All case studies
Case 10AI-Native Transformation7 weeks

Legal aid organisation

More need than capacity, permanently, with no billing model to defend and nothing structural in the way. We rebuilt intake and first-pass drafting from zero - the case where recovered capacity converts most directly into people served.

Intake triageMatter containerDrafting agents

Practice

India · 14 caseworkers, 4 lawyers

Transformation

7 weeks

Track

AI-Native Transformation

Primitives moved

Matter, Actions, Research

01The challenge

The challenge

The organisation turned away applicants it was funded to help because caseworker time was consumed by eligibility assessment and repetitive form-filling before substantive help began. Case state lived in paper files and individual memory, so a caseworker's absence stalled every matter they held. There was no leverage model, no hourly billing, and no partnership - and therefore, unusually for an organisation of this size, nothing structural preventing a full rebuild.

02Primitives most affected

Primitives most affected

The MATTER primitives this engagement actually moved. Each links to what it means.

03The transformation, phase by phase

The transformation, phase by phase

  1. Phase 01

    Name

    One week reading closed case files and mapping the path from first contact to resolution. Eligibility assessment - which consumed a large share of caseworker time - turned out to follow a decision tree with eleven branches that no one had ever written down.

  2. Phase 02

    Systematise

    A digital matter container replaced the paper file, with required fields so state stopped depending on who held it. The eligibility decision tree was written out and became a structured intake sequence. Standard applications and letters, which caseworkers had been assembling from memory, became named actions with templates.

  3. Phase 03

    Automate

    Structured intake performs the eligibility assessment at first contact and routes accordingly, with any borderline case escalating to a lawyer rather than being decided automatically. A drafting agent produces first passes of standard applications against the templates, with a caseworker or lawyer reviewing every output. No eligibility refusal is ever made without a human decision - that boundary was set in week one and never moved.

04The shift to AI-native

The shift to AI-native

The manifesto argues that everywhere legal work is done under resource constraint - which is everywhere - the model applies, and that legal aid organisations have always had more need than capacity. This is the case where that claim is least abstract. Every hour recovered here converts directly into an applicant seen rather than turned away, with no revenue cannibalisation, no compensation model, and no partner vote in the way. Structurally, a legal aid organisation of this size has more in common with a solo practice than with a similarly sized commercial firm: no pyramid to protect, decisions made by people who can make them, and nothing to unlearn. That is why the full AI-Native track was honestly available at fourteen caseworkers when it is not available at sixty lawyers.

AI-native markers this firm now meets

  • Intake is automated end-to-end
  • AI runs first, on every matter, systematically
  • The organisation gets measurably better over time
  • Turnaround is fast, often same-day
Read the manifesto this case study argues for
05Outcome

Outcome

Applicants served / month

+34%

Intake assessment

3 days → same day

Eligibility refusals automated

0, by design

We measure this in people helped, not hours billed. That number went up by a third without anyone new being hired.
Book a call