The campaign that markets the system, run by the system

The problem

Soundtrak Consulting needed to launch a new offer: an AI studio run by a senior practitioner with a bespoke AI team of agents underneath. A consultancy launch needs a full kit. Sales deck, battle cards, outreach emails, a landing page, case studies, social video, long-form articles. There was no existing campaign infrastructure, no agency, and no marketing team beyond the practitioner himself, a former CMO of Netwealth.

So we gave the brief to the system. This case study is unusual for one reason: every asset described below was produced by the system the case study describes. That makes it the most verifiable case study we have. Every artifact can be shown, every gate decision walked, every cost line opened.

What was produced

Big idea

A senior CMO commands a seven-agent system, with judgment at every gate.

One brief, two runs: once through generic AI, once through Soundtrak's AI system and team, with both outputs published side by side. The visible difference between the two is the demonstration. The practitioner's deliberation behind that difference, the configuration, the send-backs, the brand verdicts, is the moat no generic tool can clone.

Key message

"Hiring a CMO is slow. Hiring an agency is generic. A practitioner with an AI team is the third answer."

Every concept in this system must survive a 15-second radio read, roughly 40 voiced words. Not because the ad will air, but because 15 seconds is a brutal forcing function for message clarity. If the proposition cannot survive a constrained read-aloud, it is not tight enough to ship. This is the script that locked the campaign's message:

Marketing plan

Four workstreams, one proposition. Each phase of the plan ships as finished assets, a selection of which are shown below.

Key assets

Real assets from the campaign, exactly as they shipped, every one produced by the system this page describes. 29+ assets, the selection below runs from storyboard to sales deck to the email that booked the first meetings.

The 14-slide pitch deck. Built around a 7-act narrative arc, shipped as HTML plus PPTX with presenter notes. Shown: the cover and the "third answer" argument slide. Approved 4 June 2026.

Storyboards before pixels. The Noise-to-Signal video series was approved as keyframe storyboards before a single frame was rendered. Shown: the V1 strip that locked the visual grammar for all eight finished videos.

The finished animation. The V4 "visible state" video, playing here (the most recent of the eight finals): noise resolves into an open eye with one red signal — the campaign's "no black box, always a human in the loop" line made visual. Code-driven animation, no stock footage, shipped in two aspect ratios for Reels, feed, LinkedIn and Shorts.

The email that booked the first meetings. The live partner outreach email: drafted by the operator, sub-edited by the system, approved at gate. The first sends produced two partner meetings, the campaign's earliest conversion evidence.

The results

The numbers

Every figure is pulled from campaign artifacts on disk; nothing invented. The AI cost is quoted as approximate, because that is what the ledger actually says. The campaign is in flight; market results land when the KPI window closes in late July 2026, and this page gets refreshed with them.

The phases

Every Soundtrak campaign walks the same six phases. The client approves at every red gate; the system does the work between them.

On this campaign: Brief approved, Concept selected, and the 29-asset Plan locked on day one. Production then ran sales-kit-first, so the consultancy could sell before it published. Two planned assets were cut by the operator mid-flight, which is the gate discipline working. And because the client is the system itself, everything behind this page (storyboards, brand verdicts, send-backs, the cost ledger) exists as an inspectable artifact we can open in the meeting.

"Seven AI agents. The CMO is still me."