Glossary
The language of the AI-native practice.
Plain-language definitions for the terms adnah uses across the method, the audit, and the journal - so nothing here requires decoding.
- AI-native practice
- A legal practice rebuilt around AI as a default, systematic first step on every matter - not a firm that occasionally uses an AI tool. Defined by seven markers: transparent pricing, automated intake, low-friction client contact, AI running first by default, measurable improvement over time, fast turnaround, and technical capability treated as core to the practice.
- MATTER Method
- An operating doctrine claiming all legal work decomposes into six primitives - Matter, Actions, Time, Team, Evidence, Research - that must be named, then systematised, then automated, in that fixed order.
- Matter (primitive)
- The atomic unit of legal work and the container every other primitive attaches to. A practice has 'named' this primitive when a competent stranger can open any matter and understand its complete state in ten minutes without asking anyone.
- Action (primitive)
- A discrete, named, repeatable step that moves a matter from opened to resolved. Extracting a firm's action library turns tacit expertise held in one lawyer's head into a written, improvable, automatable playbook.
- Evidence anchoring
- A standing rule that no factual assertion enters a work product without a citation to a specific, checkable source - page and paragraph, not just a document name. The primitive that answers the fear of AI hallucination with a workflow rule rather than a promise about model quality.
- Research stock
- Treating legal research as a compounding, searchable body of firm knowledge rather than a one-off memo that evaporates into a folder once its matter closes. The opposite of researching the same question from zero every time it recurs.
- Billable hour
- The traditional legal pricing unit: time recorded and invoiced after the fact. Trains a practice to record time rather than manage it, and structurally cannot survive automation, because every hour AI removes is revenue the billing model has no way to recapture.
- Fixed fee
- A price quoted to the client before work starts, made survivable only once a practice has real duration data on its named actions - otherwise fixed pricing is a guess, not a policy.
- Agent (legal AI)
- Software that executes a named action - drafting, checking, monitoring - inside an explicit doer/reviewer structure, with a human lawyer retaining review responsibility. Distinct from a chatbot a lawyer pastes text into occasionally.
- Hallucination
- A confident but false assertion produced by an AI system - a citation that doesn't exist, a holding that's inverted. Addressed structurally by evidence anchoring rather than by waiting for better models.
- AI-augmented (vs. AI-native)
- A firm that has adopted AI tools without rebuilding its underlying operating model - pricing stays hourly, intake stays manual, AI use is incidental rather than systematic. The overwhelming majority of large-firm 'AI adoption' falls here.
- Legal ops
- The operational infrastructure underneath legal delivery - intake, matter management, pricing, measurement - as distinct from legal judgment itself. Treated as core, not a support function, in an AI-native practice.
- Transformation (adnah usage)
- adnah's term for a full engagement: naming, systematising, and automating an organisation's six primitives end to end, typically four to eight weeks and occasionally up to ten, producing working artefacts the organisation owns and can run without the studio afterward.
- AI-Native Transformation (adnah usage)
- The track adnah runs for organisations with no legacy pyramid to protect: pricing, intake, and delivery rebuilt from zero into a fully AI-native operating model. Distinct from AI Adoption, which installs the same primitives inside an existing structure rather than replacing it.
- AI Adoption (adnah usage)
- The track adnah runs for larger, more established, or more regulated organisations that cannot honestly complete a full AI-Native rebuild - the same six primitives, named and systematised, installed as a systematic layer inside the sign-off chains and procurement cycles that already exist. Not a lesser outcome, a structurally honest one.
- Structural audit
- The first phase of any adnah engagement: a read-only look at what an organisation uses now, how it actually uses it, and what best practices are already in place, before any prescription is written. Produces the primitive map and the AI-Native vs AI Adoption track decision.
- MatterOS
- adnah's own software for running the Matter, Actions, and Team primitives - the matter container, action library, and role chart. Deployed where it fits; swapped for a custom build or market software where it doesn't.
- LexOS
- adnah's own software for running the Evidence and Research primitives - evidence anchoring and the compounding research stock. Deployed alongside MatterOS or independently, wherever the gap is real.
- Team primitive
- The explicit assignment of doer, reviewer, and decider to every named action - the delegation structure that makes it safe to hand routine actions to a non-human doer while a human keeps review responsibility.
- Time primitive
- Treating time as a property of the matter - structured deadlines and measured action durations - rather than a product sold to the client in six-minute increments.
See it applied
Read the method, or score your own practice.