[Confidential Compliance Platform]
Multi-tenant SaaS platform that automates the EN 303 645 self-assessment process for IoT manufacturers entering the European market. RAG architecture with AI agents for document analysis, standard mapping, and verdict suggestion.
Client: Confidential | IoT Manufacturing | Northern Europe
CE Compliance is a SaaS platform for European product manufacturers who need to navigate the EU's CE marking and broader product-compliance regimes. The directives are dense, change frequently, and a single wrong answer can hold up shipments at customs.
The founders came to us with deep regulatory expertise and a thesis: if we could give a manufacturer's compliance team an AI assistant grounded in the current state of EU directives and harmonized standards, we could replace days of legal-review work with minutes of guided answers — without giving up the audit trail compliance teams need.
The challenge was building something that could be both genuinely helpful and demonstrably trustworthy.
Manual compliance audits took weeks per product, blocking time-to-market in Europe
Document analysis was inconsistent across reviewers — same evidence, different verdicts
Standard-to-product mapping was error-prone and impossible to audit at scale
This was a RAG (retrieval-augmented generation) project from day one. We built a multi-tenant Laravel + Filament platform on top of PostgreSQL, with Pinecone as the vector store for the directive corpus. Every answer the AI gives is grounded in retrieved source passages with full citations — no "trust me" answers, ever.
The multi-tenant architecture matters because different manufacturers have different product portfolios, internal documents, and compliance histories. Each tenant gets their own private corpus layered on top of the shared regulatory base, and the retrieval layer respects those tenant boundaries strictly.
We chose PostgreSQL over MySQL specifically for the JSONB-heavy compliance metadata and the row-level security model that backs the tenant isolation. OpenAI handles the generation, but the value is in the retrieval layer we built around it.
RAG-powered document analysis grounded in the EN 303 645 standard and tenant-specific evidence
AI agents for standard mapping — clauses linked automatically to product features and test units
Verdict suggestion with human-in-the-loop review, full reasoning traces, and audit trail
Each evaluated automatically against the standard
From evidence ingestion to verdict reporting
Tenant, product and document layers prefilled by AI
The technical foundation behind CE Compliance
Services we offer that built the foundations of this project
Other projects with similar shape, industry, or tech stack
From RAG architectures to multi-tenant SaaS — we build the AI compliance engines that make audits a workflow, not a project.