Agentic content marketing platform

An agentic content marketing platform that does the work between your ideas.

Brief it like a colleague. The agents research, draft, check, publish, and come back to refresh what slips, on a schedule you set, with approval gates wherever you want a say.

Request beta accessInteractive demo$50 / MONTH · BYOK · NO LOCK-IN
Trusted by
Flux RoboticsLattéanuahCXLFiskalRoiSPRMRKT
Which post did best this month? Write a LinkedIn post about it.
Search Console, last 28 days
Top page: "What is a content operating system", 1,240 clicks
GPT-6 Luna
Knowledge base
Tone of voice, audience, hard rules
GPT-6 Luna
Brief for the writer
Angle, facts to use, format: LinkedIn
GPT-6 Luna
Draft
Written from the brief, in your voice
Claude
Brand voice and anti-slop passes
2 voice gaps fixed
Claude
Slop check
1 sentence flagged and rewritten, re-check clean
Jev
Saved to Library
Draft, ready for review
GPT-6 Luna

Definition

What is an agentic content marketing platform?

An agentic content marketing platform uses AI agents to carry out the work in steps. Each agent decides which tools to call, reads your brand rules and your data along the way, and hands back finished work at the points where you want to approve it.

The agents run multi-step jobs on their own, and the guardrails are yours to set. That's what turns an AI that answers questions into a content operation that ships.

One agentic run, start to finish

Loads your knowledge base and hard rules
Checks Search Console for the opportunity
Researches from sources you approve
Drafts in the format the job needs
Runs four slop filters on the draft
Waits for you, then publishes

What happens in one agentic run.

Ask for a blog on a keyword and this is what runs underneath, in order, with your knowledge base loaded at every step.

  1. 01

    Brief

    One line in the chat, a scheduled automation, or a message from your own Claude.

  2. 02

    Plan

    The assistant picks the tools and files the job needs and shows every call it makes.

  3. 03

    Research

    Keyword data, competitor pages, and cited sources from the domains you approve.

  4. 04

    Draft

    Claude writes in your voice, against your writing guide and your hard rules.

  5. 05

    Check

    Four slop filters, then a link check. Only the flagged sentences get rewritten.

  6. 06

    Approve

    You read it, change what you like, and publish to your site in a click.

Automations repeat the run on your schedule, and refreshes land back on the same URL.

A small team of models, each on its own job.

One model doing everything is either expensive or sloppy. So the lookups go to a fast model, the writing and judgement go to Claude, and a third model reads the output looking for AI tells.

LeadOn us

OpenAI GPT-6 Luna

  • Runs the conversation in the Marketing Assistant
  • Looks things up in Search Console, Analytics, your knowledge base and your library
  • Picks the right tool and writes the brief for the writer

Fast and cheap, so it takes the legwork. It never writes a word your readers see.

WriterYour key

Anthropic Claude

  • Every draft, rewrite, and edit
  • Blog research and fact checks, workflows, and automations
  • The brand voice and anti-slop passes

Anything that needs judgement runs here, with your full brand voice loaded.

CheckerOn us

TypeSafe Jev

  • Reads finished text for the patterns that give AI writing away
  • Checks it against your brand's own never-rules
  • Points at the exact sentences to fix

A model built to check writing, so it gives a straight yes or no for each pattern.

CodeFree

Plain code

  • Counts every em dash, emoji, and banned phrase
  • Tests every link and drops the dead ones
  • Applies fixes exactly where they were flagged

Where a rule is black and white, a model has no business deciding it.

Eight agents, one brand brain.

Marketing assistantCHAT
Brief it in plain English. It picks the tools and files.

The assistant reads your knowledge base, your Search Console data, and the live web before it answers. You describe the outcome you want and it decides which tools to call, then shows you every call it made.

Reads the knowledge base before it writes a word
Calls Search Console, web search, and your files
Every tool call is shown, not hidden
SEO blog writerBLOG
Keyword in, researched article out, published to your site.

Queue a keyword or a topic and the writer researches, drafts, lints against your hard rules, resolves every link, and publishes to your site. Updates to an existing post go back to the same URL.

Editorial and listicle playbooks
Hard-rule linter runs before you see the draft
Publishes and updates at the same URL
Social writerSOCIAL
One idea, reshaped per platform, in your cadence.

One argument becomes a LinkedIn post, an X thread, and a short-form script, each written to the platform's rhythm rather than pasted across all three.

Per-platform length and cadence
Same argument, not the same sentences
Banned phrases enforced everywhere
NewsletterEMAIL
A weekly issue assembled from what you shipped.

It assembles an issue from the week's published work, adds the context a subscriber needs, and leaves the send button to you.

Built from posts you actually published
Your tone of voice file, applied
Nothing sends without your approval
PRPRESS
Announcements written the way trade press reads them.

Announcements structured the way a trade journalist reads them: the news in the first line, the proof underneath, quotes that sound like a person.

News first, narrative second
Quotes drawn from your transcripts
Boilerplate held in the knowledge base
TranscriptsAUDIO
Turn a call into source material the agents can use.

Drop in a call or an interview and it becomes a file in the knowledge base, with your own phrasing preserved for every agent to draw on later.

Calls, interviews, and voice notes
Your phrasing kept intact
Becomes a knowledge base file
Research studioRESEARCH
Live web and search data, cited, before a word is written.

Live web results and your own search data, gathered and cited first, so the draft argues from evidence instead of from the model's memory.

Live web search with citations
Your Search Console data included
Sources attached to the draft
Thought leadershipOPINION
Your point of view, argued properly, not summarised.

It takes the position you hold, finds the strongest case for it, names the counter-argument, and answers it. Length follows the argument rather than a word count.

Starts from your position, not a consensus
Counter-arguments named and answered
Written to the argument, not to a length

Approval gates

Autonomy where you want it, a gate where you don't.

Chain research, drafting, repurposing, and review into one workflow. Add an approval gate after any step and the run pauses there until you say go, so the agents do the work and you keep the final call.

Build the workflow once in plain English and reuse it every week. Less copy and paste, fewer tabs, the same result every time.

Monday LinkedIn roundup
Built from one plain-English prompt.
1
About this workflow
Every Monday, pull last week’s top-performing LinkedIn posts in my niche, write a short commentary, and save it as a draft.
2
Pull top posts
Search LinkedIn activity from the last 7 days in your niche, ranked by engagement.
3
Digest what worked
Read the top 3 posts and note the hook, the format and the angle each one used.
Pauses for your approval
4
Write the commentary
Draft a short take in your voice, referencing the pattern found above.
5
Save to queue
Save the draft to your content library, ready for review before it goes anywhere.
+ Add step

On a schedule

Put the agents on a schedule and get on with your week.

Describe the job in plain English and pick when it runs. A daily social queue, a weekly long-form draft, a monthly newsletter roundup, a Friday sweep for broken links. Anything that needs your say waits at its approval gate until you open Contengi.

Automations
3 running on schedule, 1 paused.
Weekly thought leadership
Every Monday · 7:00am
Runs in 4 days
Daily social queue
Every day · 9:00am
Runs tomorrow
Monthly newsletter roundup
1st of the month
Runs in 12 days
Broken link sweep
Every Friday · 6:00am
Paused

Your agents, inside your own Claude.

ClaudeCONTENGI MCP
Queue three blogs for this week from the gaps in Search Console.
contengi.search_console_gaps
contengi.knowledge_base
contengi.queue_blog ×3
Three briefs written against your hard rules and queued. Nothing publishes until you approve it.
Inside Claude
Your agents show up as tools in Claude. Same knowledge base, same hard rules.
Sonnet for volume, Opus for judgment
Drafting and updates run on Sonnet. Strategy, briefs, and the hard calls run on Opus. You are not paying Opus prices to write a meta description.
The tools are already wired
Search Console, web search, your knowledge base, and your site's publishing endpoint are connected on day one. There are no MCP servers for you to run.
Your key, your data, no lock-in
You bring your own Anthropic API key and pay Anthropic directly for tokens, with no markup on top. Export the knowledge base and it is a folder of files you can take anywhere.

FAQ

Common questions

What is an agentic content marketing platform?

A platform where AI agents carry out content marketing work in steps. They decide which tools to call, read your brand rules and your data, and hand back finished work for approval. Contengi is one, with agents for blogs, social, newsletters, PR, research, video, and transcripts.

How is agentic content different from using ChatGPT or Claude directly?

A chat gives you one answer to one prompt, and you do the rest. An agent runs the whole job: it checks Search Console, researches sources, drafts, runs the quality checks, and waits at the approval gate, with your knowledge base loaded the entire time.

Will the agents publish anything without me?

Nothing goes live until you approve it. Workflows pause at approval gates, and a refreshed post only republishes over its original URL, so a second copy never appears at a new address.

Can agents run on a schedule?

Yes. Describe a recurring job in plain English, pick a schedule, and the automation runs it, from a weekly Search Console report to turning each new transcript into a blog and three LinkedIn posts.

Which models run the agents?

A fast model leads the conversation and handles lookups, Claude writes and makes every call that needs judgement, and a separate model checks drafts for AI patterns. The How it works page shows which model does which job.

What does it cost to run?

$50 a month in beta plus your Anthropic usage at cost. The pricing page shows what a typical blog, post, or newsletter costs in tokens.

Keep exploring

Related features and reads

In the platform

Automations

Put the agents on a schedule. Walk away.

How it works

Which model does which job, and why.

From the blog

Agentic content workflowsAgentic content, explainedAI content workflow automationAgents as a service, explainedHow to build a content system that runs without you hovering over it
See all posts

Brief it once. Approve what comes back.

Agents that know your brand, read your data, and wait for your go-ahead before anything goes live.

Request beta accessTry it out