Why we're building Oz

Before Oz was a platform, it was the toolchain Passionfruit's operators reached for when they had a Monday morning, a decaying PDP, and twelve hours to do three weeks of work. This post is the story of how that toolchain became a product - and why we think it's the shape marketing software should take from here.

Where this came from

Passionfruit has spent years running organic growth for 500+ brands - HP, Unilever, Marsh, L'Agence, Solawave, Goli, Mokobara, Shaadi.com, and others. Across those accounts, our teams have shipped over a billion dollars in incremental organic revenue. We didn't do it with dashboards. We did it with playbooks: Reddit pushes that earned citations inside ChatGPT, programmatic pages that opened net-new territory, technical fixes that landed before the next crawl, signal-based refreshes that caught declining pages a week before they dropped off the first page.

Over time, the playbooks stopped being documents. They became software - agents that held the institutional knowledge, read the data, and did the work. Oz is what happens when you open that software up.

Claude Code for marketers

Cursor and Claude Code changed what engineers do in an afternoon. You don't write scaffolding anymore - you tell an agent the intent and watch it plan, edit, test, and ship. Oz is the same pattern for marketing. A marketer opens the builder, names an agent, writes a contract in plain English, wires a trigger, and attaches the tools. Then the agent runs.

The tools are the ones marketers already live in: Google Search Console, Google Analytics, DataForSEO, and Passionfruit Labs for AI-search tracking - plus a hundred more. The contract is the mandate: "every Monday, audit the top 50 declining pages and queue refresh briefs in the Actions tab." Triggers are flexible: on-demand, on a schedule, on a webhook, or on a signal that fires when a metric moves.

What Oz actually does

Agents take action across every surface of SEO and AEO: content, on-page, technical, Reddit engagement, off-page, programmatic SEO, net-new page creation, competitor analysis, and AI-search visibility. They read the revenue data - including attribution per keyword, which nobody else has wired up - and propose the next move. They hand off work, escalate for approval, and run on signals instead of standups.

The job isn't to write more. It's to turn marketing into a control loop.

Ads, PDP, and catalog management are next. For now, SEO and AEO are where Oz is opinionated - because that's where the playbooks started. This blog will be the working notebook: fieldwork from the agents, wins, misses, and the operating system getting built around them. Welcome.