Contract law · Carlill v Carbolic
Reason out loud…
A Socratic tutor for Indian legal education
Tark teaches law students the Socratic way, one question at a time, then maps every conversation onto a reasoning arc faculty can replay, diagnose, and grade.
The student identifies the legal issue.
Tark never hands over the answer. It asks the next useful question, adapts to how the student is thinking, and keeps them reasoning until the argument stands on its own.
TARKContract law · Carlill v Carbolic
Every move becomes a node; every gap becomes a signal. Faculty and students replay exactly where the reasoning branched, advanced, and broke down.
TRIM, the TARK Reasoning Intervention Manual, is Tark's diagnostic engine. It reads every move a student makes and grades it against a faculty-authored rubric, checkpoint by checkpoint, to find not just whether the reasoning is wrong, but exactly where and why. This is what generic AI cannot do.
A conclusion stated without connecting it to a fact in the record. Tark's next question targets this exact gap; it never just supplies the missing fact.
TRIM: TARK Reasoning Intervention Method.
Most tools tell a student whether an answer is right. Tark names why the thinking failed, from a taxonomy of eight reasoning gaps that recur across every area of law.
The student never spots the legal question that actually decides the case.
In practice: arguing about damages before asking whether a contract was ever formed.
Every session becomes competency evidence. Faculty see the concepts a class most needs taught, the gap types driving it, and which students need intervention, mapped to the standards your accreditation already speaks.
Available in English, Hindi, and French, built for multilingual legal education.
Contracts I · 38 students
Most frequent reasoning gaps
Across every Socratic session, Tark evaluates how students think, not simply whether they reach the right answer. Each move is graded against a faculty-authored rubric and rolled up into competency evidence.
Issue, rule, application, evidence, counter-argument, authority, synthesis, challenge.
A taxonomy of reasoning failures Tark names, not just flags as wrong.
Grades reasoning against the rubric, checkpoint by checkpoint.
Authored by your faculty, so every diagnosis reflects your curriculum.
Every session replayable as a branching tree of moves and gaps.
Patterns across a whole class, surfaced as competency reports.
Competency evidence framed the way NEP asks institutions to show it.
Reasoning competencies mapped to European qualification levels.
Reading builds knowledge. Generic AI gives answers. Only Tark teaches the reasoning, and shows faculty how it develops.
“Tark made me realise I wasn't actually reasoning, I was just reciting.”
“The replay showed me exactly where I lost the thread. No professor has time to do that.”
“Finally a tool that teaches law the way the Socratic method was always meant to work.”
Tark runs paid institutional pilots, scoped with a named faculty champion and measured against a control group. The pilot succeeds when reasoning improves, not when minutes are logged.
Talk to usPriced per seat or per department, scoped with your registrar.
No, that is the one thing it is built never to do. Tark asks the next question, and a scaffold ladder escalates help one rung at a time. Direct teaching happens only when reasoning genuinely stalls, and every release is checked against answer-leak evaluations.
TRIM stands for TARK Reasoning Intervention Manual. It is Tark's proprietary diagnostic engine: it grades each reasoning move against a faculty-authored rubric and classifies failures into eight reasoning gap types. Generic AI can produce an answer; TRIM diagnoses the thinking: where it broke down and why.
A cohort-level competency report showing the concepts the class most needed taught and the gap types driving it, plus drill-down into each (anonymised) student's reasoning arc. Faculty can validate or correct any diagnosis; the correction loop is part of the product.
Yes. Tark's reports are framed as competency evidence (issue spotting, application, evidence-based reasoning), aligned to NEP 2020 and mapped to Bologna EQF qualification levels, giving institutions visibility they otherwise have no way to see.
English, Hindi, and French today, on a Hindi-first architecture built for multilingual legal education. The substantive law changes across jurisdictions and languages; the reasoning Tark diagnoses is the same.
With three people: a faculty champion, a signer who controls budget, and a semester start date. Write to tarkedullp@outlook.com and we'll scope a paid pilot for one cohort.
The takeaway
Socratic dialogue that never gives the answer. A reasoning arc you can replay. Diagnostics your institution can act on.