An AI marketing agent that works your growth plan with your team

An AI marketing agent is a model that takes a marketing goal, decides the steps and carries them out with tools, instead of handing you a list of ideas. The useful ones read real data, act on it and report what they did.

Most products sold under the name fall into two groups. Some are autonomous agents that run campaigns or post content on their own. Others are chat assistants inside a marketing suite that answer questions about your data. LogNorm takes a third route: the agent is a teammate inside a shared plan, with the same permissions and checks as a person, and a human signs off on anything that goes public.

That matters for a startup because the expensive mistakes in growth work are rarely about writing speed. They come from working on the wrong thing, shipping claims nobody checked, or fixing an SEO issue that never reaches the live site. An agent that only writes faster makes those mistakes faster.

LogNorm finds the work and ranks it; your agent does the work; LogNorm verifies and measures it. The agent is the coding agent you already use, connected to LogNorm as a remote MCP server.

LogNorm gathers the evidence: a crawl and SEO audit, a GEO audit for AI crawlers, keyword research, Google Search Console, tracked competitors and what ChatGPT, Gemini and Google AI Overviews answer your buyers. Every finding becomes a move, and moves are ranked head to head into a weekly Growth Plan. LogNorm ranks every move with its own decision model, whoever does the work.

Your agent reads that plan, claims a move so nobody else picks it up, and gets to work. Claims are atomic: if two agents reach for the same move, the first wins and the second is told who holds it. A claim lapses two hours after the agent's last action, so a move never gets stuck with an agent that stopped.

1. 1
   LogNorm finds and ranks

   Audits, Search Console, keywords, competitors and AI answers become moves, ranked against each other for this week.

2. 2
   Your agent claims and works

   It takes a fix, an article or a research task, does it in your repo or in LogNorm, and comments what it changed.

3. 3
   People approve

   Drafts go to review. A person approves and publishes to your CMS or merges the pull request.

4. 4
   LogNorm verifies and measures

   Fixes are re-checked on the live site, and shipped moves are measured at 28 and 90 days.

The agent gets the same freedom as a person on the team, inside the role and permissions you give it at connect time. Its role is editor or contributor, never above yours, and it works on one website.

Publishing is the line that matters most. The agent saves drafts and sends them for review; people approve and publish. See what the agent can do for every tool and the permission it needs.

You paste one sentence into your agent and click Allow once in your browser. There are no API keys to copy.

1. 1
   Paste the connect sentence

   Give your agent Connect to LogNorm: follow https://lognorm.com/connect.md. It adds LogNorm's MCP server and the LogNorm skill with one native command each.

2. 2
   Sign in and allow

   LogNorm opens a consent screen with your workspace, website and the usual permissions already filled in. Press Allow and join team, or Customize to change the role, permissions or idle timeout.

3. 3
   Restart once

   Restart the agent so the tools and skill load. It joins with a name of its own, such as Sam's Panda, and your teammates get a notification.

4. 4
   Give it the week

   Paste the starter prompt from the Agents page. Sub-agents can split the work, shown under their parent as Panda · Pip.

Start the agent in the codebase that builds your site, so it can fix templates, metadata, robots.txt, sitemaps and llms.txt directly. The connect page covers the commands for each client.

An autonomous agent decides and acts alone; an agent on a LogNorm team works a plan the team can see, steer and approve. The difference shows up in three places: who picks the work, who can see it, and who checks the result.

If you want hands-off posting at volume, an autonomous tool is the better fit. If you want an agent your team can trust with your site and your brand, keep a person on the publish button. Our post on AI teammates for growth goes deeper on that choice.

Everything the agent does is written down and credited to it, live on your dashboard. You can read its reasoning on the work itself instead of digging through chat logs.

• Its plan, decisions, changed files and pull request links as comments on each move.
• A one-line live status while it works.
• Credits on the work, such as Planned by Sam's Panda on a topic plan.
• Notes beside its drafts in the editor: what it changed and which sources it chose.
• Growth reviews it runs, saved as strategist sessions you can reply to.
• Every action in the activity log under the agent's name.

Remove an agent from the Agents page and its access ends at once; its claims go back to the team. Idle agents leave on their own after the period you picked.

The agent's own thinking and writing cost no LogNorm credits and need no AI key. Plans, briefs, drafts, comments, reviews and drawn SVG images run on your Claude or Codex plan.

What LogNorm runs for the agent costs what it costs in the app: research runs, keyword lookups, re-checking a fix, tracking a competitor or a generated image. Drafting and publishing are on paid plans, and new workspaces start with a 14-day Growth trial. Compare plans on the pricing page.