Service 03 · AI Agents for Marketing

Shipped agents, not demos or prompts.

A library of 60+ marketing agents built, tested, and deployed inside real client stacks. Research, briefing, reporting, competitive intel. Automation that does the work, rather than automation that looks good in a sales call.

Why this service exists

Every marketing team has been pitched AI tools. Most have tried a few. Few have an actual operating stack of AI doing meaningful daily work.

The gap isn't the tools. It's the operator. Companies are being sold AI; what they need is someone who knows which work is worth automating, builds the agents properly, and maintains them when they drift. That's what this service is.

So we're aligned on terms

What an AI agent actually is.

Three things. If a tool only has one or two of them, it's a prompt, a workflow, or a dashboard — not an agent.

— 01

It runs autonomously.

On a schedule, a trigger, or an event. Not because a human clicked a button. Monday morning, a new competitor page ships, a new search query trends — the agent runs itself.

— 02

It produces a defined output.

A report, a brief, a flagged alert, an update in your CRM. Every agent has a clear "this is what it produces and who reads it" answer. If it can't be named, it can't be built.

— 03

It plugs into your stack.

Notion, Slack, Google Drive, your CRM, your analytics. Agents that produce output into a silo get ignored. Agents that appear where your team already works get used.

The library

Six categories. Sixty-plus agents.

Built for specific marketing jobs. Most engagements use three to six agents — enough to cover the biggest time-sinks without creating a stack nobody maintains. Full catalogue on request.

Keyword & topic research
12 agents
  • Intent cluster mapper — groups SERP results by underlying intent, flags shifts weekly
  • Emerging query monitor — surfaces new queries in your ICP's space before competitors cover them
  • Gap analyst — compares your coverage vs. top 5 competitors, produces prioritised gap list
Content strategy
10 agents
  • Brief generator — produces structured briefs tied to your voice guide and ICP
  • Angle finder — finds underexplored angles on your target topics using your positioning
  • Content refresh scorer — flags published pieces losing rankings, with refresh recommendations
On-page & technical
9 agents
  • Internal link optimiser — suggests new internal links as you publish, updates old pieces
  • Schema generator — builds and validates structured data for every new piece
  • Site health monitor — weekly crawl, flags broken links, slow pages, indexation issues
Link building & outreach
8 agents
  • Prospect researcher — sources qualified link prospects based on your ICP and target topics
  • Outreach drafter — drafts genuinely-researched emails (edited by a human before sending)
  • Backlink auditor — flags toxic backlinks, monitors for new referring domains
Competitive intel
11 agents
  • Competitor content tracker — watches competitor blogs, flags meaningful posts, not every post
  • Pricing & positioning watcher — monitors competitor pricing and messaging changes
  • Hiring signal reader — infers strategy shifts from competitor job postings
Reporting & planning
10+ agents
  • Weekly update writer — narrative-style weekly report pulling from analytics, GSC, and CRM
  • Monthly board report — assembles a board-ready marketing summary from your sources
  • Anomaly detector — flags unusual patterns in your traffic and conversion data
The agents above are examples drawn from the library. Full catalogue and per-agent specs available on request. New agents are added when a client need is general enough to warrant one; custom agents for specific needs are also built as part of engagements.
How an engagement runs

Four steps from scope to shipped.

Most engagements take four to six weeks from the first call to the first running agent. No six-month "discovery" phase.

Week 01

Scope

We look at your actual workflows, identify the biggest time-sinks, and pick the 3–6 agents to start with.

Week 02–04

Build

I build the agents, wire them into your stack, and test against your real data.

Week 05

Deploy

Agents go live. Your team is trained on what each one produces and where the output lives.

Ongoing

Operate

Monthly review. Agents that stop earning their keep get retired. New ones added as needs shift.

Pricing

Priced per agent.

Setup fee covers the build; monthly fee covers ongoing operation, monitoring, and maintenance. Typical engagement runs three to six agents. Cancel any agent quarterly.

— Setup per agent
One-time build fee

Scoping, build, integration into your stack, and testing against your data.

— Monthly per agent
Recurring operating fee

Keeps the agent running, monitored, and updated as your stack or priorities shift.

— Typical engagement
3–6 agents

Enough coverage for the biggest time-sinks, without creating a stack nobody maintains.

Honest answers

Questions I get. And how I actually answer them.

The things prospects ask on the first call, answered in advance so you don't have to.

Can't we just use ChatGPT and build this ourselves?

You can. The question is whether your marketing team will. Most of the companies who hire me have already tried — a few people built a few prompts, it worked for a week, then nobody maintained it and the habit died. Agents are a discipline problem, not a tools problem.

The value I add is that the agents stay alive because someone outside the team owns their uptime.

Do we need engineers on our side?

No. The agents are built on standard tooling (n8n, Zapier, custom scripts where needed) and plug into SaaS your team already uses. The only thing I need from your side is credential access to your analytics, CMS, and wherever the agents need to read or write.

What happens when an agent breaks?

I'm monitoring all running agents. If something breaks, I fix it — that's part of the monthly operating fee, not billed separately. For genuinely new failure modes I'll flag to you before changes, but the day-to-day uptime is my problem, not yours.

What if an agent isn't pulling its weight?

Kill it. Every quarter, we look at each running agent and ask: "if we switched this off, would anyone notice within two weeks?" If the answer is no, it gets retired. Paying for dead automation is worse than not automating at all.

Do the agents replace our marketing team?

No, and anyone selling them as team replacements is lying to you. Agents replace specific repetitive tasks — research, reporting, monitoring, first-draft generation. Your team gets hours back and spends them on strategy, editing, and judgement. Which is the work you want them on anyway.

Can we use these with your Content Engine?

Yes, and many clients do. Several of the library agents are effectively building blocks for the Content Engine — brief generation, competitor tracking, content refresh scoring. If you're running both services, there's no duplication; the agents are already part of how the engine works.

Next step

Let's see if we're a fit.

One conversation. No pitch deck. By the end of it, we'll both know whether this is worth building on.