Jeremy Henricks · Knowledge & AI

AI is only as good as the knowledge behind it.

I've spent 30 years proving that — leading Knowledge Management and KCS at Symantec and Broadcom, where structured knowledge powered AI chatbots, self-service, and $30M+ in annual support savings. Today I build with Claude, Cursor, and modern AI tooling every day. I work at the intersection most teams are still trying to find.

$30M+
Annual support savings
160K+
Articles governed
1,500+
KCS publishers mentored
20-35%
Case deflection
AI Practice

Not AI-curious. AI-operational.

Plenty of people talk about AI strategy. Fewer have run AI in production for enterprise support, and fewer still ship their own products with it. I do both.

AI in production, at enterprise scale

At Broadcom I managed an AI chatbot handling 1,500+ daily interactions on WebEx and Zoom, integrated with a unified KCS knowledge base of 160K+ articles. Built Python/NLP dashboards to find content gaps before customers did. Result: 20-35% deflection gains with mid-90s CSAT.

Hands-on builder

I design and ship software myself. KnowledgePeak.ai turns KCS article structures into guided AI generation. A USA Archery platform rebuild migrated 2,900+ articles into Payload CMS with a citation-backed support assistant. Shimmer is open-source audio restoration built on classical DSP. Working software, not slideware.

The methodology layer

My core thesis: AI output quality is a knowledge-structure problem. I write and speak about using KCS article structures for LLM fine-tuning, governance for agentic AI, and the change management that makes AI adoption stick instead of stall.

Daily toolkit Claude Cursor Perplexity Lovable Midjourney Adobe Firefly Python / NLP SearchUnify Azure AI

Building a knowledge or AI team?

I'm open to roles in knowledge management, KCS, and applied AI — anywhere structured knowledge and intelligent systems meet. Program leadership or hands-on practitioner work, depending on where the interesting problem is.