I treat AI as core infrastructure, not a tool. Over the last two years, the way I work has shifted from running individual queries against an LLM to operating a connected stack at the system level: production workflows that handle volume work, a reasoning hub that orchestrates everything end to end, and a clear point of view on which work humans should hold and which should move to agents.
Two pillars carry the operating model. AirOps is my production engine, the place where I run the five Playbooks that defend, extend, and monitor my content function automatically. Claude is the orchestration hub on top of that, connected via MCP to the data and tools where my work actually lives, doubling as the place I execute quick tasks, mock up designs, draft strategy, and write code.
Together, the stack collapses what used to be eight tabs and three calendars into one operating model. Below is how the layers work.
Layer one
AirOps: the production engine
I run five Playbooks in concert at AirOps. Each one handles a distinct surface of the content function, and together they cover the full lifecycle from ideation through optimization to defense.
Content optimization and republishing
An always-on workflow that monitors existing content performance against live search and LLM signals. When a piece falls below threshold, the Playbook identifies what changed, generates an updated draft using Claude inside AirOps, and routes the revision into a human review queue for republishing. The library defends itself instead of decaying silently between quarterly audits.
New search content
A multi-step build workflow that produces new SEO and AEO-optimized content with rubric-based evaluation. The Playbook takes a topic and target keyword set, runs research and competitive analysis, drafts an outline aligned to user intent and AEO-friendly structures, generates a draft, and loops the work through the quality rubric until it clears the bar or surfaces gaps for human edit.
Thought leadership exploration
A research workflow that surfaces emerging topics, market signals, and POV opportunities worth building around. The Playbook scans across signal sources, identifies the takes worth defending, and produces structured briefs that feed executive voice work, category-defining content, and original research framing.
Customer media mentions tracking
A scheduled Playbook that scans the open web and AI-generated answers for customer mentions, flagging new appearances and surfacing them into a structured grid for review. The output feeds case study sourcing, competitive intelligence, and authority signal monitoring across owned and earned channels.
Search prompt monitoring
Tracks the priority prompts the brand needs to win across ChatGPT, Gemini, and Perplexity. The Playbook monitors mention rate, share of voice, average position, and sentiment for each prompt on a running cadence, and feeds the data that triggers the content optimization and republishing workflow upstream.
I am also using AirOps to launch programmatic SEO pages at scale, designing the page systems, schema, and quality controls together as one motion rather than as separate engineering and content workstreams.
Layer two
Claude: the central hub
Claude is my central working partner, connected via MCP to GA4, Google Search Console, SEMRush, Ahrefs, AirOps, and Google Docs. The connectors collapse what used to be six to eight tabs into one thinking surface. A typical session might start with a question in Claude, pull live GA4 traffic and GSC query performance, cross-reference SEMRush and Ahrefs for keyword opportunities, query AirOps for current AEO health on the priority prompts, and draft the next round of optimizations directly into Google Docs for the team.
Beyond the orchestration work, Claude is where I run almost everything else day to day: quick research and analysis tasks I would have once opened ten tabs for, design mockups and visual ideation for landing pages and reports, code work for personal ventures and content systems, strategic thinking through cover letters and pitch documents, and almost every long-form writing task I ship.
The split with other AI tools is deliberate. Claude carries depth: the strategic thinking, the sustained reasoning, the long-form writing, the code, anything that benefits from holding context across a session. Other tools handle specific utility moments where they outperform Claude on the narrow task.
The metrics matter less than what they represent: an AI-native operating model running in production, in a regulated category, against real competitors, with measurable outcomes.
Justin Guinn | Content & Search Consultant 