@openpjHi, I'm Piergiorgio, here's what I'm building
Hi everyone, I’m Piergiorgio, a Solution Architect based in Rome and I’m thrilled to join the AI Founders community!
Over the last few months, I’ve been talking to enterprises and they all share the exact same frustration: they want to deploy GenAI and RAG, but they are terrified of Data Leakage.
Traditional ingestion pipelines and Vector DBs strip away Access Control Lists (ACLs) when parsing corporate documents. The result? An intern querying the company's internal LLM could accidentally retrieve chunks of a confidential board meeting or the CEO’s contract.
To solve this, I’m building OpenCrawling the next-generation Zero-Trust RAG orchestration engine for the enterprise.
Instead of just scraping text, OpenCrawling bridges complex legacy systems (Alfresco, Hyland, SharePoint, Flowable) to modern AI infrastructure while strictly preserving business metadata and security permissions at the chunk level.
Here is what we have achieved recently and why it’s a game-changer for B2B AI adoption:
🛡️ Zero-Trust by Design: We extract original ACLs (via XACML/CMIS) and inject them directly into the Vector DB. Your AI only answers based on what the specific user is actually allowed to see.
🏛️ Global Standardization: We are currently working with the OASIS Consortium to formalize our underlying data structure into the Open Ingestion Standard (OIS). It’s not just a tool; we are building the industry standard for secure AI data ingestion.
⚙️ Modern Stack: Built on Java 25 and Spring Boot 4, leveraging Virtual Threads for massive concurrency. It also features a native Model Context Protocol (MCP) server to instantly connect corporate knowledge to LLM agents and UI explorers.
📦 "Migrate Now, AI Later" Pattern: We just introduced a migration mode targeting Apache Ozone (S3-compatible). Enterprises can stream legacy data into modern data lakes using our "OIS Sidecar" format (bypassing expensive embeddings initially) so they are 100% AI-ready when the time comes.
Traction & Ecosystem:
We’ve successfully validated POCs with core developers from major BPM vendors (like Flowable) using a collocated pgvector architecture, and we are rapidly expanding our connectors (Apache SeaTunnel, StormCrawler, CMIS, Hyland OpenArch, OnBase and CIC).
Why I’m sharing this here:
I’d love to connect with other founders in the B2B/Enterprise AI space. Whether you are building governance tools, agentic workflows or exploring the Commercial Open Source (COSS) business model, I’d love to exchange ideas, get your brutal feedback or just grab a virtual (or physical, for the Rome crew!) coffee. ☕️
Check out the website: https://opencrawling.org/
Check out the repo here: https://github.com/opencrawling/opencrawling
Read our Manifesto: https://github.com/opencrawling/open-ingestion-standard/blob/master/MANIFESTO.md
Let’s build the secure future of Enterprise AI together!
