Greg Isenberg thinks the most valuable person in tech over the next 18 to 24 months won't be an engineer.
It'll be a marketer who builds like one.
He calls it a marketing engineer. Some people say forward deployed marketer. Others say AI growth operator. His take: the name will change, the job won't. It's one person doing the work of a whole marketing team with AI agents.
And his salary prediction is wild: $250K, $500K, up to $1M a year. Later in the video he goes further and says $1M is conservative for the best ones.
I watched the full episode so you don't have to. Here's the whole playbook, with the parts I'd actually steal.
What a marketing engineer actually does
Greg has started and sold three venture-backed companies. One in the web era, one in the social era, one in the mobile era. His observation: every time the technology changes, the most valuable kind of marketer changes with it.
His timeline goes like this:
→ Don Draper era: make people care through story and psychology. Print and radio.
→ Digital era (around 2005-2006): Facebook ads, landing pages, pixels. The best marketer acquired customers through channels you could measure.
→ Growth hacking era (roughly 2008-2011): activation, referrals, retention. Dave McClure's pirate metrics. Marketing moved inside the product.
→ Now: marketing engineering.
The definition he lands on: a marketing engineer is the person who turns market signal into pipeline using AI agents, data, code, and taste.
The new part is that last piece. The old skills still matter (positioning, customer understanding, judgment). But now you also build the system behind the marketing. A system that keeps learning.
Most companies already have the pieces: dashboards, sales calls, a CRM, content calendars. The problem is the learning is scattered. Sales hears one version of the market, support hears another, and the founder can't stop thinking about the one emotional customer call from last week. Everyone walks into the growth meeting with a different version of reality.
The marketing engineer pulls all of that into one system and turns it into growth.
The first thing to build: a growth repo
This is the most tactical part of the episode, and it costs nothing.
Greg says the first build is a GitHub repo (or honestly just a structured folder) called something like Growth OS. It becomes the company's marketing memory.
The problem it solves: most people use AI in random chats. You ask ChatGPT for 10 posts, copy one into a doc, and the work disappears. Next week the AI starts from scratch. What it needed was your performance data, your voice, the objections from sales calls, the language that actually got replies.
Inside the repo:
→ Customer truth: sales call notes, support tickets, churn notes, interviews.
→ Content engine: founder voice guide, winning hooks, what performed before.
→ Outbound engine: your ICP, account research, trigger events, approved angles. Even banned language (AI outbound gets weird fast if you let it talk like an overexcited SDR).
→ Creative testing: ad angles, landing page tests, offers, results.
→ Agents: the job descriptions for your AI workers.
Then the prompts change completely. Instead of "write me 10 LinkedIn posts" you say: read the customer truth file, read the founder voice file, read the last five posts that drove qualified replies, and draft five new posts around pains buyers mentioned this week.
Totally different level of output. The agent has real context.
I run my own version of this for Profitable Founder. This blog, the outreach, the content pipeline: agents reading from files that get corrected every week. It compounds in a way random chats never do.
The tool stack (and why it matters less than you think)
Greg's lineup, mapped to lanes:
→ Grok bot as the growth OS that lives close to the internet. One bot watches competitors, one watches customer language on X and Reddit, one watches creators in the niche, one watches ads and landing pages.
→ Claude and Codex to build the repo, generate landing pages, write scripts, build small internal tools.
→ Hermes-style workflows for scheduled operations with memory and approval. Every Monday, build me a market brief. Every Friday, review the experiments.
→ Creative models (he names fal.ai and Higgsfield) for ads, thumbnails, mockups, video concepts.
→ Local AI for anything sensitive: private transcripts, pricing plans, regulated notes.
His caveat, which I agree with: the tools will keep changing. The workflow is the thing to learn.
If you want a deeper dive on the agent side of this stack, I broke down his earlier episode with Cody Schneider in marketing agents are too good now, and the origin of the "forward deployed" framing in the forward deployed engineer story.
Agents need job specs, not vibes
My favorite idea in the episode: write out every agent's job like you're hiring a person.
Here's the data source. Here's when you run. Here's what you filter out. Here's what good output looks like. Here's what needs human approval. Here's the metric that matters. And here's where you write the result so the system gets smarter next time.
His SEO example shows the gap. The beginner version is asking AI to write a blog post about a keyword. The marketing engineer version: the agent checks Google Search Console, pulls keyword data from Ahrefs or Semrush, looks inside the CMS to see if the post already exists, ranks opportunities by volume and buyer intent, researches what's ranking, drafts the post with the founder's point of view, writes the meta title, suggests internal links, and sends the whole thing for approval.
Or the competitor engager agent: every weekday morning, check 20 LinkedIn accounts, pull the people who commented on new posts, enrich them, drop the bad fits, and draft 10 messages tied to the specific post they engaged with.
The metric is positive replies from qualified accounts. Messages sent is activity. Qualified replies is signal.
And you train agents like new hires. Small tasks first. Watch the work. Correct mistakes and add the correction to memory. If the outbound agent writes a first line that sounds fake, that becomes a rule in the repo. If the customer truth agent makes a claim with no evidence, new rule: every insight needs a quote, a link, or a source.
The HVAC example: six systems
To make it concrete, Greg walks through a vertical SaaS startup selling to commercial HVAC contractors. Real B2B market, messy workflows, specific language.
The sharp insight: the marketing problem is never "we need more content." It's which pain gets the owner to book a demo. Maybe dispatch chaos. Maybe late invoices. Maybe the technician finishes a service call, spots a replacement opportunity, and the follow-up quote never gets sent.
"Stop losing replacement revenue after every service call" beats "run your HVAC business better" every single time.
The six systems he'd build around that company:
1. Customer truth. A file called what-the-market-is-telling-us.md (he tweeted this idea and it went viral) that updates from sales calls, tickets, churn notes, even Stripe movement. Not vague summaries. Receipts. A good memo reads like: five sales calls this week mentioned emergency dispatch, but the calls that converted all talked about missed follow-up quotes.
2. Founder content engine. Record the founder, pull from podcasts, extract the strongest ideas, watch what performs. One insight becomes five posts, a short video, a landing page line, a cold email angle, and a lost-revenue calculator.
3. Outbound signal engine. Bad outbound starts with a spreadsheet of names. Good outbound starts with timing: who raised money, who's hiring dispatchers, who just got a bad review. Painkillers, not vitamins.
4. Creative testing engine. One offer, 20 hooks, 10 ad angles, results recorded. "Facebook ads don't work for me" usually means "I didn't test enough creative."
5. AI search visibility. Over a billion people ask ChatGPT things now (Greg cites Sam Altman's number). Is your company even understandable to those systems? Getting cited by AI is a wide-open lane.
6. The growth cockpit. A weekly memo: what content worked, which objection came up again, what test won, what to try next. His sample line: the lost-replacement-revenue angle drove fewer clicks than the dispatch angle, but twice as many demo requests from owners with more than 20 techs.
That last memo is the kind of thing an executive pays real money to wake up to.
Four ways to get paid for this
→ Become the person inside a company. This work sits directly next to revenue. Someone who can say "I'll save $2M in spend and double conversion" gets to name their price. That's the $500K hire logic.
→ Consulting. Embed with a founder for 30, 60, 90 days. Build one growth system, sell the outcome. Greg's range: $5K to $30K a month depending on what you build.
→ Productized service. Pick one wedge and repeat it: outbound signal engines for vertical SaaS, founder content engines for B2B CEOs, customer truth repos for seed-stage startups. The tighter the wedge, the easier to sell and deliver.
→ Software. The biggest outcomes, but he'd start with services first. Build the same system by hand for 5 to 10 companies, notice the pain that repeats, then productize. That's also how you avoid building something nobody wants.
And they stack. Consult to learn, productize the pattern, turn it into software at the end.
The 30-day plan if you're starting from zero
→ Week 1: audit one real company. Yours or a friend's. Study the website, the offer, the ICP, the founder's content, sales calls if you can get them. Output a market map: who's the customer, what pain do they describe, what words do they use, where does the funnel leak.
→ Week 2: build the growth repo. Five files: customer truth, founder voice, experiments, agent jobs. Paste 20 real customer notes and ask the agent for one job: tell me what changed, show me the receipts, suggest one test that creates pipeline this week.
→ Week 3: build your first machine. One system, not five. One working system beats five half-built ones.
→ Week 4: results. Did replies improve? Did meetings get booked? Did conversion lift?
At the end of the month, your case study sounds like: I audited this company, built a customer truth repo, found one high-intent pain they didn't know about, turned it into an outbound engine that shipped 75 targeted messages, got 9 warm replies, and booked 3 calls.
That's how you get hired. That's how you get clients.
My take
The part of this episode that will age best isn't the salary prediction. It's one line near the end: the agents are going to be a commodity. Your judgment about what to point them at is the moat.
I've felt this building Profitable Founder. The agent that writes is cheap. Knowing which customer sentence deserves an article, which angle is sharp and which is generic, that's still the job.
So if you're a marketer, this is how you become the person your company can't run without. If you're a founder at $5K-$50K MRR doing all 15 jobs yourself, marketing is probably the one worth systemizing first.
The window is open because most people haven't built the machine yet.
FAQ
What is a marketing engineer?
A marketing engineer is one person doing the work of a marketing team by building AI agent systems: a shared growth repo, agents with written job specs, and feedback loops that turn sales calls, tickets, and test results into content, outbound, and positioning. Greg Isenberg defines it as turning market signal into pipeline using AI agents, data, code, and taste.
Do you need to know how to code?
Not really. Greg says even non-technical people can build the growth repo (it's markdown files in a folder). Tools like Claude and Codex handle the building. What you can't skip is marketing judgment: knowing your customer, writing sharp positioning, and spotting when agent output sounds fake.
How is this different from growth hacking?
Growth hacking used product and data to build loops (activation, referrals, retention). Marketing engineering keeps those fundamentals but adds a system that learns on its own: agents connected to live business data, with memory, running on schedules, getting corrected like new hires. The growth hacker ran experiments. The marketing engineer builds the machine that runs them.
How much can a marketing engineer earn?
Greg's prediction is $250K to $1M a year for the best ones inside companies, and he calls $1M conservative. Independent routes: consulting retainers of $5K to $30K a month, or a productized service focused on one system for one niche.
Want more breakdowns like this? Every week on the Profitable Founder Podcast I talk to bootstrapped founders making $100K to $10M a year about exactly how they did it. Real numbers, real playbooks.
