AI Marketing Content Studio
Ten picks for the marketing team running an AI-assisted content factory: keyword research, SEO audit, blog publishing, social scheduling, ad-creative generation, attribution, and privacy-first analytics — wired in a deliberate pipeline from idea to dashboard.
What's in this pack
This is the stack a real marketing team puts together when they decide AI is going to do the rough draft of every blog post, every ad headline, every social caption — and a human will still own the final call on what ships. Ten picks chosen to cover the full content factory: keyword research and SEO audit at the top, blog and CMS in the middle, social scheduling and paid creatives on the distribution side, and an attribution + analytics layer that tells you which of those bets actually paid off.
The pack assumes you're running multi-channel by default. The same brief becomes a long-form blog post, three social posts, two ad variants, and a newsletter blurb. The picks deliberately overlap on content generation (Content Marketer agent + Marketing Skills bundle) so you have one general-purpose agent for first drafts and a specialist skill library for the moments where you need a copywriting move you can't ad-hoc.
It also separates execution from measurement. Ghost / Postiz / Together AI handle output; Plausible and the Marketing Attribution Analyst handle the read-back. Keep that split clean — teams that conflate the two end up with dashboards their content people can't act on.
Install in this order (keyword research → SEO → blog → social → ads → analytics)
- Claude Code Agent: Content Marketer — the top-of-funnel agent. Turns a one-line idea or a tracked keyword into a content brief: outline, hook, target keyword density, internal-link plan, and platform variants. Start every project here so the brief exists before the writing does.
- Claude Code Agent: SEO Specialist (Technical SEO Audit) — the audit pass. Crawls your site, calls out crawl budget waste, broken canonicals, duplicate-title clusters, and missing schema. Run it monthly; run it again the week before any big launch.
- Marketing Skills — 34 CRO, SEO & Growth Skills for AI Agents — the specialist library. 34 prompt-skills covering on-page SEO, meta tag tuning, headline tests, landing-page CRO, email subject lines, ad-copy formulas, and lifecycle nurture. Use it as the deep-bench when the generalist Content Marketer agent feels too generic.
- Ghost — the publishing platform. Long-form blog, public SEO page, and newsletter from one post. Self-hostable and OSS, so the subscriber list and the URL structure are yours. This is where the SEO investment from picks 2-3 actually accrues.
- Postiz — open-source social scheduling. Schedule the same campaign across X, LinkedIn, Bluesky, Threads, Mastodon, Reddit, Pinterest, and YouTube Shorts from one calendar. Self-hosted so you control the data and aren't paying Buffer/Hootsuite per seat.
- Together AI Image Generation Skill for Claude Code — the ad-creative and blog-hero engine. Hosted Flux/SD with a clean API and predictable pricing. Generate 8 ad-creative variants in a minute, drop them into your A/B test, see which CTR wins.
- Lighthouse — Automated Web Performance Auditing by Google — the page-experience guardrail. Core Web Vitals, accessibility, SEO, and PWA scores from one CLI run. Wire it into CI so a slow page never ships, because slow pages quietly tax every other pick in this stack.
- Plausible Analytics — the read-back surface marketers actually look at. Privacy-first, cookieless, lightweight enough that it doesn't tank your Lighthouse score. UTM-tagged campaigns from Postiz and Mautic land here for the daily read.
- Claude Code Agent: Marketing Attribution Analyst — the synthesis layer. Pulls Plausible + ad-platform exports + CRM data into one weekly answer to "which content earned which conversions." Stops the "last-click wins" trap.
- Mautic — the automation backbone. Lead segments, drip campaigns, lifecycle emails, scoring. Marketing's outbound muscle, open source, integrates with Ghost for newsletter-to-nurture handoff.
How they fit together (ASCII marketing pipeline)
┌── Content Marketer Agent ─────┐
│ (idea → brief → outline) │
└────────────────┬───────────────┘
▼
┌── SEO Specialist + Marketing Skills ──┐
│ (audit + on-page polish) │
└────────────────┬────────────────────────┘
▼
┌── Ghost ──┐
│ (blog + │
│ newsletter)│
└─────┬──────┘
│
┌────────┴────────┐
▼ ▼
┌── Postiz ──┐ ┌── Together AI ──┐
│ (social) │ │ (ad creatives) │
└─────┬──────┘ └────────┬─────────┘
│ │
└─────────┬─────────┘
▼
Lighthouse (page exp guardrail)
│
▼
┌── Plausible ──┐
│ (UTM read-back)│
└────────┬───────┘
▼
┌── Marketing Attribution Analyst ──┐
│ (weekly synthesis: what worked) │
└──────────────┬────────────────────┘
▼
Mautic
(nurture the winners,
retire the losers)
The two critical joins are: Content Marketer → Ghost → Plausible (one brief shipped, measured end-to-end with UTM) and Together AI → Postiz / ad platform → Attribution Analyst (creative variants tied back to actual conversions). Get those joins right and "what should we make next quarter" becomes a data question instead of a meeting.
Tradeoffs you'll hit
- Content Marketer agent vs the Marketing Skills bundle. The agent is fast and general — drop in a topic, get a brief. The Skills bundle is 34 specialist prompts you reach for when generic fails (headline test, meta description rewrite, landing-page CRO pass). Run the agent for 80% of briefs; reach for the bundle when a piece needs to actually convert.
- Ghost vs WordPress. Ghost is a Node app focused on publishing + newsletter; WordPress is a kitchen sink. For a content-led marketing team that doesn't need 60 plugins, Ghost is faster, cleaner, and the same product is the blog and the email engine. If your team is married to Yoast/Elementor and a plugin ecosystem, stay on WordPress.
- Postiz vs Buffer/Hootsuite. Postiz is self-hosted, per-team flat cost, and the integrations cover the platforms a 2026 marketing team actually publishes to. Buffer/Hootsuite are SaaS, per-seat, polished UI, with enterprise integrations. For a 3-10 person team running their own stack, Postiz pays back the ops cost inside a quarter.
- Plausible vs Google Analytics 4. GA4 is free but loads a heavy script, requires cookie consent, and the reports got worse in the GA4 migration. Plausible is lightweight, cookieless, EU-friendly, and the dashboard is one screen a marketer can read at standup. The cost: GA4's deep funnels and enhanced ecommerce are not 1:1 in Plausible — for heavy ecommerce, run both.
- Mautic vs HubSpot. Mautic is the open-source alternative; you own the data and the cost is mostly hosting. HubSpot is the polished managed option with a sales tier wired in. Choose Mautic when marketing operates independently of sales tooling; choose HubSpot when marketing and sales need to share one record.
Common pitfalls (AI copy that smells like AI, attribution traps)
- Your blog posts start sounding AI. Symptoms: every intro opens with "In today's fast-paced digital landscape," every CTA is "unlock the power of." Cause: the model's marketing-blog priors leaking through. Fix: keep a
brand-voice.mdyour Content Marketer agent reads first, with explicit "never use these phrases," two sample paragraphs in your real voice, and three of your hot takes the model should defend. Update it after every editorial review. - Skipping the SEO audit until after the redirect. Running the SEO Specialist agent the day after a migration is too late — the canonical chaos is already indexed. Audit before launch, audit again two weeks after, set up Lighthouse in CI so regressions get caught at PR time.
- Postiz cross-platform copy without rewriting per platform. Posting the same 280-char tweet to LinkedIn and Bluesky reads like a bot. Use the Marketing Skills bundle's platform-rewriter prompts: same idea, native tone per platform.
- Together AI creatives that violate the brand system. AI image gen will happily render a brand-coloured button in a slightly-off red. Lock your brand hex + font choices into the system prompt; do a 30-second human pass before any creative ships to the ad platform.
- Last-click attribution lying to you. Plausible's default report tells you which referrer closed the deal; the Marketing Attribution Analyst's job is to tell you which earlier touchpoint set it up. Wire UTM source/medium/campaign consistently across Postiz, Mautic, and ad platforms — without that, the Analyst is guessing.
- Mautic drip campaigns that never get retired. Mautic makes it cheap to launch a sequence and easy to forget it. Calendar a quarterly review: any sequence with open rate below your baseline gets killed, not "optimised again."
10 assets in this pack
Frequently asked questions
Do I need all 10 tools or can I start smaller?
Start with four: Content Marketer agent for briefs, Ghost for blog + newsletter, Postiz for social scheduling, Plausible for the read-back. That gives you a closed loop: brief → publish → distribute → measure. Layer in the SEO Specialist agent and Lighthouse the moment organic search matters to revenue. Add Together AI when you start running paid creatives at any cadence. Add the Marketing Skills bundle when generic copy stops converting. Add the Marketing Attribution Analyst once you have at least three channels firing and need to decide where to double down. Add Mautic last, when the email sequences are the bottleneck. The full 10 only makes sense for a marketing team running blog + social + paid simultaneously.
What does this actually cost per month for a 5-person marketing team?
Realistic baseline: $20-50/mo Hetzner for Ghost + Postiz + Plausible + Mautic on one VPS (they all run as Node/PHP apps and stay within a small box), Together AI image gen ~$10-30/mo at indie-to-SMB volume, Lighthouse is free, and the four Claude Code agents/skills run on whatever Claude or Codex subscription the team already pays for. Call it $50-100/mo of infrastructure plus your AI subscription cost. The line that grows fastest is Together AI if you're A/B testing ad creatives aggressively; cap it with a hard budget on the API key. Compare this to a SaaS stack of HubSpot + Buffer + Ahrefs + Google Analytics 360 and the savings are real.
Will AI-written marketing copy hurt my SEO?
Not on its own. Google's position as of 2026: content quality is what matters, not whether a human or a model typed the first draft. AI-assisted content that has a clear point of view, original framing, and useful information ranks fine. What gets penalised: thin AI content with no editorial input, duplicate content across thousands of programmatically-generated pages, and pages obviously stuffed with a target keyword. The Content Marketer agent's job is the draft; your editorial judgment is the moat. Pair every AI draft with a human pass that adds at least one of: original data, a real opinion, or a specific anecdote.
Why Plausible instead of Google Analytics 4?
Three reasons. (1) Page weight: Plausible's script is ~1KB vs GA4's ~50KB, which matters for Core Web Vitals — the metric you're optimising for with Lighthouse in this same pack. (2) Privacy law: Plausible is cookieless and ships compliant with GDPR / CCPA / PECR by default, so no cookie banner gates your conversion flow. (3) The dashboard: GA4's report builder is powerful but the daily-read surface got worse in the GA4 migration; Plausible's one-screen dashboard is what a marketer can actually read at standup. The trade-off: GA4's deep ecommerce funnels and audience segmentation don't have 1:1 equivalents in Plausible. For content-led marketing, Plausible wins; for heavy ecommerce, run both.
Can the same brief drive the blog post, the social thread, and the ad headlines?
Yes — that's the whole point of wiring this pipeline. The Content Marketer agent produces a brief with target audience, hook, key claims, and platform variants. The long-form goes to Ghost. The social variants go to Postiz across X / LinkedIn / Bluesky / Threads with per-platform rewrites from the Marketing Skills bundle. The ad headlines + Together AI hero image become the paid creative. Every touchpoint carries a UTM that tracks back to one brief in Plausible. The Attribution Analyst then synthesises which channel earned the conversion, which feeds back into next quarter's brief selection. The human stays in the loop for judgement calls — final headline pick, brand-voice pass, creative review — but the manual repurposing work is gone.
12 packs · 80+ hand-picked assets
Browse every curated bundle on the home page
Back to all packs