← ALL CASE STUDIES
DOSSIER 01 · SHIPPED SYSTEM — K–12 LEARNING PLATFORM

The drafting tool that never says “I don’t know.”

Problem class: constrained generation. An AI whose correct output is sometimes a refusal, enforced by a control layer rather than by prompt compliance.

PROBLEM CLASSConstrained generation
CLIENTK–12 learning platform
ARCHITECTUREFive agents, supervisor control layer
STATUSShipped · peer-reviewed
§1 — THE PROBLEM IN YOUR FIRM

The document looked finished, so nobody looked twice.

SCENARIO

An estate-planning client completes intake. Somewhere in the free-text notes: a second marriage, and a child from the first who is mentioned once and never again.

The drafting assistant assembles a will. It produces a clean, complete, internally consistent document — because a complete document is what it was built to produce. The ambiguity about which children are provided for was not flagged. It was resolved. Quietly, by whichever reading made the template close.

The attorney reviews it. Nothing looks wrong, because nothing looks wrong — a confident document reads as a correct one. It is executed. The defect surfaces at probate, years later.

FAILURE MODESilent completion

A language model optimised to produce output will produce output. Given an incomplete instruction it does not stall — it interpolates, fluently, and the fluency is indistinguishable from correctness at review time. The same mechanism produces the intake assistant that answers a question constituting legal advice, the review pass that resolves a contract ambiguity instead of surfacing it, and the citation to a case that does not exist. These are not four problems. They are one problem wearing four hats.

Instructing the model not to do this does not work. A refusal held only by a prompt degrades in two directions: under pressure it capitulates and answers anyway, or it over-triggers and refuses routine work until staff stop using it.

§2 — WHERE WE HAVE MET THIS BEFORE

We built a system whose entire value depended on withholding the answer.

A K–12 learning platform engaged us with a product requirement that is, structurally, the one above. The AI's job was to advance a student's reasoning without resolving it for them. Give away the answer and the product has no reason to exist.

A single model with a well-written prompt could not hold the line. We tried. It capitulated under pressure and stonewalled on routine input. What worked was moving the decision out of the generating model entirely: five specialised agents orchestrated in LangGraph, release authority isolated in a supervisor that is not a generating model, retrieval grounded in a vector store, every step traced. No agent that generates text also holds the authority to release it. That separation is code, not instruction.

50%LOWER INFRASTRUCTURE COST
30%FASTER AI RESPONSE
60%LOWER DATA-LABELLING COST
Peer reviewedARCHITECTURE PUBLISHED
THE ARCHITECTURE WAS PUBLISHED AS A SUPERVISED MULTI-AGENT PEDAGOGICAL FRAMEWORK AND SUBMITTED TO REVIEW. THE RESEARCH IS PEDAGOGICAL — THE ARCHITECTURE TRANSFERS; THE DOMAIN VALIDATION DOES NOT, AND WE WON'T PRESENT IT AS THOUGH IT DOES.
ON THE RECORD

“Our biggest challenge was taking a complex ‘teachable agent’ concept where the AI helps students learn without just giving away the answers. They didn’t just build the software. They engineered a robust AI framework that solved the actual teaching challenges.”

FOUNDER & CEO · K–12 LEARNING PLATFORM
§3 — THE APPROACH FOR A FIRM

Move the release decision out of the model that writes the document.

Structurally identical, with the agents re-specified for legal drafting: retrieval over firm precedent and actual authority, a drafter limited to cited clauses, a conflict checker, a gap detector — under a completeness supervisor whose halt path is a first-class output. A halt is a successful output. The system is measured on what it refuses to produce, not only on what it produces.

WHAT WE BUILT
WHAT IT BECOMES IN YOUR FIRM
Supervisor holds release authority
The escalation rule is code. A gap stops the document whether or not the model “chose” to comply
RAG grounded in a vector store
Clauses drawn from firm precedent and actual authority. No source, no clause
Evaluation agent scoring progress
Conflict and omission checks that run before a human ever opens the draft
Graduated hint release
Three distinct outcomes: proceed, assist, halt and escalate
Trace on every interaction
When a partner asks why a paragraph reads that way, there is an answer — in the log, not from memory
INDEX All case studies NEXT STUDY Nine sources, and one document you sign →
DWG. FINAL

Start with the plan.

The $2,000 deposit is credited toward the build and refundable for two weeks after the plan lands. The plan is yours either way.

Book a scoping call →