastructure) compete here. The agent layer is where the AI yield accumulates and where the value capture fight is happening today.
  • On-chain instruments — Emerging crypto rails: smart contracts as agreement infrastructure, IP NFTs as priority proofs, DAO wrappers, on-chain attestations, tokenised cap tables, stablecoin settlement for legal fees and escrow.

Assets the Industry Operates On​

Platform broadens past tech. Industries also operate on physical, data, and IP capital.

Data assets​

  • Case law corpora — Westlaw + Lexis own the largest indexed corpora. Public-source case data is fragmenting and being re-aggregated by new entrants. The largest moat any incumbent holds.
  • Statute + regulation corpora — Public-source; private indexing. Lower moat than case law because the source is universally available.
  • Filing + court records — PACER (US federal) + state equivalents. Fragmented; access-controlled; emerging analytics layer.
  • Firm playbooks (proprietary per firm) — The contracts a firm has negotiated; the redlines they accept and reject; the positions they hold. The data moat that survives commoditisation of the source-corpus layer.
  • Contract corpora (proprietary per firm) — The actual signed contracts. Subject to confidentiality and privilege; firm-internal AI training corpus.
  • Counsel network + conflict-clearance — Currently fragmented across firms; emerging marketplace + protocol layer to make this queryable.

Physical + capital assets​

  • Office real estate (BigLaw + MidLaw) — Significant carrying cost; AI yield questions whether the footprint should shrink as productivity rises.
  • Document-management infrastructure — Historically on-prem; migrating to cloud + self-hosted hybrid as confidentiality requirements clarify.
  • E-discovery hosting capacity — One of the largest operating-cost items for litigation-heavy firms. Where the cloud + AI migration story is most advanced.

IP assets​

  • Firm playbooks (as IP) — Treated as trade secrets; the asset that AI-augments most directly.
  • Practice-area methodologies — How the firm runs M&A diligence, securities offerings, regulatory engagement. Codifiable; therefore AI-replicable; therefore subject to commoditisation pressure.
  • Brand + reputation — The asset that endures as everything else commoditises. Built over decades; lost in one bad outcome.

The AI Vendor Landscape​

The agent layer is where the industry's AI investment is concentrated. Two waves competing for the same buyer.

Wave 1 — closed proprietary​

Vendor-locked platforms with premium pricing, opaque benchmarks, enterprise-only deployments.

  • Harvey — BigLaw AI assistant. $5B+ valuation. Defines the wave-1 platform: deep integration, premium pricing, closed model architecture.
  • Spellbook — Contract drafting + redlining inside Microsoft Word. SMB-focused; subscription model; closed.
  • Lexis+ AI — AI-augmented legal research from the incumbent corpus owner. Combines the existing research moat with generative output.
  • Westlaw Precision — Thomson Reuters' equivalent. Same play.
  • Everlaw / Relativity aiR — E-discovery with AI. Mature; expensive; effective.
  • BigLaw-internal builds — Most AmLaw-100 firms have an internal AI platform team. Some published; most not. Closed by default.

Wave 2 — open infrastructure​

Self-hostable, composable, multi-model, transparent.

  • MikeOSS — Open-source legal document assistant. Next.js + Express + Supabase + multi-LLM (Anthropic / Gemini / OpenAI). Self-hosted; bring-your-own-key; clear schema. The wave-2 stance is encoded in its satirical benchmark article — a deliberate send-up of vendor-curated leaderboard culture.
  • case.dev — Legal-AI infrastructure platform: compliance, jurisdiction, audit, firm playbooks. "Operating system for the agent era". Powers CaseMark Workspace, CaseMark for Court Reporters, Structured Summaries. Used by AmLaw firms + medical-malpractice insurance carriers. The infrastructure layer wave-1 vendors did not build.
  • OpenLaw — Smart-legal-contract tooling bridging code-as-law and traditional contract language.
  • Plus a long tail of open-weight models (Llama-class) plus jurisdiction-specific open datasets.

The wave-1-vs-wave-2 split mirrors the cloud-infrastructure story of 2010–2015. Vendor-locked wins early on enterprise relationships; open + composable wins long-term on cost, transparency, and developer adoption. Legal AI is in the analogous transition.

Crypto Rails — Emerging Industry Instruments​

Track. Selective bets where the regulatory posture allows.

PrimitiveWhat it changes for the industryStatus
Smart contractsCodifiable terms self-execute; courts not required for enforcementLive in finance + commerce; growing services
IP NFTsCryptographic timestamp of creation; transferable evidence of priorityEmerging; not yet primary evidence
DAO + wrapper entitiesOn-chain governance with traditional legal personalityWyoming, Marshall Islands, Cayman, Swiss active
On-chain attestationsCompliance status provable without paper trailEarly; pilots underway
Stablecoin settlementCross-border legal fees + escrow without correspondent-bank frictionLive in finance; legal-fee use early
Tokenised equity / cap tableReal-time cap table; programmable vesting + restrictionsLive in some jurisdictions; restricted in others
Zero-knowledge identityCounsel-marketplace + conflict-clearance without revealing client identityEarly experiments
Verifiable credentialsBar admission, conflict status, specialty certification — on-chainPilot stage

What to Skip​

Hand-rolled smart contracts for routine commercial agreements where traditional contracts work fine. Smart contracts add value when deterministic execution adds value the traditional contract cannot (escrow, milestone payments, programmable governance, cross-border settlement). For a standard service agreement between two parties in one jurisdiction with established case law, the smart-contract complexity adds risk without adding leverage.

Self-service consumer legal AI for high-stakes decisions without a lawyer in the loop. The asymmetric-field principle says use AI as a leverage multiplier, not as a lawyer substitute. AI drafts; humans judge at the boundary. Multiple jurisdictions have issued unauthorised-practice-of-law warnings on consumer-direct AI legal tools.

Closed vendor lock-in for high-volume volume work. Once the AI yield is realised, the value migrates to the firm's proprietary playbook + data. Closed vendors that own the AI layer and the data layer extract that yield. Architect for portability of data + multi-model architecture from day one.

Stack by Operating Stage — Industry Layer​

The industry's tech stack thickens by buyer segment, not by time.

  • Solo + small firm (1–10 attorneys): Open-source where mature (MikeOSS-style self-hosted or hosted at low cost); general-purpose LLM with retrieval-augmented generation; spreadsheet-grade CLM; simple compliance calendar. AI drafts most contracts; partner reviews.
  • MidLaw (10–500 attorneys): Commercial CLM platform; one or two AI vendors (wave-1 or wave-2 depending on procurement appetite); e-discovery platform with AI; jurisdiction-specific compliance tools. Internal AI training on firm corpus emerging.
  • BigLaw (500+ attorneys): Internal AI platform team; firm-corpus training; multi-vendor stack; full e-discovery + analytics; jurisdiction-specific tooling. Smart-contract review + token-law specialisations.
  • Corporate in-house (operating businesses): CLM + AI layer at scale; specialist tooling for the largest exposure areas (employment, regulatory, IP); counsel-marketplace integration for external work.

The stack does not justify itself until a Performance gauge it serves is breaking. Add platform tools when the gauge fires, not before.

Context​