Automate Your Weekly Newsletter with Claude and Make
Last winter, I spent almost every Thursday evening huddled in a coffee shop near Hongdae in Seoul, manually copying links from my content feeds, writing short summaries, and pasting HTML into my email marketing platform. It took roughly three hours every single week just to assemble a single weekly roundup newsletter for one of my publishing sites. Multiply that across three niche sites, and I was spending nearly a full working day just formatting text and moving links around.
When you run multiple digital publishing channels as a solo operator, three hours spent on routine manual curation is three hours you are not spending on deep research or product development. I needed a system that could gather my bookmarked items from the week, feed them into an AI model capable of matching my brand voice, format the text cleanly, and drop a draft directly into my email provider.
By combining Claude’s API with Make (formerly Integromat), I reduced my weekly newsletter production time from three hours down to about ten minutes of manual review. Here is the exact blueprint I used to automate newsletter claude make workflows, along with the technical traps to avoid.
The Architecture of an Automated Newsletter
Before touching any software, it helps to understand how the data flows through this pipeline. A completely automated newsletter workflow requires four distinct stages:
- Data Collection: Gathering your saved articles, tweets, notes, or published posts throughout the week into a centralized database like Notion or a spreadsheet.
- Workflow Trigger: A scheduled trigger in Make that runs once a week at a designated day and time.
- AI Generation: Sending the raw collected text to the Claude API with a detailed system prompt that handles summary generation, tone matching, and structural formatting.
- Draft Delivery: Sending Claude’s output into your email service provider (such as ConvertKit, Mailchimp, or active campaign software) as a pre-formatted draft ready for review.
Notice that I specified dropping the content in as a draft rather than auto-sending it. Setting up an auto-send sequence without human review is one of the quickest ways to damage subscriber trust when an AI hallucination inevitably slips through.
Step 1: Setting Up Your Content Ingestion Hub
To feed Claude clean source material, you need a single place where you save interesting links or write quick notes during the week. I use a simple Notion database called ‘Newsletter Inbox’, but a Google Sheet or an RSS feed trigger works just as well.
Your database only needs a few basic fields:
- URL: The original link you want to feature.
- Title/Topic: A quick identifier for the story.
- Personal Take: A 1-2 sentence quick note on why this matters (optional, but gives Claude your original angle).
- Status: A select property set to ‘Ready for Newsletter’ or ‘Processed’.
During the week, whenever I read an insightful article or publish a new post on my own blogs, I save it to this Notion database using a browser extension. If you do not add a personal take, that is fine, but adding even twenty words of raw thoughts gives Claude enough context to write in a voice that sounds distinctly yours.
Step 2: Building the Make Scenario
Log into Make and create a new scenario. Make offers a flexible visual automation builder, with a free tier that includes a limited number of monthly operations. Paid tiers generally start in the $9 to $10/month range depending on usage, but you should always check the official Make pricing page for current limits and billing details.
Your scenario will consist of four main modules in a linear chain:
1. The Scheduled Notion Search Module
Set the first module to Notion: ‘Search Objects’. Configure it to run on a schedule—for instance, every Thursday at 9:00 AM. Set the filter condition to pull all database entries where Status equals Ready for Newsletter.
2. The Text Aggregator Module
Because Notion returns each row as an individual bundle of data, you need to combine them into a single string before sending them to Claude. Append a ‘Text Aggregator’ module directly after the Notion module. Map the Title, URL, and Personal Take fields into the text structure, separating each item with a clear divider line like ---.
3. The Claude API Request Module
Connect the official Anthropic Claude module (or an HTTP module configured for the API) to the scenario. You will need an API key from Anthropic. API usage is billed based on token consumption rather than a fixed monthly fee, which usually amounts to a few cents per newsletter issue. Check Anthropic’s documentation for exact per-token rates.
Select claude-3-5-sonnet (or the current equivalent production model) for the best balance of stylistic nuance and structural adherence. Claude 3.5 Sonnet excels at following strict formatting rules without adding generic AI buzzwords.
Step 3: Engineering the Claude System Prompt
The quality of your newsletter depends entirely on the instructions you pass to the API. Never ask the model to ‘write a newsletter’ without strict guardrails, or you will end up with generic text full of fluff like ‘In today’s fast-paced digital landscape.’
Here is a structural template similar to what I use in my own Make scenario:
- Role: Act as an editor for a minimalist email newsletter aimed at solo creators.
- Source Material: Use ONLY the provided list of links and notes below. Do not invent outside facts or statistics.
- Tone: Direct, practical, concise, and conversational. Avoid hype, metaphors, and grand claims.
- Format Requirements: Output clean, semantic HTML wrapped in simple paragraph and subheading tags. Output a subject line at the very top labeled
SUBJECT:followed by three variations. - Structure for each item:
<h3>Title linked to URL</h3>, followed by a 2-3 sentence breakdown explaining why it matters, followed by a bulleted takeaways list if applicable.
By specifying HTML output directly from Claude, you skip the step of converting Markdown to HTML later in the automation chain.
Step 4: Creating the Draft in Your Email Provider
The final module in your Make scenario connects to your Email Service Provider (ESP). Whether you use ConvertKit, Mailchimp, Brevo, or an automated WordPress newsletter plugin, select the module action named ‘Create Broadcast Draft’ or ‘Create Campaign’.
Map the text output from Claude into the body field of your draft template. If you configured Claude to output subject line options at the top of its response, you can use Make’s built-in text parsing tools to strip out the primary subject line for the campaign’s subject field, or simply map the full output into the draft body so you can manually select your preferred subject line during review.
Finally, add a concluding module that updates your Notion database rows, changing their status from Ready for Newsletter to Processed so they are not included in the following week’s execution.
Comparison: Claude API vs Alternative AI Engines
While you can run similar workflows using ChatGPT or Gemini, each language model handles structured text generation slightly differently. Here is a brief look at how they compare for newsletter generation:
| AI Engine | Strengths for Newsletters | Weaknesses |
|---|---|---|
| Claude (3.5 Sonnet) | Exceptional style matching, follows complex HTML constraints without extra formatting tags, avoids corporate fluff. | Can occasionally be overly conservative with creative phrasing if prompts are too restrictive. |
| ChatGPT (GPT-4o) | Fast execution speeds, strong integration options across multiple platform tools. | Tends to default to enthusiastic marketing cliches unless heavily penalized in the prompt. |
| Gemini (1.5 Pro) | Massive context window makes it easy to process dozens of long-form articles at once. | Can sometimes struggle with precise HTML markup output consistency in standard API payloads. |
My Take
When I first set out to build an automated newsletter system, my goal was pure hands-off automation: run a script on Friday, send the email on Friday night, never look at it. I set up a prototype that automatically sent emails directly to my subscriber list without my intervention.
It was a mistake. Within three weeks, a link broke, Claude incorrectly summarized a nuanced technical change in a software tool, and the tone felt slightly clinical because I had omitted my personal commentary notes during a busy week. Subscribers noticed, and my click-through rate dropped.
My take after running automated pipelines for over a year across multiple channels is simple: automate the collection, drafting, and formatting, but never automate the final publish button.
The value of this workflow is not that it completely replaces you; it is that it eliminates the friction of starting with a blank document. Instead of spending three hours searching for links and typing out HTML tags, you log into your email software on Thursday afternoon, spend five minutes reading through a fully formatted draft, tweak two sentences to add your own real-time voice, select a subject line, and hit send. That small human check maintains your editorial standards while keeping 90% of your time savings intact.
FAQ
How much does it cost to run a Make and Claude newsletter setup?
The cost depends on your email list size and workflow frequency, but for a weekly newsletter, platform expenses are low. Make offers a functional free tier, with entry-level paid plans around the $9-$10/month range for higher task limits. The Claude API operates on pay-as-you-go pricing based on tokens processed; generating a weekly draft typically costs less than a few cents per run. Be sure to check the official pricing pages for Make and Anthropic for current rates.
Can Claude format HTML correctly for email templates?
Yes. Claude 3.5 Sonnet is effective at generating clean standard HTML tags like paragraph, heading, list, and bold tags without including unnecessary markdown wrappers or broken markup. In your system prompt, explicitly tell the model to return raw HTML code without using markdown code fences (like “`html) so that Make can map the raw text directly into your email platform.
What happens if Make encounters an error during the automation execution?
If a API endpoint times out or a module fails (for example, if Notion is temporarily unreachable), Make will flag the scenario as failed and pause execution depending on your settings. You can configure error-handling routes inside Make—such as ‘Break’ or ‘Resume’ directives—or simply set up email alerts within Make settings to notify you whenever a scenario fails to complete so you can manually rerun it.
