Agentic Workforce.
Automated search operations for growth-stage companies. Keyword research, on-page optimization, brand-voice content pipelines, and multi-engine Generative Engine Optimization (GEO) tracking. We run the agents. You get the output.
Growth-stage marketing teams are rate-limited by the same repetitive workflows.
Keyword research takes weeks. Optimizing 200 meta tags takes an intern a month. Tracking AI-engine citations takes a human running manual queries across five different interfaces.
None of this has to be human work in 2026. The inputs are structured. The outputs are structured. The only reason it still takes a traditional agency a quarter is the hourly-billing model.
We automate the operational layer of modern search.
Five workflow categories. Each one is a Claude Skill (or a set of Skills) we run today against live client work.
- Keyword research at scale. Skills produce keyword universes with search intent, competitor mapping, and URL-level targeting. A traditional process that takes weeks ships in hours. seo-keyword-research, winston-keyword-mapping.
- On-page optimization at volume. Title tags and meta descriptions for hundreds of pages in minutes, not days. Schema injection and GEO chunking recommendations ship alongside. winston-geo-onpage.
- Content pipelines with brand voice. GEO-chunked articles written in your voice at the cadence you need. We run winston-geo-article for general use plus brand-specific writer Skills we have built for a handful of clients.
- GEO tracking and reporting. Skills generate hundreds of tracking prompts per brand. Agentic workflows run them across ChatGPT, Claude, Perplexity, Google AI Overviews, and Gemini. Delta reports land on a schedule. geo-prompt-research, geo-competitor-analysis.
- Holistic search strategy. Most search now happens outside Google. We pull Reddit, TikTok, YouTube, and LLM citations through Application Programming Interface (API) scrapers to give you the complete picture. aeo-reddit-strategy, aeo-youtube-optimizer, tiktok-refresh, aeo-content-auditor.
All of it plumbed with Model Context Protocol (MCP) so the Skills talk to your tools and our data sources in one stack.
The engagement has three beats.
- Month 1 baseline. We audit your site, your top competitors, and your current AI-engine citation state. Output: a 20-page baseline report plus a ranked backlog.
- Monthly delivery. You get a scoped output bundle every month. Typical mix: one keyword research refresh, on-page optimization for 50 to 300 pages, 4 to 12 pieces of brand-voice content, a GEO tracking report across the engines you care about, and holistic search monitoring (Reddit, TikTok, YouTube, LLMs).
- Quarterly review. We adjust the mix based on what moved and what did not. Retire workflows that are not earning their keep.
- Your access. Every prompt, every Skill output, every raw data file lives in a workspace you can read. No black box.
What you get.
- Keyword research workbooks and URL maps
- Title and meta optimization sheets (CSV or direct Content Management System (CMS) push)
- Brand-voice content drafts ready for your editor
- Monthly GEO tracking reports across ChatGPT, Claude, Perplexity, Google AI Overviews, and Gemini
- Reddit, TikTok, YouTube, and LLM citation monitoring dashboards
- Quarterly strategy review meetings
- A Notion or Google Drive workspace where every output lives
The stack we run.
Named tools, not vague capabilities.
Who this is for.
Growth-stage companies ($1M to $50M in revenue) with an active marketing function spending too much time on the repetitive operational layer. You have at least one senior person who can review Skill outputs and make editorial calls.
Who this is not for.
Teams without editorial capacity to review what the Skills produce. One-person marketing operations that need a generalist, not a volume engine. Clients who want "just ads" or "just blog posts" as a standalone buy.
And if your internal team cannot commit one senior operator to review monthly deliverables, this engagement is not going to land for you.
How we compare on the axes that matter.
Four vendor archetypes. Six honest questions.
| Axis | Traditional agency | AI-powered agency | Dev shop | /winston |
|---|---|---|---|---|
| Ships keyword research in hours, not weeks | No | Sometimes | N/A | Yes |
| Delivers on-page optimization at scale (hundreds of pages) | No | Limited | N/A | Yes |
| Monitors AI citation on a schedule (not ad-hoc) | No | No | No | Yes |
| Covers Reddit, TikTok, YouTube, and LLMs | No | No | No | Yes |
| Publishes methodology | No | No | Rarely | Yes |
| Your data and content in workspaces you control | Varies | Varies | Varies | Yes |
Frequently asked questions.
Do you build AI agents that we own and run ourselves?
No. We operate the Claude Skills and MCP connections on our side and deliver the output. Workbooks, optimization sheets, content drafts, GEO tracking reports, citation dashboards.
Some clients want owned skills and in-house deployment. That is a separate engagement, not this one.
How is this different from a traditional agency retainer?
Output per hour. A traditional retainer staffed with humans can ship keyword research in weeks and optimize 50 pages a quarter.
Our Skills ship keyword universes in hours and optimize hundreds of pages in an afternoon. The delivery cadence is the product.
What is MCP and why does it matter?
Model Context Protocol (MCP) is an open spec from Anthropic. It lets our Skills talk to data sources (DataForSEO, Ahrefs, Firecrawl, Apify, Google Search Console) and to client tools (Notion, Google Drive, Slack) through one standardized interface.
MCP is the plumbing. A single Skill can pull Search Engine Results Page (SERP) data, scrape competitor pages, and write results to your Notion in one pass, without bespoke glue code per client.
What AI engines do you track for GEO?
ChatGPT, Claude, Perplexity, Google AI Overviews, and Gemini.
We generate tracking prompts per brand with the geo-prompt-research Skill, run them across the engines on a schedule, and produce delta reports week over week.
Can you show me real Skill outputs before we sign?
Yes. The free 48-hour audit produces real Skill output against your domain. A baseline keyword map, an on-page optimization sample, and a first-pass GEO citation scan across the engines.
You see the deliverable format before you commit to the engagement.
Related reading.
- See the technical components. AI Marketing hub. Custom GPTs and Skills, Tool Integration, Content Workflows, Strategy Assessment, Agentic Web Transformation.
- Meet our own agents. About Winston. The section listing the named Skills /winston runs internally.
- Read the playbook. Why agentic workforce beats dev shops. The argument with a worked example.
> /winston workforce --audit=yoursite.com
Start with a free 48-hour audit.
Submit your site and top competitor when you book. We ship a prioritized 20-page audit in 48 hours and walk you through it on a call with John.