Solutions

Outcomes Over Activity.

Most marketing ops functions are busy. Few are effective. Here is what effective looks like.

Delivered

STMicroelectronics·Senior Manager, Marketing Programs

Built $20M in Pipeline by Standing Up STMicro's First Global Lead Management Program

The Situation

No global lead management framework existed. Demand generation was siloed, unscored, and unmeasured.

The Approach

Architected the end-to-end lead management system from scratch: defined scoring models, built automation, partnered with IT on infrastructure, and drove adoption globally through training and change management.

The Outcome

Generated over $20M in new sales pipeline within the first six months of launch. The framework became the operational standard for all demand generation globally.

SalesforceAdobe CampaignCross-functional AlignmentChange ManagementGlobal Rollout

From zero lead processing to a global framework. The four-phase journey from alignment through build, pilot, and rollout.

State at Arrival
Salesforce and Adobe Campaign were live — leads captured and stored, zero processing. No scoring, routing, SLAs, or measurement.
Salesforce ✓Adobe Campaign ✓Lead Scoring ✗Routing ✗SLAs ✗Measurement ✗
BEGIN PHASE 1
Phase 1 — Alignment
🏢
Marketing Leadership
Executive sponsor
🌍
Regional VP Sales
EMEA · APAC · AMER
⚙️
IT + Marketing Tech
Salesforce + Adobe owners
Sold the program by projecting pipeline impact. Secured sign-off on scope, ownership, and project plan before a single line of configuration was written.
Output: Approved program design + project plan
BEGIN BUILD
Phase 2 — Build
🎯
Lead Scoring
Behavioral + demographic model in Salesforce + Adobe Campaign
🔀
Routing
Score-threshold rules by region and segment
⏱️
SLAs
Response time commitments agreed with each regional sales team
📊
Measurement
Pipeline attribution dashboard tracking lead-to-opportunity by region
Output: Fully integrated and tested system
BEGIN PILOT
Phase 3 — Pilot EMEA
🏛️
Geneva HQ
EMEA selected — closest to HQ for fast feedback loops
📚
Enablement
Materials developed and delivered to EMEA sales managers
🚦
Expansion Gates
Clear criteria defined before rolling out to next regions
Output: Validated process + early pipeline results
BEGIN GLOBAL ROLLOUT
Phase 4 — Global Rollout
🌏
APAC
EMEA playbook replicated — same enablement, same gates
🌎
AMER
EMEA playbook replicated — same enablement, same gates
📈
Global Dashboard
All three regions tracked in unified measurement framework
Output: Global lead management program live
Result — First 6 Months After Launch

Global lead management program live across EMEA, APAC, and AMER. Framework became the operational standard for all demand generation at STMicroelectronics.

$20M
New sales pipeline
1 of 7

Built

Products and Tools I Have Shipped

I do not just advise on AI and automation. I build with it. These are real projects, shipped and live.

Full-stack SaaS

Pathwise

A workflow optimization tool for RevOps and MarTech teams. Users build workflow canvases, run an AI-powered alignment assessment, and generate a structured improvement blueprint. Built solo using Next.js, Supabase, React Flow, and the Claude API.

Live at pathwise.works
Full Claude-powered assessment engine
Guest and authenticated experience
PDF blueprint export
Next.jsTypeScriptSupabaseClaude APIReact FlowVercel

The full build workflow from concept to live product — built solo with Claude as the primary development partner.

Phase 1Foundation & Documentation
✍️ AI-Assisted Planning — One Time Setup
🧑‍💻
Product Vision
Initial concept, goals, target users, and core feature ideas
You
Draft
🤖
Gemini
Developed the initial PRD from concept conversations
Initial PRD
Refine
Claude
Refined the PRD, developed the full TRD, and generated all baseline project files and folder structure
PRD + TRD + Scaffold
🔄
Automated Documentation Process
After every git push, npm run update-docs uses the Claude API to analyze the commit, then updates the PRD, TRD, and Project State docs automatically. No manual doc maintenance required.
BEGIN BUILD PHASE
Phase 2Ongoing Build Loop
👥
Early Users
RevOps and Martech professionals — direct demos, walkthroughs, and conversations shaping product direction
Customer Discovery
Demos + Feedback
🧑‍💻
You
Product decisions, design direction — shaped by direct feedback from early users, then brought to Claude
Windows PC
CONTEXT + PROMPTS
↕ EDITS + FEEDBACK
Claude.ai
Strategy, visual feedback, project state review, direct file edits via MCP, and precise Cursor prompts
AI Co-pilot
DIRECT EDITS VIA MCP — OR — PASTE PROMPT INTO CURSOR
🖥️ Local Dev Environment
Cursor
AI code editor — writes and edits files
Code Editor
Next.js
Live preview at localhost:3000
localhost:3000
🗄️
Supabase Dev
Isolated dev database — safe to experiment
Dev Project
🌐
Browser
Preview + screenshots → Claude in Chrome
Chrome + Claude
📁 Project Files
app/components/docs/.cursorrulesupdate-docs.js
🚀 Production Environment
Vercel
Auto-deploys on every git push — live in ~60s
Production Host
🗄️
Supabase Prod
Separate DB with RLS policies — real user data
Prod Project
🌍
Live Site
What real users see — pathwise.works
pathwise.works
🔐 Dev/Prod separation enforced via env vars. No data commingling between environments.
git add . → git commit → git push → npm run update-docs
☁️ Shared Cloud Infrastructure
🐙
GitHub
Source of truth — every push triggers Vercel deploy
Auto-deploy
Vercel
Builds Next.js and deploys to production in ~60s
Production Host
Anthropic API (Claude)
Powers Waypoint (in-app assistant), health check, assessment interview, blueprint generation, and automated doc updater. Sonnet 4.6 for all main flows, Haiku 4.5 for parallel review agents.
Authentication — Supabase Auth + OAuth Providers
🛡️
Supabase Auth
Central auth hub — manages sessions, JWTs, OAuth callbacks, and RLS enforcement. Configured identically in both dev and prod.
Session + RLS
Handles
🔵
Google OAuth
Google Cloud Console
OAuth
🪟
Microsoft OAuth
Azure App Registration
OAuth
🔗
LinkedIn OIDC
OpenID Connect
OIDC
✉️
Email Magic Link
Supabase built-in
Magic Link
↔️
Guest vs. Authenticated Experience
Guests can build workflows and run the mechanical health check — canvas state lives in localStorage only. Authenticated users unlock the Claude alignment check, assessment interview, blueprint generation, and full Supabase persistence. On sign-in, any guest workflow is claimed automatically. New users complete onboarding to capture department, tools, and skills — injected into every Waypoint chat.
Observability Stack — Coming Soon
📊
PostHog
Product analytics — tracks key user events across the build → assess → blueprint lifecycle. Session recording and funnel analysis.
Coming Soon
🔗
LangSmith
LLM tracing and evaluation — monitors every Waypoint chat, assessment agent, and blueprint generation call end-to-end.
Coming Soon
🚨
Sentry
Error monitoring and alerting — captures frontend exceptions and API failures across both dev and production environments.
Coming Soon
⚡ The Full Build Loop — Repeated Every Session
1. Idea or screenshot2. Claude reviews + plans3. Claude edits via MCP or writes Cursor prompt4. Cursor edits files5. Preview on localhost6. Screenshot → Claude in Chrome7. Iterate until done8. git push9. update-docs🌍 Live on pathwise.works in ~60s
AI-powered portfolio project

This Site

growthbydesign.cv is itself a project. Built in Next.js with an AI pipeline that researches a company and role, then generates a personalized company page.

AI research and page generation pipeline
Visitor identification via Snitcher
IP-to-org resolution via ipinfo
Magic link auth via Supabase
Next.jsTypeScriptClaude APISupabaseRB2BipinfoResendVercel

The AI pipeline that powers this portfolio site.

Layer 1Consulting Site
Next.js 14
App Router
Framework
📘
TypeScript
Type safety
Language
🎨
Tailwind
Utility-first CSS
Styling
🧩
shadcn/ui
Component library
UI
Vercel
Deployment
Hosting
Claude
Co-pilot via MCP filesystem access and Chrome extension for iterative development
AI Co-pilot
Edits
🧑‍💻
You
Design decisions, content, and deployment
Developer
LAYER 2 — AI APPLICATION ENGINE
Layer 2Company Page Creation — 6-Step Pipeline
JD + URL
Input
Research Agent
Claude researches company
HM Research
Optional LinkedIn layer
Draft Agent
Claude generates page
Steve Reviews
Admin chat refinement
Published
/[company] live
☁️ Infrastructure
🗄️
Supabase
application_pages + page_views tables
Database
Anthropic API
claude-sonnet-4-20250514 for all AI steps
AI
👁️
RB2B
Visitor identification
Analytics
🌐
ipinfo.io
IP resolution
Identification
LAYER 3 — WHAT EACH PAGE CONTAINS
Layer 3What Each /[company] Page Contains
🔍
Company Research
Mission, pain points, key language
Content
🎯
Role Alignment
Mirrors JD language
Content
📊
Proof Points
3 selected from library
Content
🤝
HM Context
Connection points if LinkedIn provided
Content
🖼️
Clearbit Logo
Company logo from domain
Asset
📈
Visit Tracking
RB2B + page_views
Analytics
⚡ The Application Loop — 8 Steps
1. Paste JD + company URL2. Research Agent (company analysis)3. HM Research (optional LinkedIn)4. Draft Agent (page generation)5. Steve reviews in admin6. Chat refinement (iterate)7. Publish8. Live at growthbydesign.cv/[company]

Want to see what this looks like for your team?

Every engagement starts with understanding your specific situation.