Cofounder/CEO of Cape Networks (acq. by HPE/Aruba), grew to $10M/y
3 years building AI products: agents, evals, 300K+ users
Next: hands-on product builder/leader at an AI-first company
I've been building things since I convinced my 9th grade
teacher to let my friend Ross and me mod a video game
instead of doing our biology assignment. Only one problem:
neither of us could code! We spent months of long nights
figuring it out – and later founded two startups together.
The first company, Cape Networks, started in a basement in
Cape Town. We designed and built our first hundred hardware
units by hand – team-building consisted of smashing rubble
in the backyard with a sledgehammer. We sold to Aruba/HPE
and moved to San Francisco. I led about fifty people at
peak, including product, design, engineering, and sales –
and grew the business to $10M/y revenue. Aruba's founder
offered me a Chief Design Officer role to stay, but my
cofounders and I wanted to build something new.
My second company, Hunch, was a deep dive into AI: agents,
eval systems, and what it takes to make AI products that
actually work in production. We had real traction – 300K users in two weeks on one project – but couldn't find a
venture-scale business model and wound it down.
What's next
I want to lead and build, hands-on, making products people
love. Every product leader should be building with the
best AI tools right now, to get more done and keep up
with where the frontier is.
Two themes run through my career:
Making complex work accessible. Cape gave
non-experts instant visibility into network health. Neon
Data (pre-Hunch) made data exploration fluid for
non-analysts. Hunch put multiple AI models and tools in one place
anyone could use. I love making powerful tools feel simple
and fun.
Shipping under pressure. AI is rewriting
how every product works, which means a lot of product teams
are back in 0→1 mode whether they expected to be or not.
Navigating ambiguity and iterating fast is where I'm at my
best.
If you're working on an interesting problem, I'd love to
hear about it.
Visual AI workspace and background AI agents. Started
with Hunch, a beautiful canvas to create workflows with
any AI models. Evolved into Overclock: agents that
review each other's work, learn from feedback, and are
actually reliable in production. Hunch is still used by
thousands of people – here's a
demo
from when we won the Latent Space AI UX award.
Viral year-in-review product.
300K users in two weeks
(2024).
Idea to production in 4 days,
then
60+ versions in 17
days. Rebuilt it in 2025, charged for it, and drove $150K ARR to Coauthor, an AI writing partner, in two weeks.
Also designed and shipped a free
carousel
generator
in less than 2 days.
Neon Data
Super fast data exploration for regular humans.
Demo
of automatic visualization, instant cross-filtering,
auto-joins. More about the inspiration:
Latency is the Mind
Killer.
Cape Networks
Hardware + SaaS for network performance management.
Founder/CEO. Acquired by HPE/Aruba.
$10M/y revenue. I'm proudest
of how beloved our product was by
2,000+ customers including
Microsoft, Amazon, Home Depot, and Disney. Disney even
shot a
testimonial video in front of the
Falcon, a big moment for the little boy in me from Cape Town!
AI tools I'm using most
In addition to using Cursor Agent and Claude Code, I run
~1,000 prompts per week across different models, mostly in
Hunch and ChatGPT. It's how I develop intuition for what
each is good at.
A thinking partner for hard or nuanced
problemsClaude Opus 4.5
Research-intensive or “boil the ocean”
tasksGPT-5.2 Pro
Working with a ton of content fast &
cheaplyGemini 2.5 Flash
Complex/precise image analysisGemini 3 Pro
Wrangling messy, iterative projectsHunch
Async coding from my phoneCodex
Agentic coding at my deskClaude Code
Updated January 2026
Personal
I live in San Francisco with my wife and two daughters.
Outside work: piano, Star Wars LEGO and other projects with
my kids, sneaking in cricket and rugby highlights when I
can, and movies at the Alamo Drafthouse.