# Create Instagram and TikTok posts with the AI subscription you already have

> Markdown version of https://postmakr.com/en/create-posts-with-ai, as of 21 October 2026. Publisher: postmakr, Vienna, hello@postmakr.com.

*Last updated on 30 October 2026, published by postmakr in Vienna.*

With postmakr you create Instagram and TikTok posts using the AI subscription you already pay for, such as Claude, GPT or Gemini. You connect your AI model to your postmakr workspace once via MCP, let it suggest ideas for your brand and turn the ones you pick into finished carousels, reels or stories. Every post lands as a draft in the dashboard, and nothing goes live until you approve it.

## What do you need?

You need two things: an AI subscription with a client that can connect to MCP servers, and a postmakr workspace for your brand. postmakr deliberately doesn't ship its own AI model. Production runs through your own account with your AI provider, so you don't pay for AI twice and you decide which model you trust.

postmakr is free during the beta, and access is available on request. You don't need to know how to code: connecting takes one command or a short entry in your client, and after that you talk to your model in plain sentences.

## Step 1: Request beta access and set up your workspace

Enter your email address on the [home page](https://postmakr.com/en/#contact). We'll get back to you with your access and set up your workspace together with you: you describe your brand, and your AI can read your website and create the brand and style guide in your workspace.

Every brand gets its own workspace. It holds the style guide, format catalog, post archive and learnings, which is everything your AI model needs later so the posts look and sound like your brand.

## Step 2: Connect your AI model via MCP

MCP (Model Context Protocol) is an open standard that lets AI models use external systems. postmakr is an MCP server, and your AI client connects to your workspace at this address:

```
https://mcp.postmakr.com/mcp/
```

### Claude Code in the terminal

A single command is enough, and you don't need a key for it:

```
claude mcp add --transport http postmakr https://mcp.postmakr.com/mcp/
```

On first access your browser opens with the postmakr login. Once you've signed in, the terminal remembers the connection. The guide [Connect postmakr to Claude Code](https://postmakr.com/en/mcp/claude-code) explains it in detail.

### Claude Desktop

Under **Settings → Developer → Edit config**, add this entry (it requires Node.js) and restart the app:

```
{ "mcpServers": { "postmakr": { "command": "npx", "args": ["mcp-remote", "https://mcp.postmakr.com/mcp/"] } } }
```

On first access you sign in once in the browser, and postmakr then shows up in the tool list. The guide [Connect postmakr to Claude Desktop](https://postmakr.com/en/mcp/claude-desktop) explains it in detail.

### Other AI clients

In other clients, add postmakr wherever MCP servers or custom connectors are managed. The setting is usually called "Custom connector", "MCP server" or "Integrations", and all it needs is the address above, because sign-in happens via browser login. If your client can't do OAuth, create a key in the dashboard under **Connection & API keys** and add it as a header:

```
Authorization: Bearer fp_YOUR_KEY
```

If your client doesn't allow custom headers, use the `mcp-remote` bridge as with Claude Desktop. Your key belongs to exactly one workspace and can be revoked at any time, and postmakr never stores keys from your AI provider. All variants are also on the [MCP page](https://postmakr.com/en/mcp).

## Step 3: Let it suggest ideas

Once the connection is in place, tell your model something like *"Suggest three ideas for my next posts."* or start the guided ideas flow. Your AI saves every idea in your workspace, together with a fully written hook as a possible opener and a recommendation for the best-fitting format.

For every idea you decide: implement, keep or discard. Ideas you keep stay in your workspace, so none gets lost, even when you're busy. You can add ideas yourself as well, and your AI builds the ones you implement.

## Step 4: Let it build the post

Your AI model then turns an idea into a complete post, for example when you ask *"Build a carousel from the first idea."* or start the guided `/create` flow. Your model doesn't freestyle. It follows your workspace's format catalog (carousel, animated carousel, reel, story or sharepic), and each format comes with its own spec for structure, slide count and dramaturgy.

As a style reference it uses your brand's style guide and the two or three most similar posts from your archive. It writes the copy, builds the slides and has them rendered on the postmakr server, as PNGs at post resolution or, for animated slides, as video. It saves the result as a draft in your workspace.

Ideas you have kept can be turned into posts at any time: tell your AI to implement one, and all that's left for you is the approval.

## Step 5: Review, revise and approve

Every post lands as a draft in the dashboard. There you see the finished draft and decide whether to approve it or send it back for revision. For a revision, you tell your model what should change, and it reworks the draft based on your feedback.

Review every post yourself before approving it, because AI-generated content can contain errors or inaccuracies, and you are responsible for what gets published. Once you approve it, the post moves into the archive and is ready to be published on Instagram and TikTok. Nothing goes live without approval, and that is the hardest rule in the system.

## What happens after publishing?

After publishing, every post's metrics flow back into your workspace: views, likes, comments, shares and the points where people drop off. They become learnings that your AI model receives with the next idea and the next post. That way each post gets a little more on target, and the dashboard shows you why.

How the system behind it works in detail is explained in the [documentation](https://postmakr.com/en/doku). Which brands and teams the workflow suits is covered in the [use cases](https://postmakr.com/en/use-cases).

## Frequently asked questions

### Can I create Instagram posts with Claude?

Yes. postmakr connects via MCP to the AI model you already use, for example Claude in Claude Code, Claude Desktop or claude.ai. Other models such as GPT or Gemini work as well if their client can connect to MCP servers. Your model builds the posts in your postmakr workspace, and you approve them in the dashboard.

### How long does setup take?

The MCP connection takes three steps and about five minutes. We set up the workspace for your brand together with you when you get beta access.

### What does it cost?

postmakr is free during the beta. For AI production you use your existing AI subscription, so you don't pay for AI twice. Fair pricing comes after the beta, and early users get special conditions.

### Does a post go live without my approval?

No. Every post lands as a draft in the dashboard, and only your approval makes it ready for publishing.

## Try it in the beta

Enter your email address on the home page and we'll get back to you with your access. The beta is free.

[Request beta access](https://postmakr.com/en/#contact)

Read on: [What makes postmakr different](https://postmakr.com/en/what-makes-postmakr-different), [use cases](https://postmakr.com/en/use-cases) the [MCP setup](https://postmakr.com/en/mcp) and the [glossary](https://postmakr.com/en/glossary).