AI Fluency

I don't just use AI. I build the tools my team uses.

I cut campaign documentation from two hours to five minutes — a 98% reduction — by pioneering Claude in my field marketing org, then scaled that process across the entire team. I build the calculators that put a real dollar figure on what a cyberattack costs a city, a utility, or a municipality running unmanaged credentials. And I design the Claude Skills that take the manual grind out of lead processing, microsite builds, and brand-compliant document production. I was one of the first here to bring Claude and Gemini into daily marketing work, and I didn't stop at using them.

See the programs

98%
Less time on campaign docs: Org-wide
Scaled the process: 3
Public sector risk calculators: 3
Claude Skills for my team

Putting a number on it

Risk calculators for the public sector

Government buyers don't move on urgency — they move on a figure they can defend in a budget meeting. So I built the tools that produce the figure: standalone, customer-facing calculators with estimates grounded in published research.

Municipal Revenue Loss

What does a city actually lose when its revenue systems go down?

Estimates the revenue a municipality stops collecting when the systems behind taxes, fees, and payments are taken offline by an attack — turning an abstract cyber risk into a dollar figure a city manager can act on.

Input: Name a city; the calculator does the rest
Built for: City managers, finance directors, and the people who approve security budgets
What backs the numbers: Published research and studies on municipal revenue and breach impact

Public Utility Revenue Loss

Same question, for the utility that keeps the lights on.

Applies the same model to public utilities — quantifying the billing and collection revenue at risk when a utility's systems are compromised.

Input: Utility or service area
Built for: Utility leadership and public sector technology decision-makers
What backs the numbers: Published research and studies on utility billing and outage impact

Standing Privileged Access

What is a municipality paying for credentials nobody is managing?

Calculates the cost of standing privileged identities in a municipality — the always-on administrative credentials that accumulate quietly and carry real exposure until someone puts a number on them.

Input: Municipality profile
Built for: Security and IT leaders making the case for identity governance
What backs the numbers: Published research and studies on privileged access exposure

The work

Claude Skills I designed for my team

A Skill packages a workflow — the steps, the rules, the examples — so it runs the same way every time, for anyone. These are the three I built to take the grind out of event marketing.

Skill 01

Lead Processing

The problem

Event leads arrive as messy exports — inconsistent fields, duplicates, job titles that don't map to anything. Cleaning them by hand is the slowest step between an event ending and sales getting a name to call.

What it does

Takes raw post-event lead exports and returns clean, deduplicated, qualified records ready for handoff — normalized fields, consistent formatting, and the routing logic applied the same way every time.

Who uses it

Anyone on the team running an event, not just the person who built it

Claude Skills
Lead qualification logic
CRM handoff

Skill 02

Microsite Generation

The problem

Every event wants a landing page. Waiting on a web queue means the page ships after the audience has already stopped paying attention.

What it does

Spins up event microsites from a program brief — structure, copy, and layout consistent with how our events are supposed to look, without a design or web ticket in the middle.

Who uses it

Event marketers who need a page live this week

Claude Skills
Event brief → page
Template system

Skill 03

Brand-Compliant Documents

The problem

Brand guidelines only work if people follow them. In practice, a distributed team produces a hundred slightly-off documents and someone spends their week correcting them.

What it does

Generates on-brand documents with the voice, formatting, and guideline rules built into the Skill itself — so compliance is the default output rather than a review step at the end.

Who uses it

The entire team — this one gets the most reuse

Claude Skills
Brand guidelines
Document templates

Where it shows up

Where AI does real work in my programs

Not a tools list. These are the parts of an event program where I have actually changed how the work gets done.

Designing Skills, not prompts

A good prompt helps one person once. A Skill encodes the workflow so the whole team gets the same result without knowing how it works.

Breaking a manual marketing process down into steps a model can execute reliably
Writing the instructions, examples, and guardrails that make output consistent run to run
Packaging it so a teammate can use it on day one with no prompt-writing skill of their own

Early adoption, then enablement

I brought Claude and Gemini into my work before there was a playbook, then turned what I learned into something the team could actually adopt.

Evaluated Claude and Gemini on real marketing work rather than demos
Translated what worked into shared tooling instead of keeping it as personal shortcuts
Coached teammates through the shift from 'ask the chatbot' to 'run the Skill'

Making the risk a number

Public sector buyers don't move on urgency. They move on a figure they can defend in a budget meeting.

Built calculators that turn cyber risk into lost-revenue estimates for a named city or utility
Quantified the cost of standing privileged access rather than describing it abstractly
Gave sales and marketing a concrete number to open a conversation with, instead of fear

Operations & automation

Removing the manual steps between an event ending and pipeline landing in the CRM.

Lead cleanup, deduplication, and qualification as a repeatable workflow
Post-event deliverables that assemble from source data instead of from scratch
SOPs and program kits written once, then run consistently across a distributed team

Knowing where it stops

Government audiences are the least forgiving place to get AI wrong. Scoping what stays human is part of the design.

Reviewing every output that reaches a customer, sponsor, or executive
Keeping sovereignty, procurement, and security constraints in scope from the start
Choosing the workflows where consistency matters more than creativity — and leaving the rest alone

Proof

And the programs behind them

The Skills came out of real work. Here is where that work happened — including the assistant running on this page.

Campaign documentation, rebuilt

Campaign documentation took two hours a pop. Multiply that across a field marketing org running 50 events a year and it is a genuine tax on the team's time.

What I built

Pioneered the use of Claude for campaign documentation, rebuilt the process around it, and then scaled that process across the entire field marketing organization rather than keeping it as a personal shortcut.

Outcome

98% reduction in campaign documentation time — two hours down to five minutes.

Claude
Process design
Team enablement

Federal AI Conference

Federal agencies needed a practical answer to how AI serves the mission — not a vendor keynote.

Read the case study →

What I built

Produced a two-day federal AI conference at Amazon HQ2 during an active government shutdown, with three parallel tracks for senior leaders, technical practitioners, and cleared personnel.

Outcome

The largest event the facility had ever hosted, delivered at record scale despite the shutdown.

Multi-track program design
Federal stakeholder engagement
AI narrative

This site's portfolio assistant

Recruiters skim. A static portfolio makes them hunt for the one detail they actually care about.

What I built

A Claude-powered assistant embedded in this site, grounded in my resume, impact data, and every case study — so a hiring manager can ask a direct question and get an answer sourced from real content rather than generated from nothing.

Outcome

Answers scoped strictly to documented work, with the whole portfolio compiled into its context at build time.

Claude Haiku 4.5
Anthropic API
Next.js
TypeScript

The stack

What I reach for

Models & tooling

Claude
Claude Skills
Gemini
Anthropic API
Standalone web tools

Marketing systems

Salesforce
Marketo
GoToWebinar
Asana
Google Analytics

Practices

Skill design
Risk quantification
Workflow automation
Brand governance
Team enablement
Human-in-the-loop review

How I think about it

The judgment stays human

Public sector audiences are the least forgiving place to get this wrong. My rules for working with AI are the reason I can move fast in it.

If only I can run it, it isn't a win

A clever prompt saved in someone's notes app helps one person on one day. That's why I build Skills — the workflow gets written down, packaged, and handed over, so the whole team produces the same quality without needing my help to do it.

Put the guardrails inside the tool

Brand compliance shouldn't be a review meeting at the end. I build the voice, formatting, and guideline rules into the Skill itself, so the on-brand version is the easy version and nobody has to police it after the fact.

The output is mine, not the model's

AI drafts. I decide. Nothing reaches a government audience, a sponsor, or an executive without my judgment behind it — which means I read every line and I own every number.

Public sector rules still apply

Government audiences come with sovereignty, procurement, and security constraints. In this space AI is a compliance question before it's a productivity question, and I scope the tooling accordingly.

Try it yourself

Ask my assistant anything about my work.

It answers from my real resume and case studies — no invented credentials, no filler. Ask it what I did at the Ottawa Symposium, or how I cut planning time by 90%.

Let's talk