Built for CBRE conversation · April 2026
Head of AI Products
Strategy, demos & how I work
Enterprise AI product leadership means turning complex opportunities into safe, measurable capabilities—from discovery and evaluation through adoption—partnering across product, data, engineering, and governance so GenAI, predictive insight, and agentic automation land where they actually matter.
Proof points
- Trimble — Agentic AI Platform (Principal AI Solutions Architect): Shipped 2,000+ API nodes, 1,500+ MCP tools, and 50+ production AI workflows—platform scope tied to $14M+ in opportunity—while scaling adoption to 150+ internal and external teams and ~40% improvement in decision-making across those teams (program metrics).
- Revenue & enablement: Influenced $4M ARR closed-won by aligning platform capabilities with 200+ stakeholders; authored 15+ guides and templates that reduced implementation errors ~25%; presented at 4 industry conferences (5,000+ attendees) to build adoption and credibility.
- Integrations + GenAI engineering: Delivered 21 production C# connectors mapping 2,500+ endpoints ($10M+ pipeline; $4M+ proposed development scope). Built Ludex AI—LLM fine-tuning, RAG, vector embeddings, and custom MCP servers—for NL→SQL analytics: the same architectural patterns I bring to enterprise GenAI products.
Live proof-of-concept demos
These links go to working environments I stood up on LibreChat (conversational AI, agents, MCP-style tooling) and n8n (workflow automation and integrations)—the same categories of products enterprises already run at scale. The intent is a tangible reference: “what it feels like” when you pair modern chat + orchestration with your data and governance—not a CBRE production system, but a credible pattern you can react to and tear apart together.
Why these stacks: Organizations such as Spotify have publicly associated with LibreChat for internal AI; Trimble (where I ship agentic platform work) uses n8n in production alongside other enterprise automation—so this isn’t a toy stack; it’s what serious operators already standardize on.
What I'm building in this POC
- LibreChat — Targeted agents and MCP integrations that mirror how I'd pilot grounded GenAI (eval-friendly, traceable tool calls)—so you can judge product judgment, not slide theory.
- n8n — Flows that show cross-system coordination with clear ownership between steps—how I'd demonstrate orchestration before we commit capital to platform-wide rollout.
Ideas to explore together
CBRE's roadmap and constraints are yours to define—these are conversation starters, not instructions. I'm listing hypotheses I'd validate with product, data, engineering, and risk partners so we can quickly separate signal from theater.
- Document-heavy workflows (GenAI): Retrieval + citations across leases, OMs, work orders, and tickets—with role-aware access and evaluation gates—often the fastest credible win when data is messy but high-value.
- Agent + MCP patterns: Assistants that do work via governed tool calls into internal systems, not just chat—paired with logging and human-in-the-loop approvals where outcomes are material.
- Orchestrated operations (n8n-class): Repeatable runbooks for facilities, portfolio, or service-line processes—retries, routing, and audit trails—so automation scales without losing the enterprise bar on reliability.
If this resonates, I'd rather spend our time pressure-testing one or two themes against your reality than debating slide decks— that's how I tend to turn proof into credible product bets.
Selected videos
Play inline on this page—no tab-hopping. Where a clip is bookmarked, it picks up at that moment. Nothing autoplays; each player loads as you scroll.
About Charley
I build AI and automation products end-to-end: clarifying use cases, designing evaluations and guardrails, and shipping experiences people actually adopt. These demos reflect how I think about reliability, traceability, and responsible rollout in real enterprise settings.