The rise of the marketing engineer, and what your next twelve months should look like
Every time the technology shifts, the most valuable kind of marketer changes with it. We are one shift in. Here is what the next twelve months look like if you intend to be on the right side of it.
Founder of Agentique, where he leads delivery by turning commercial strategy and business analysis into specs the AI engineers build agents against.
The marketing industry is having an identity crisis. If your head is spinning from the sheer velocity of AI announcements, you are not alone. Plenty of non-technical marketers are asking themselves the same anxious question: how do I evolve over the next twelve months so I don't get left behind?
To answer it, stop looking at the latest AI widgets and look at the history of technology cycles instead. Every time the technology shifts, the most valuable kind of marketer changes with it.
Four waves, and the marketer each one made valuable
- Wave 1The Don Draper eraPsychology, storytelling and brand packaging, built to capture attention on print, radio and television.Paid a premium forThe brand storyteller
- Wave 2The digital marketerWebsites, SEO, analytics, pixels, Facebook ads. Suddenly everything after the click was measurable.Paid a premium forFunnel and acquisition literacy
- Wave 3The growth hackerProduct and software merged with marketing. The marquee names built loops around activation, referral, retention and pricing.Paid a premium forLoop and product thinking
- Wave 4 · nowThe marketing engineerAgents, data, code and taste. The execution layer is automated; the design of the system is the job.Paid a premium forSystems architecture and judgment
The fourth wave has a few names: marketing engineer, AI growth operator, forward-deployed marketer. The definition is the same whichever you use.
And here is the thesis underneath the whole shift. Anyone can open ChatGPT, ask for “ten generic social posts about my product”, and hit publish. When the cost of raw execution falls to zero, generic execution stops being a moat.
AI is about to make average marketing unbelievably cheap.
So: twelve months, four milestones. Each one is a mindset shift first and a system second. That order matters, because the systems do not survive the old mindset.
- Month 3Growth OS + Customer TruthStop starting from scratch. Give the company one place its marketing memory lives.
- Month 6Content and outbound loopsWrite agent job specs instead of creative briefs. Ship two engines against real signal.
- Month 9Testing and AI visibilityTrade campaigns for a compounding learning loop, and get cited by the models buyers ask.
- Month 12The growth cockpitGovern the engine. Report pipeline, not activity.
Month 3 · Establish your Growth OS and ingest customer truth
The mindset shift: from disconnected chats to compounding memory
Most teams use AI in fragmented, ephemeral ways. Open a chat window, generate a blog post or an email draft, copy-paste it into a document, close the tab. Next week, start completely from scratch.
The AI has no memory of which copy performed well, what the founder's voice actually sounds like, or what objections the sales team heard on a call yesterday. By month three, that has to end. You need a centralised Growth OS: a structured repository, a Git repo or honestly just a disciplined directory, where the company's marketing memory lives.
- Customer Truth/
- what-the-market-is-telling-us.mdrewritten weekly
- sales-transcripts/
- Content Engine/
- founder-voice-guide.mdthe voice agents must match
- Outbound Engine/
- ICP-and-trigger-events.md
- Creative Testing/results written back here
- Agents/
- job-specs/
The system: the Customer Truth Engine
Instead of relying on gut feeling, the first system a marketing engineer builds is a Customer Truth Engine. You deploy an agent to continuously read the messy, scattered signals: sales call transcripts, support tickets, CRM notes, product feedback.
What matters is the output standard. A generic AI summary like “customers want better collaboration” is worse than useless, because it sounds like insight. The engine instead maintains what-the-market-is-telling-us.md: concrete weekly realities, quote snippets, and raw event counts.
The difference is easiest to see on a narrow business. Take a vertical B2B software startup selling dispatch and invoicing tools to commercial HVAC contractors.
| Same week, same calls | Gut feelingThe traditional approach | Customer truthThe marketing engineer approach |
|---|---|---|
| What you notice | Five sales calls mentioned “emergency dispatch”, so dispatch is the theme. | Five calls mentioned emergency dispatch, but the three that closed spent twenty minutes on missed follow-up quotes after the technician left the site. |
| What you write | “Run your HVAC business more efficiently.” | “Stop losing replacement revenue after every service call.” |
| Why it works | It is true of every competitor in the category, so it competes on nothing. | It names a specific leak the buyer has already described out loud, in their words. |
The engine's job is not to summarise the calls. It is to notice which calls closed.
Month 6 · Deploy founder content and signal-based outbound
The mindset shift: from creative briefs to agent job specs
In the old system, a marketer writes a creative brief for a copywriter or an SDR. In this one, you write agent job specs, and you treat an agent the way you would treat a new human hire, which means being specific about the things new hires get wrong.
| Spec field | Every agent, before it shipsWhat it has to answer |
|---|---|
| Inputs and sources | Where does it look? Name the actual files and systems, not "our data". |
| Trigger | When does it run? A schedule, or the event that wakes it. |
| Filter | What must it ignore? Bad-fit accounts, off-the-record material, stale records. |
| Output standard | What does good look like? Brand guidelines, format, and the banned language list. |
| Human approval | Where does draft work go for sign-off before anything ships? |
| The metric that matters | How is success measured? Positive replies from qualified accounts. Never "emails sent". |
Six fields. An agent without all six is not a system, it is a habit.
The systems: two loops worth having early
- The Founder Content EngineReplacesGhostwriting from an empty document and hoping it sounds like the founderCapture the raw material founders already produce: podcasts, customer calls, quick audio memos. An agent pulls out the strongest, most opinionated concepts, checks them against the founder voice guide, and drafts social posts, video scripts and landing page angles from them.
- The Outbound Signal EngineReplacesSpray-and-pray sequences sent to everyone on a list at onceGood outbound is a timing problem, not a volume problem. Point an agent at trigger events: target accounts hiring for a role, opening a location, collecting bad reviews in public. It enriches the lead, drafts a message tied to the specific trigger, and files it for human approval.
Month 9 · Run creative testing and get visible in AI search
The mindset shift: from campaigns to continuous learning
By month nine your role stops being campaign management and becomes supervision of automated testing loops. Creative work is no longer a manual treadmill you have to keep walking; it is a system that compounds what it learns.
The systems: velocity, and being the cited answer
The Creative Testing Engine. Instead of three ad variations and a hope, you use image, copy and video models to spin up dozens of variations, hooks and landing page tests from your core positioning. The system records what happened, writes the results back into the creative testing repo, and uses those learnings to shape the next generation. The repo is the point. A test whose result is not written down is a test you will run again next quarter.
AI search visibility. Your buyers are no longer only searching Google; they ask conversational models for software and service recommendations directly. A marketing engineer does not answer that with generic keyword blogs. They rank buyer intent from Search Console, Semrush and Ahrefs data, then restructure the site's content architecture, write genuinely in-depth syntheses, and inject the founder viewpoints nobody else can copy, so the models index, cite and recommend them.
Month 12 · Govern from the growth cockpit
The mindset shift: from executioner to system architect
After twelve months you are not in the weeds writing drafts or maintaining spreadsheets. You are something closer to a blend of marketer, product manager, data analyst and systems architect. The work is deciding what the system should be doing, not doing it.
The system: the growth cockpit
Your control centre is a single weekly board that puts unvarnished business truth in front of the executive team.
Notice what is not on it. No impressions, no emails sent, no follower count. A cockpit that reports activity lets a broken engine look busy. This one reports pipeline, qualified replies and conversion: the three things that survive contact with a board meeting.
What this means for your career, and your company
- If you are a marketer. Stay a traditional digital marketer and your execution tasks get commoditised underneath you. Becoming a marketing engineer is how you stop being an expense line and start being the person who designed the pipeline engine.
- If you are a founder. You do not need a large marketing team. You need someone who can build and orchestrate an agentic growth engine that learns faster than your competitors notice.
The tools are commodities. The taste, the judgment and the systems architecture required to steer them belong entirely to the human behind the screen. The window to build the compounding version of this is open right now, while most teams are still copy-pasting out of a chat window.
We train non-technical marketing teams into this role, on your own workflows.
See the workshopsMore on agentic work
Business analysts, marketing experts and AI engineers in one team, building agents that understand the business and hold up in production.
Three offices in Belgium. The mapping sessions happen in yours.
AntwerpMechelsesteenweg 271b2018 AntwerpGhentDok Noord 59000 GhentBrusselsCours Saint-Michel 30a1040 Brussels