The AI Content Stack I Actually Use (And What I Have Stopped Paying For)

People ask me about my tool stack often enough that this post is overdue.

I will tell you exactly what I use, what I have stopped paying for, how the pieces fit together, and what I would recommend to someone starting from scratch. No affiliate links. No tools I have been asked to mention. If I include something here, it is because I open it every week and it earns its cost.

I will also tell you what I tried and dropped — because the decisions to stop paying for something are often more instructive than the decisions to start.

The Primary Tools (What I Use Every Day)

Claude (Anthropic)

My primary AI writing environment. I use Claude for strategy thinking, long-form drafts, editing passes, and anything requiring nuanced reasoning about marketing positioning or audience psychology.

The reason I use Claude over GPT-4 as my primary tool: it handles complex instructions better, maintains context over longer conversations, and produces prose that is less recognisably "AI-written" in its default mode. For B2B marketing writing in particular, where voice authenticity and precision of language matter, Claude consistently produces better first drafts.

What I specifically use Claude for: rewriting client content that has been over-AI-ified by someone else (restoring voice and specificity), drafting LinkedIn post series from a briefing document, synthesising research into editorial frameworks, and generating structured content briefs from client discovery notes.

What I do not use Claude for: final publication without human editing. No AI output goes live without an edit pass that replaces generic examples with real ones, adjusts the voice to match the client, and removes the subtle structural tells that mark AI content to a trained reader.

Ahrefs

The SEO workhorse. I use Ahrefs for: keyword research and gap analysis, competitive content audits, site crawl and technical SEO auditing, and backlink research for content promotion.

The free Ahrefs Webmaster Tools tier covers most single-site needs: you get site audit functionality, keyword ranking data for your own domain, and backlink monitoring. The paid tier is necessary for competitive research — understanding what keywords a competitor ranks for, finding content gaps, and prospecting for link-building targets.

If I could keep only one paid SEO tool, it would be Ahrefs. Nothing else in the market provides equivalent depth of keyword and competitive data at a comparable price point.

Make.com (formerly Integromat)

The automation backbone. Make connects my content tools into workflows so that manual steps I would otherwise repeat become automatic.

Specific workflows I run: when a new blog post publishes, Make triggers a draft LinkedIn post in my content queue with the URL and key excerpts pre-populated. When a contact form submission arrives, Make logs it to Notion and sends an acknowledgement. When I tag a Notion entry as "ready to publish," Make adds it to the scheduling queue.

Make is not the kind of tool you notice on a day-to-day basis because it is doing its work invisibly. But its absence would cost me approximately three hours per week of manual work that currently happens automatically.

Notion

Where the content operation lives. I use Notion for: editorial calendar, content brief library, client context documents, swipe files, prompt library, and performance tracking.

The specific Notion setup I use for content: a database of all planned and published content, with fields for status, platform, target keyword, assigned persona, publication date, and engagement metrics. Each piece of content links to a brief. Each brief links to client context. Everything is connected.

This connectivity is the thing that makes Notion irreplaceable for content operations. A Google Doc library is faster to write in; Notion is faster to operate as a system.

LinkedIn (native)

I schedule and publish LinkedIn content natively — not through a third-party scheduler. The reason: LinkedIn's algorithm demonstrably reduces reach on content published through third-party APIs compared to content published natively. The difference is not enormous, but it is consistent enough across my testing to make native posting worth the extra friction.

For clients with multiple accounts or content teams, a scheduler becomes necessary. For a solo practitioner managing their own presence, native posting is the better choice.

What I Have Stopped Using

Jasper

I used Jasper in 2021–2022 when it was genuinely ahead of the market. Claude and GPT-4 have now surpassed it on every dimension I care about: output quality, instruction-following, voice preservation, and contextual reasoning. The subscription was not renewed when it came up for renewal.

Surfer SEO

A strong product with legitimate use cases — I stopped using it because Ahrefs covers the SEO layer I need and Claude handles on-page optimisation recommendations that Surfer previously provided. For a team running high-volume content with strict SEO targets, Surfer's content editor is genuinely valuable. For my workflow, it was redundant.

Buffer

Switched away from Buffer for LinkedIn after noticing a consistent organic reach reduction on scheduled posts versus native. Buffer remains excellent for Twitter and Instagram. For LinkedIn specifically, native beats scheduled in my testing.

Midjourney

I use Midjourney occasionally for client work requiring custom visuals. It is not in my core stack because most B2B content performs on substance rather than imagery, and most clients have brand visual guidelines that require human design work anyway. For consumer-facing brands, Midjourney is more essential — for B2B content, it is optional.

Clearscope and MarketMuse

Both excellent SEO content optimisation tools that I tested and ultimately dropped. My workflow uses Claude for content planning and Ahrefs for keyword research — the Clearscope/MarketMuse middle layer was adding a step without adding enough value to justify the subscription.

How the Stack Works Together

The workflow for a typical piece of client content:

1. —

   Keyword research in Ahrefs — identify the target keyword cluster, understand search intent, audit competitor coverage.

2. —

   Brief creation in Notion — build the content brief using the keyword research, the client's ICP definition, and the content positioning framework.

3. —

   Draft in Claude — load the brief, the client voice doc, and two to three examples of existing content that represent the target voice. Generate a first draft with a specific prompt that includes the brief.

4. —

   Edit pass — replace generic examples with client-specific ones. Adjust voice. Remove AI structural tells (parallel constructions, formulaic conclusions, hedge phrases).

5. —

   Publish natively — blog posts published via the CMS, LinkedIn posts published natively, email via the email platform.

6. —

   Make automation — captures the published URL, adds to the tracking database in Notion, triggers the social distribution workflow.

Total time per typical 1,500-word piece: approximately ninety minutes, versus four to six hours without the system.

What the Right Stack Looks Like If You Are Starting From Scratch

If you are building a content operation from zero, here is what I would recommend:

Month one: Claude (free tier is sufficient to start) + Notion (free tier) + LinkedIn native publishing. Cost: zero. This is enough to build and publish a content operation.

Month two: Add Ahrefs Webmaster Tools (free) to monitor your own site. Start understanding your keyword position before paying for competitive intelligence.

Month three onward: Evaluate whether your volume justifies Make.com automation and Ahrefs paid tier. These become cost-effective once you are publishing consistently enough to feel the friction of manual steps.

The most expensive mistake in building a content stack: buying tools before you have the workflow those tools are supposed to support. A tool is only as valuable as the consistent behaviour it enables.

The Honest Version

The stack is approximately fifteen percent of the system. The other eighty-five percent is clarity on what you are trying to say, consistency in showing up to say it, and editorial quality control that keeps your output sounding like a human with expertise rather than a machine with access to training data.

I have seen content operations with eight premium tools and zero consistent publishing. I have seen a founder with Claude, Notion, and a Google Doc swipe file outpublish agencies three times their size.

Get the workflow right first. The tools support the workflow — they do not create it.

FAQ

Is Claude worth paying for versus using the free tier? For occasional use, the free tier is sufficient. For daily use with long documents and complex prompts, Claude Pro ($20/month) is worth the upgrade — the expanded context window and priority access during peak times are meaningful for professional use.

What is the best tool for scheduling LinkedIn posts if you must use a scheduler? Taplio and Shield are the two I have heard most positive reports on from clients who manage LinkedIn at scale. Both are LinkedIn-specific and designed to minimise the algorithmic reach reduction associated with third-party posting.

Do I need a keyword research tool if I am just starting out? Not immediately. In the first three months, the constraint is publishing consistently, not keyword optimisation. Ahrefs Webmaster Tools (free) gives you enough data to understand what you are ranking for. Upgrade to paid when you need competitive intelligence.

How much should I budget for tools as a solo practitioner? My current monthly tool spend for the stack described above: approximately $120–150 per month. This covers Claude Pro, Ahrefs Starter, Make.com Starter, and Notion Plus. An argument could be made for starting at essentially zero for the first three months while building the habit.

---

See the full resource guide and prompt library I have put together →