Run the growth loops a marketing engineer would build. Without the hire.

The marketing engineer is the new $250K to $1M growth hire: someone who builds AI agents that find customers, write outbound, test angles and get smarter every week. Finding the right one takes months. ProductClank puts those systems in your org tomorrow. Your agent triggers them, real people who already hold your audience’s trust do the work, and you only pay for actions we can verify.

12,000+ builders and creators in the network. 1,000+ campaigns run. 84% completion rate.

The six systems a marketing engineer builds, and how to run each one today

Greg Isenberg laid out the job in six systems on The Startup Ideas Podcast. Here is each one, and the campaign that runs it on ProductClank. The difference from doing it with agents alone: the last mile is done by a real person, so the reply lands and the content is not slop.

SystemWhat the marketing engineer buildsHow you run it on ProductClank

Customer truth

Run it now

A living file of what the market is telling you: the exact words customers use, ranked by how often they show up, with receipts.

Structured collection campaign. Real people in your audience answer a fixed set of questions or interview a customer for you. You get quotes and counts, not survey summaries.

Founder content engine

Run it now

Record the founder, extract the strongest ideas, watch which hooks hold attention, feed the winners back.

Content campaign. You post once. Creators and builders in the network clip, quote and repost it in their own words. Each post is verified on the platform it went out on before it is paid.

Outbound signal engine

Run it now

Timing beats copy. Watch for people who just posted about the pain you solve and reply while it is fresh.

Communiply. Discovery finds the posts on X, LinkedIn, Reddit and YouTube, drafts the reply, and a real community member posts it from their own account. No bot, no spam folder.

Creative and angle testing

Run it now

One offer, many angles, kill anything that does not earn attention, keep spending on the winners.

Content campaign with an objective. Give three angles. Creators each post one. The platform scores what landed, and you keep both the winning hook and the creator who found it.

AI search visibility

Run it now

Show up when someone asks ChatGPT, Claude or Perplexity for a tool in your category.

AI visibility panel. Real people ask the models about your category and submit structured answers. You see where you appear, where you do not, and what the models say instead. Skill-Leads corpus campaigns handle the offensive side.

Eval loop

Built in

Grade every output against replies, pipeline and conversion. Adjust the job spec. Rerun.

Verify, then settle. Every action is read off the platform it happened on before a reward is released. What got claimed, what got approved and what drew engagement feeds the next run.

How a loop runs

The same cycle every time, whichever system you start.

Your agent triggers it

Connect ProductClank over MCP and describe the loop in plain language. Or start it from the web app.

Real people do the work

Builders and creators from the network claim the task and execute it from their own accounts.

Every action is verified

The platform checks the result on the network where it happened. Self-reporting is not enough.

Rewards split automatically

Verified work is paid. Unverified work is not. You never fund a claim.

Three ways to get these systems into your org

Same loops. The difference is who designs them, how long until the first one is running, and whose audience it reaches.

Option 1 · Hire

Hire a marketing engineer

$250K to $1Mper year, plus three to six months to find them
  • Best case: one person who owns the whole system
  • Everything stalls while the seat is empty
  • Builds agents. Still needs distribution for the output to reach anyone

Option 2 · Done with you

We design the loops for you

Scoped engagementfixed scope, first loop live in weeks not quarters
  • We sit with your team, pick the one or two loops that move your number, and write the brief
  • We build and run them on ProductClank with you, then tune off real results
  • You keep the loops. We hand over the campaigns, the briefs and the network access
Book a call to scope itor email sales@productclank.com

Option 3 · Self serve

Run it yourself on the platform

Pay per verified actionno seat, no retainer, start today
  • Running tomorrow. Trigger a loop from your agent or the web app
  • 12,000+ people who already hold your audience’s trust as the execution layer
  • Every action verified before it is paid
Start a campaign →

Where the six systems come from

Greg Isenberg defined the role and the systems. Listen to the episode or read his thread, then come back and start one.

The episode: Making $$$ as a Marketing Engineer

Open on Spotify

The write-up: the four agents a marketing engineer builds

Read the full thread on X

Questions people ask before they start

What is a marketing engineer?

A growth person who builds AI agents and automated systems that find customers, write outbound, test creative and improve every week from results. Greg Isenberg calls it the forward deployed engineer of growth and expects the best ones to earn $250K to over $1M a year.

Growth engineer or marketing engineer?

Growth engineer is the older name from the growth hacking era and usually sits inside product: funnels, experiments, retention. Marketing engineer is the newer name for the same instinct applied across the whole marketing stack with agents. Most teams use them interchangeably, and this page answers both.

Can you design and build the loops for us?

Yes. That is the middle option above. We sit with your team, pick the one or two loops that actually move your number, write the briefs, and run the first cycles with you on the platform. You keep the campaigns and the network access when the engagement ends. Book a call and tell us what you are trying to move, or email sales@productclank.com.

Is this just bots posting for me?

No. Agents do the planning and drafting. The posting, replying and researching is done by real people from their own accounts, and each action is verified on the platform where it happened before anyone is paid.

Do I need to know how to code?

No. If you use Claude, Cursor, ChatGPT or any agent that speaks MCP, connect ProductClank and describe what you want. Everything is also available in the web app.

Which loop should I start with?

Outbound signal, through Communiply. It is the fastest to show a result you can measure: warm replies from people who were already talking about your problem this week.

One working loop beats five half-built ones. Start one.

Add the MCP endpoint to your agent: productclank.com/mcp