DiveIn

A 2-person team doesn't have time to run marketing. So we built one that runs itself.

A 15-agent marketing department that reports to the two people who own it.

15
agents, each with one job
1 CMO
agent that reviews every output
Telegram
seamless communication
1Problem

Two people, the workload of a company

Okeanos wants to 4x its installs, but two people run marketing, sales, ads, content and follow-up, on top of real installs.

MarketingSalesAdsContentFollow-upbacklog ↑2 people can't keep up

Leads go cold, opportunities lost

A lead goes cold before anyone replies. Ad budget spends on what isn't working. Content never gets made. Not because no one cares, because there's no one left.

Today

Jordan M.

Pool quote · inbound

NEW
0hno reply
COLD
2Solution

Meet DiveIn, a marketing department that runs itself.

What DiveIn is, and the CMO agent that keeps fifteen of them in line.

The team

Fifteen agents, each with one job

One scores leads, one writes social, one watches ad spend. A full marketing department that runs the function for a two-person company.

A marketing department, minus the headcount.

The boss

A CMO agent above them all

It sets the weekly plan, reviews every agent's output, catches conflicts, and brings the humans in when it matters.

AI

DiveIn CMO

online · Telegram

Weekly plan is set. 3 campaigns live, content queued through Friday.

One thing needs you: a 1-star Google review came in. Reply drafted — approve to send.

Approved. Send it.

Reports in like a coworker, over Telegram.

3How it works

Fifteen specialists, one CMO, one company voice.

Each agent owns one job. The CMO plans the week, reviews every output, and catches conflicts before they ship.

2 owners · Telegram

AI CMO Orchestrator

plans · reviews every output · catches conflicts

Creative Supply Chain

Growth & Traffic

Sales & Nurture

Intelligence & Success

4AI-enabled redesign

Almost everything automated. The line is risk.

The same lead, by hand and through DiveIn. Routine scoring and replies run on their own; a reputation-risk reply gets drafted, then waits for an owner.

By handWith DiveIn
Lead in
missed
Lead lands while both owners are on an install
✦ AI does thisinstant
The Qualifier scores the lead the moment it lands
Score
no triage
No one scores or prioritizes it
✦ AI does thisauto
Routed by fit and urgency
Reply
✕ Where it broke48h
First reply, if any, comes two days later
✦ AI does thisminutes
A follow-up in Okeanos's voice goes out
Escalate
lost
The lead has already gone cold
Human signs offby design
Reputation-risk replies wait for an owner
5Success metrics

The numbers I'd track in production to know it's working.

Median lead response time

From 48+ hours to minutes.

Cost per qualified lead

Where shifted ad spend pays off, or doesn't.

Content cadence hit rate

Posts that went out versus the plan.

Human-escalation rate

How fast owners clear the review queue.

6Learnings

What I'd do differently next time.

If I ran this again, I wouldn't open Cursor first. I'd map the journey, be honest about what the company is ready for, and decide the human-machine mix before writing a line of code. Then build agents, probably fewer.

01

Build a dial, not a switch

We chose full autonomy for them. Higher-placing teams gave the business a toggle and let them choose what to hand over. We decided for the client instead of letting them decide.

02

Match the build to what they're ready for

Okeanos was running six guarded agents. We jumped them to fifteen autonomous ones. That's not the next step, it's three.

03

Pitch the experience, not the architecture

We led with the stack because it seemed impressive. They were asking whether it helps their company. The most impressive system and the right one aren't the same.