Traceability & compliance

Release Notes

Full version history and technical traceability of the ClinIAMind platform.

This page documents the platform's technology stack, AI models used, and all changes made to date. It is designed to support internal audits and regulatory due diligence requirements.

Technology Stack

Core frameworks, libraries and services used in production.

Framework & language

Next.js 16.0.7

React 19.2

TypeScript 5

UI & design system

Tailwind CSS v4

Radix UI

Database & storage

Supabase (PostgreSQL + pgvector)

Authentication

Clerk v6

Billing & payments

Stripe v20

Internationalisation

next-intl v4

AI SDK

Vercel AI SDK v5

AI Models Used

Large language models and embedding models integrated into the platform.

OpenAI GPT-4o

OpenAI

Semantic search, document chat, metadata extraction, auto-annotation.

@ai-sdk/openai ^2.0.77

Claude (Anthropic)

Anthropic

Structured report generation, long-form summarisation, evidence synthesis.

@ai-sdk/anthropic ^2.0.53

Gemini (Google)

Google

Alternative LLM backend for report generation and summarisation tasks.

@ai-sdk/google ^3.0.2

text-embedding-3-small

OpenAI

Dense vector embeddings for semantic similarity search via pgvector.

openai ^6.10.0

Version History

Chronological record of all platform versions and changes.

v1.1Security patch

14 March 2026

Security Audit & Hardening

Full security audit of all API routes and infrastructure configuration. No critical vulnerabilities found. 17 files hardened, missing protections added.

Generic error messages on all API routes (17 files) — internal details no longer leaked to clients
ILIKE query escaping in semantic search to prevent injection via special characters
Clerk authentication added to /api/ai/embed endpoint to prevent unauthenticated cost exploitation
Content-Security-Policy header added in next.config.ts
Verified: webhook signature verification (Stripe & Clerk), file upload validation, no secrets exposed client-side
Pricing page removed; all pricing links now point to #pricing section anchor

Implemented by Claude Sonnet 4.6 (Anthropic)

v1.0Major update

14 March 2026

Landing Page Redesign & Internationalisation

Complete redesign of the public landing page with full multilingual support across 5 languages and mobile-first responsive layout.

Full landing page redesign: hero, features, how-it-works, testimonials, personas, pricing, FAQ, CTA, footer
Multilingual support implemented for EN, FR, DE, ES, IT via next-intl lp namespace (~100+ keys per language)
Mobile responsiveness: DashboardMockup overflow fixed, floating cards hidden on mobile, persona tabs flex-wrap, MobileMenu links synchronised
Hero h1 and CTA h2 font size reduced on mobile (text-[1.85rem]) with break-words to prevent text overflow
Anonymous testimonials replacing real names; trust bar with fabricated logos removed
Annual pricing toggle removed; only monthly pricing displayed

Implemented by Claude Sonnet 4.6 (Anthropic)

v0.1Initial release

Q4 2025

Initial Platform Release

First functional version of the ClinIAMind platform. Core document management, AI-powered search, annotation workflow and report generation.

Document library: PDF upload with automatic metadata extraction (title, authors, journal, DOI, year)
Semantic search powered by OpenAI text-embedding-3-small + Supabase pgvector
AI annotation system with highlight extraction and evidence tagging
Structured report generation (IVDR SVR format) via Anthropic Claude and Google Gemini
Multi-tenant architecture with Clerk authentication and Stripe subscription management (Free, Starter, Pro, Enterprise)
Internationalisation infrastructure (next-intl) with EN, FR, DE, ES, IT locale support

Implemented by Human development team

Compliance & Audit Note

ClinIAMind is designed to support literature-heavy regulatory workflows such as IVDR Clinical Evaluation and Performance Evaluation. This changelog page provides the traceability layer required by quality management frameworks. All AI-generated outputs must be reviewed and validated by a qualified regulatory professional before use in any regulatory submission.