I recently wrote that Growth Engineer is the new Growth Marketer and it really resonated with readers, so I’m back with part 2. TLDR on that essay: AI has spawned a new type of marketer that’s part growth marketer, part engineer. They aren’t solely thinking about experimentation, but also the infrastructure that enables rapid experimentation (once the job of an engineer).
As a startup founder, I’m running marketing, ops, sales, finance, recruitment, and the list goes on. But naturally as a career-marketer, I still spend a huge chunk of my time running paid ads, writing content and experimenting with our positioning. In parallel, I’ve been heavily obsessed with Claude and how it can empower my team.
This leads me to how I’ve been acting as more of a Growth Engineer these days, versus a typical Growth Marketer (my job when I was at Uber).
I thought it’d help to share what my daily workflows look like and how it differs widely from my time in growth at various tech companies.
Thinking in systems
The largest shift is in how I’m thinking about the structure of our marketing team, context sharing and experimentation.
Thought process pre- Growth Engineer era:
I have new messaging angles I want to test on Meta
Go to competitor ad libraries to research what they’re running
Write down all the angles we should test and send to design team
Launch campaign on Meta
Review results in 1-2 weeks and report to the team on wins/ losses
Rinse and repeat manually
The thought process post- Growth Engineer era becomes “How can I create a system that enables rapid experimentation on Meta?” So, the steps are instead:
Build an agent on Claude that automatically scrapes competitor ads weekly
Compiles a list of winning angles that we should test in PDF format
Design agent builds 50 variations of the same ad with copy tweaks
Another agent launches these ads automatically in Meta
That same agent reviews the ads daily and pauses immediate losers
Receive a report every Friday with messaging angle test results
In this setup, there’s a system in place, along with agents that are running the show with the messaging tests. This allows for much faster experimentation, that’s more dependable and runs 24/7 (I need my 8hrs of sleep lol).
As a former growth marketer, I was always thinking about how I could gain alpha on competition by using tools like Zapier to automate repetitive tasks. The issue with this is that those tools never really had a mind of their own. This changes with AI. We can now have our brains embedded within systems that are constantly thinking like us.
Context lakes will be the norm
I’ve been on marketing teams of two, all the way to 150 and the problem has always centered around context sharing– both within the marketing team and cross-functionally.
“Where’s the latest brand guidelines?!”
“What are the sales team hearing on their calls this week!!?”
“How many tests did we run last quarter and what were the results?!”
These questions always led to an endless pursuit of trying to locate items from people and scattered Google Drive links.
Growth Engineers now think in systems for their workflows, but also with how their teams share information. Instead of organizing a Google Drive with the latest marketing wins and ad collateral, this’ll all live inside of a system like Github.
Typical software engineers have been far ahead of marketers in how they operate within systems and shared repos within Github. This allows for version control, context sharing across machines and a source of truth where all software code lives.
In the same way that engineers use Github to organize code, marketing teams will use Github to house all their marketing collateral, experimentation data, brand guidelines, etc.
This allows marketing teams to:
Share a system of files and folders that constantly update with latest info
Work with the latest skills/ agents the team is running on, ICP data, etc
View a running log of all changes being made to the shared system (repo)
This style of team will outperform a scattered Google Drive or hundreds of Slack messages to get to an answer. Speed will win in this new era. Growth Engineers will own the build and maintenance of this foundational system.
Building and orchestrating agents
Regardless of the model or platform (Claude, GPT, Gemini, etc), it’ll be table stakes to build agents within these systems. They will run with schedules first, and then autonomously in the future based on its knowledge and parameters. For example, a Growth Engineer may set up a competitor research agent within a Research system that runs weekly at 8am. Eventually these agents will learn to run when they need and optimize themselves.
Competitor research agent → Research system → Marketing OS system
The flow is simple if you look at the multi-step workflow with the research agent feeding data into a research system (folder within Github), that then feeds to the entire OS system.
Now marketing teams will be able to all share information and can prompt their LLM of choice with questions like:
“What’s the latest research on company X?”
“Have there been any recent pricing changes for our competitors?”
No more searching, reading through files or asking others these questions. Instead these agents will arm teams with more information than we’ve ever had, just a question away.
And just like a Growth Marketer would empower their direct reports with knowledge, train them up and make them smarter—Growth Engineers do the same for these agents. That’s the orchestration portion of the job duty. It’ll become critical to understand how to properly build these agents, make them as efficient as possible (srry, no unlimited token glitch yet), and have them output the most reliable and important information.
Just because AI is heavily automated, the agents will only be as good as your last training session and tweak that you made. Rather than 1:1s with direct reports or a team, Growth Engineers will spend that time training the agents they’ve created, connecting them within the systems and analyzing their outputs. This’ll prevent AI slop and hallucinations.
On top of agent builds, there’s also the skillset needed to create bespoke software to support marketing that wasn’t available before (eng resources get prioritized to product). This role will require marketers to think through building versus buying software for their team. I don’t believe teams will build out their own CRMs and ESPs anytime soon, but they’ll build out custom ad hook creators, SEO content recyclers, influencer coordination pipeline dashboards, etc.
Putting it all together
We still have a Growth Marketer at the core, but they’re now thinking differently and operating within a new world of AI possibilities. Digital Marketers evolved into Growth Marketers with the rise of mobile, testing product virality loops, creating referral mechanisms, optimizing CAC/ LTV and working cross-functionally. It’s another radical shift, but will make marketing teams stronger than before.
When you put it all together, this is how a Growth Engineer looks in a modern marketing team:
Growth Engineer operator/ orchestrator
Foundational system (marketing OS)
Bespoke systems (experimentation, research, analysis)
Agents working inside these systems
This is in stark contrast with how the old way a Growth Marketer operated in a marketing team:
Growth Marketer executing on tests
Disparate file systems across Notion, Drive, etc
Some automations across Zapier, CRMs, ESPs
More junior marketers running research and analyses
One thing to note is that there can be multiple Growth Engineers on a team, similar to how there are multiple Growth Marketers on teams.
My message to all marketers is to embrace the change. It’ll lead to huge tailwinds on your marketing team. And finally, I predict that Growth Engineers will be the hottest new role in tech and a huge force for startups in the years to come.