01 — CM COPY REVIEWER · AUG–OCT 2025 · HEALTHTECH · CONSTITUTIONAL AI

Teaching a healthtech AI to write like its brand

Role

Prompt Architect · AI Constitutionalist · Vibe Coder

Team

XFN collaboration with the HealthTech marketing team

Timeline

Aug – Oct 2025

A healthtech marketing team needed AI that could review Instagram captions with brand voice accuracy, medical compliance, and Singapore cultural specificity — not generic suggestions from a general-purpose model. I designed the Constitutional AI framework that made it possible, then built the tool they use daily.

WHAT I DID

Audited the marketing team's copy workflow across 12 use case types to map where AI assistance would reduce overhead without adding new error risk.

Designed the CM (Constitution Maker) Principles framework from scratch — 12 principles covering brand voice, medical compliance, CTA structure, and Singapore cultural specificity — iterated across 12 versions over 8 weeks.

Invented and ran a dual-session evaluation method: one Claude conversation acting as copywriter, a second as evaluator — both operating on the same CM Principles, benchmarked against human-written captions from the marketing team.

Used Claude's /compact command to branch conversations and preserve principle JSON across testing sessions, enabling versioned iteration without losing prior evaluation results.

Made the call to narrow scope from 12 use cases to IG captions only after v1 testing revealed that broad scope degraded the precision of every individual principle.

Built and deployed the HealthTech Copywriter AI webapp on Google AI Studio, shared directly with the marketing team as an active workflow tool.

THE CHALLENGE

How do we build a GenAI copy reviewer that assists a marketing team with brand adherence and medical compliance — across a spectrum of audiences?

The team produced 12 types of marketing copy for Raffles Connect. Each had different compliance requirements, audience registers, and brand voice standards. Three constraints made this genuinely hard:

Switching between 12 use cases

IG posts, EDMs, articles, press releases — each required different tone and compliance standards. A single reviewer built for all of them would be too broad to be precise.

Generic AI output

General-purpose LLMs did not know Raffles Connect's brand voice, could not apply Singapore cultural context, and had no awareness of MAS/MOH copy guidelines.

Low GenAI literacy on the team

The marketing team could not prompt-engineer effectively. Any solution needed to be deployable as a no-code interface — usable without understanding the system behind it.

THE WORK

Reviewing IG Caption

Adherence

—

The CM Copy Reviewer in action: the constitutional AI scores each IG caption against four principle dimensions in real time. Hover to pause.

The Claude Project that became the artifact: CM Principles JSON files across 12 versions, constitutional reference documents, and evaluation sessions organized by iteration round.

I used Claude's Developer Workbench to customise a System Prompt that served as the foundational instruction set for my creation and testing phase. The marketing team provided feedback at demos; I used that to align on expectations and refine CM Principles. Eventually we agreed to focus solely on IG captions — this let me build a constitution with the precision the use case demanded.

Evaluation testing in action: scoring AI-generated captions against the CM Principles framework, benchmarked against the marketing team's human-written baseline.

Part 2 of the evaluation session: the second Claude conversation acts as evaluator, using the same CM Principles to judge output from the first.

To not just maximise token utility but preserve the core essence of my testings, I had to "branch / version" iterations using the /compact command. This allowed me to add completed testings and CM Principles as JSON to the Claude project's overview — and maintain a clean evaluation history without losing earlier results.

CM Principles versioning — tracking evolution from v1.0 through v12.0 using /compact to branch conversations and preserve each constitutional milestone as a JSON snapshot.

CM Principles v12.0 — 396 lines. The JSON structure tracks principle evolution from v8.0 ("Initial Framework — 12 core principles established") through v12.0 ("Production-ready with enhanced cultural context and formal content optimisation").

The deployed HealthTech Copywriter AI — built on Google AI Studio with Claude-tested CM Principles. Each suggestion is tagged by principle number and includes reasoning on why it works. Singapore-specific cultural hooks (Ah Ma, heaty, TCM, polyclinic) are applied automatically per the constitution.

INTERACTIVE PROTOTYPE — try the scorer

Try it — paste an IG caption and score it

This prototype demonstrates the scoring dimensions from the CM Principles framework. Paste any IG caption to see how it scores.

OUTCOMES

Tested against baseline IG captions written independently by the marketing team — no AI assistance. Testings and evaluations were done concurrently using a second Claude conversation with the same System Instructions, acting as evaluator and benchmarking against every human-written caption. After 8 weeks of iteration:

80–90%

adherence to Constitution Principles — compared to the team's 75% baseline from human-written copy

75% faster

copy creation and review turnaround — 50 minutes with the AI workflow vs 2+ hours previously

+15%

estimated improvement in CTA click-through rate on IG posts

REFERENCE

ConstitutionMaker: Interactively Critiquing Large Language Models by Converting Feedback into Principles

GenAIConstitutional AIPrompt EngineeringClaudeHealthTechGoogle AI Studio