How to build a content library that never runs dry

Short answer: stop guessing what to post. Teach Claude how you talk by pointing it at your real client calls and emails, then show it what is already working in your niche, and have it suggest ideas every week in your voice. Four steps: two to learn your voice, two to learn what performs.

There are two ways to decide what to post. Guess, or look at what already works and say it in your own voice. This is the second one, set up so it runs every week without you.

Four steps. The first two teach Claude who you are, the second two show it what is working.

Step 1: Let Claude learn how you talk

Point it at your client meetings and your emails. Not your website copy, which is written in marketing voice, but the way you explain things when somebody is confused in front of you.

Read my client meeting transcripts and the last 12 months of my email.

Write me a short guide to how I talk:
- The words and phrases I use often
- How I explain things: analogies, examples, the order I put them in
- Sentence length and how formal I am
- What I never say

Keep it to one page. I am going to reuse this every time you write for me.

Save what it gives you. That page is the difference between content that sounds like you and content that sounds like everyone else.

Step 2: List the accounts you look up to

Five to ten Instagram accounts in your space that are genuinely good. Not the biggest, the ones whose posts you stop for.

Be honest here rather than aspirational. An account with ten times your following has a different game. Pick ones you could realistically be in two years.

Step 3: Connect Apify so Claude can see their numbers

Apify is a scraping service. The part you need pulls public Instagram data: posts, captions, likes, comments, views.

Connect it in Claude's Connectors. The free tier covers a handful of accounts a week, which is all this needs.

This is the step that turns opinion into evidence. Without it you are guessing at what did well.

Step 4: Run a weekly analysis

Every Monday, look at the last 30 days of posts from these accounts: [list them].

For each account, find the three posts that did best relative to their normal numbers.

Then tell me:
1. What the post was about
2. The hook, word for word
3. Why I think it worked: the format, the promise, the emotion
4. Whether the topic applies to my business

Do not give me the ones that only worked because of who posted them.

Now cross it with you. That is the actual output.

Take the winning ideas from this week's analysis and cross them with how I operate, using my voice guide and my meeting transcripts.

Give me 10 post ideas that:
- Use a proven angle from the analysis
- Are about something I actually do or believe
- Sound like me, not like the account I took the angle from

For each one, give me the hook and the three points.

Frequently asked questions

Is scraping Instagram allowed?

Apify reads public data, the same posts anyone can see. Keep it to public accounts and public numbers, and do not republish anyone's content.

What does Apify cost?

There is a free tier that covers a weekly run over a handful of accounts. You would only pay if you scaled it up considerably.

Will my posts end up looking like theirs?

Not if you do step 1. The angle is borrowed, the content is yours. Without the voice guide, yes, it drifts.

How long before it is useful?

The first run gives you ten ideas. It gets better after a month because it can compare weeks and see which of your own posts landed.

The other half of this system is turning every client meeting into three posts.

Related reading

Once you know what to make, how to turn Claude into your video editor covers getting it filmed and out the door. Your client calls are the other source worth mining: see how to turn client meetings into social content.

If you want this set up around your business, have a look at how I work.