Applying for: Director, Growth Marketing — OpenRouter
Paid, search, AEO and conversion are part of the job. I connect them to product adoption, retained usage and enterprise growth, then build the team around what works.
Amy Wilkinson
Ex-CMO · Built growth at WalletConnect · 11 years across technical products
$400B+
transaction volume scaled
Eleven years, six teams
$400B+
transaction volume scaled
300M+
ecosystem users reached
11 yrs
in marketing, 4 in leadership
10x
team growth at WalletConnect
Builder — Jan 26 — now
tap a line
Eleven years across six teams, almost all of it technical. Founding growth lead at WalletConnect, content and GTM lead at Radix DLT and Mode, growth at Cleo AI (800K to 2M users), founding marketer at Donut (150 to 35,000 users), then CMO at Aragon. Today, three live AI products I built and run myself. Developer infrastructure, fintech and AI tooling, not brand campaigns for consumer retail.
WalletConnect is the clearest case, and the parallel to OpenRouter is close enough to be useful. I built the growth function from a blank page: planning rhythms, campaign architecture, reporting, and a team that scaled 10x on top of it. The hard part was never volume. The protocol was already the default connection layer across hundreds of wallets and apps, so distribution was not the constraint. The work was converting embedded, largely invisible usage into a company developers could name, a category with an owner, and a brand they chose on purpose rather than inherited through a dependency. That meant ecosystem partners, technical audiences, product launches and a token launch, all sequenced against real usage rather than impressions.
I have run paid acquisition and SEO before, and I run them now. Roam operates paid social with rapid creative iteration. AI Jobs Board runs programmatic SEO plus community distribution, which reached 5K activity in its first two weeks with zero paid spend. AEO and SEO run across the wider product ecosystem, including FounderGrowthOS. These are live engines I operate weekly, not decks about channels.
At WalletConnect I built the planning cadence, campaign architecture and reporting system the function ran on, and kept it honest through a token launch, a 1.5K-attendee conference and every product launch. Across my current products the same layer is automated: planning, campaign structure, execution and reporting run through systems I designed, with GA and product data feeding back into what gets built next.
At WalletConnect the team grew 10x under me and eventually stopped needing me in every decision, which was the point. I build and operate the systems myself first, so I know what the job actually takes before I hire around it. That is still how I work: three products live, most of the execution automated, and nothing delegated that I could not run alone.
Three acquisition engines, live
Since stepping out of my last full-time role, I've built three live products, each running a different acquisition engine: BD systems, network-led acquisition and AEO; programmatic SEO and community distribution; paid social and rapid creative iteration. The planning, reporting and execution layer underneath all three is automated by systems I designed. They aren't side projects. They are how I keep a practitioner's read on what actually moves acquisition and usage in 2026.
10+ years of growth practice encoded into an automated execution OS: planning cadence, campaign structure, reporting and outbound, run by AI agents. Deployed to multiple clients as the marketing function they can't yet hire for.
growth model
BD systems + network-led acquisition + AEO
A programmatic acquisition product built on the FGOS sourcing engine: crawl, structure and publish AI jobs at scale, then seed the listings across the communities where jobseekers already gather. Two weeks, 5K activity, zero paid spend, and the channel data feeds back into the OS.
growth model
Programmatic SEO + community distribution
The consumer counterweight: an AI agent that aggregates flight, hotel and miles data into daily trip offers, wrapped in a brand built for rapid creative iteration at scale. Deliberately the opposite growth model to the other two.
growth model
Paid social + rapid creative iteration
I built a clone of myself that runs as an MCP server, so my team can call it in Slack, Claude Code or their editor for a judgement call at any hour. It doesn't replace me. It is AI-native operating leverage: my judgement available at the speed the work actually moves, instead of bottlenecked on one person in one timezone.
Judgement at scale (prewritten answers, not a live chat)
Cut it to three outcomes tied to usage, not activity. Name an owner for each, kill anything that can't be traced back to one of them, then publish the list so nobody has to guess what got dropped.
Fix the one blocking other people first, contain the rest with a date, and be honest in writing about which four are waiting. Ambiguity costs more than delay.
That they always knew what they were shipping this week and why it mattered, and that the operating system never turned into ceremony.
I know OpenRouter, and I know this shape of company. At WalletConnect I built growth for infrastructure that was already everywhere while the company, category and brand were still being deliberately built. I built the function from a blank page and grew the team 10x. OpenRouter has the same unusual opportunity, in a market I am already working inside. The advertised role solves an important part of it: acquisition, search, conversion and measurement. The larger job is owning the whole route from discovery to first credits, production usage, expansion and enterprise pull, then building the growth organisation around what the data proves. I can do the role as written. I am applying because I can build the version OpenRouter will need next.
I run paid, SEO and AEO engines right now, so the advertised remit is covered. It is not my whole identity. I am a growth leader who builds the function and stays hands-on in it, which is why I can connect those channels to product, ecosystem, retained usage and expansion instead of running acquisition as an isolated machine that stops mattering after signup.
First, I'd map the real journey and economics: source to signup, first credits, first production workload and expansion. Then I'd identify where growth is actually constrained and invest accordingly. Anything more specific before seeing the data would be a guess dressed as strategy.
No. Two already run near-autonomously, and FounderGrowthOS remains a small contracting structure outside working hours. They keep my skills current without competing for my calendar.