Hi everyone! 👋
I’m Francisco, a computer scientist working on production-grade AI systems. I recently completed the Google + Kaggle AI Agents Intensive and built LexFabric Agents, a deterministic multi-agent pipeline that pairs LLM extraction with Python-based reasoning for accurate timeline reconstruction.
I’m here to share what I’m learning about multi-agent architectures, context engineering, and reproducible AI workflows — and to learn from this community as well.
When I enrolled in the Kaggle AI Agents Intensive Course with Google, I wasn’t looking for another abstract walkthrough of agent tools or a high-level tour of LLM capabilities. I came in with a very real problem: LLMs hallucinate timelines, and in high-stakes domains like law, compliance, and investigations, that’s unacceptable.
Real-world evidence doesn’t arrive neatly sorted. It comes as mismatched PDFs, emails with partial timestamps, handwritten notes, system logs, and fragments created months apart. I’ve spent years in environments where answering “What happened, and when?” determines outcomes, safety, and truth.
My goal during the Intensive was to build a system that solves that challenge — not with prompt engineering alone, but with architecture.
A system where LLMs read the evidence, but deterministic software enforces the truth.
That system became LexFabric Agents.
Looking forward to connecting with you all! 👋
Top comments (0)