Gentle Pills

Engineering & clinical AI

AI & modern homeopathy

Gentle Pills is an AI-assisted clinical workspace built specifically for qualified homeopathic doctors. These are the technical details: the Azure OpenAI architecture underneath it, why it's built the way it is, and how each AI-assisted tool actually works.

New here? Try the no-sign-up AI demo first — it's the fastest way to see what these articles describe in action.

Architecture

How Gentle Pills Uses AI: A Technical Overview of Our Clinical Assistant Architecture

The full request path from a doctor's click to a stored, evidence-linked AI suggestion — and why every step is designed around doctor confirmation, not automation.

Read article

Infrastructure

Tiered LLM Routing on Azure OpenAI: Why Gentle Pills Runs Three Model Sizes, Not One

Routing every AI task by failure cost rather than perceived difficulty, across fast/mid/reasoning Azure OpenAI deployments.

Read article

Building the Embedding Layer: Semantic Search Across Classical Materia Medica

What's already live (3072-dimension Azure embeddings), what's still lexical search, and the honest roadmap for wiring pgvector into retrieval.

Read article

Engineering

Schema-Derived Prompting: Why We Generate LLM Output Contracts From Code, Not Prose

A real production bug — and the structural fix (z.toJSONSchema in the prompt, not a hand-written description) that eliminated a whole class of model-swap failures.

Read article

Trust & Safety

Grounded AI: How Evidence Citations Prevent Hallucinated Answers in Clinical Use

Why Gentle Pills' AI only cites passages it was actually given, stores them as inspectable evidence rows, and says so plainly when it has nothing to cite.

Read article

Why Our AI Never Prescribes Autonomously: A Human-in-the-Loop Design Philosophy

The concrete mechanisms — labeling, dismissal, explicit confirmation, audit logging — that keep every AI suggestion a suggestion, never a silent write.

Read article

Modern Practice

From Free-Text Case Notes to Structured Symptoms: AI-Assisted Clinical Documentation

How modern homeopathic practice can keep the doctor's own words intact while still getting searchable, structured symptom data — with AI doing the tedious part.

Read article

AI-Assisted Repertorization: Rubric Suggestion and Transparent Coverage Ranking

Speeding up rubric selection against a classical repertory without turning the analysis into a black box — the math stays open, the decision stays the doctor's.

Read article

Remedy Matchmaker: Using AI to Explain Classical Remedy Relationships

Turning terse classical complementary/follow-well/antidote relationships into reasoned, case-aware narratives a doctor can actually use at the point of prescribing.

Read article

Miasmatic Pattern Analysis with AI: Applying Kent's Classical Framework at Scale

How pattern recognition over structured symptom data can surface candidate miasmatic layers for a doctor to evaluate, without ever asserting a diagnosis itself.

Read article