Published
11 MIN READ
Zernio + Claude: AI social media post automation
Zernio is a social media API with an MCP server that lets an agent like Claude publish, schedule and measure posts on 16 platforms through a single connection. At Wama, it closed the last link in the content process: Claude writes and adapts, a person approves, and the tool publishes at the right time.
This article shows how this social media post automation works in practice, with the process that has been running here since August 25, 2026, the numbers from my account, what Zernio costs and what is still a human decision. It is a behind-the-scenes look, not an ad: the mistakes are here too.
What is Zernio and why is it part of Wama's process?
Zernio is a unified social media and messaging API, with a hosted MCP server, that lets an app or an AI agent publish, schedule and analyze content on 16 platforms, including Instagram, LinkedIn, X, Threads, TikTok, YouTube, Pinterest and Reddit. It is part of Wama's process because it speaks Claude's language: instead of a dashboard to click through, it exposes tools the agent calls on its own.
According to Zernio's official page, updated on August 24, 2026, the same connection also covers seven ad networks and a unified messaging inbox. The MCP server, introduced on the company blog on May 29, 2026, exposes more than 280 tools and works with Claude Desktop, Claude Code, Cursor and any client compatible with the protocol.
For a one-person agency, that changes the math. The part that used to eat a whole afternoon, opening each network, pasting, adjusting, setting a time, becomes a tool call at the end of a flow that already existed. If you want to see the dashboard before deciding, create your Zernio account: the first two connected accounts are free, no card required. The Zernio links in this article are referral links: if you subscribe through them, Wama earns a commission at no extra cost to you.
How does post automation with Zernio + Claude work?
Post automation with Zernio and Claude works as a chain: a master piece of content becomes a campaign, Claude adapts the copy for each network with its own format and voice rules, a mechanical quality check verifies limits and links, a person approves, and only then does the platform receive the posts as a draft, a scheduled post or an immediate publication. The order matters more than the tool.
At Wama, the chain has six steps, and each one has an owner:
Master content. An idea, a behind-the-scenes story or a voice transcript becomes a source text with a thesis, arguments and proof. It is the only file I write from scratch.
Adaptation per network. Each network has its own skill in Claude: a carousel script for Instagram, running text for LinkedIn, a thread of up to 500 characters for Threads, a dry 280-character post for X. The voice, the prohibitions and the limits are written down, not improvised.
Quality control. A script checks length, media, links and sections; Claude itself reads for consistency with the source. The report flags issues, I decide.
Approval. A human checkpoint, per network or overall. Approved content is still not an order to publish.
Publishing plan. A tool-agnostic file states the network, mode and time of each item. Switching platforms in the future means changing a mapping, not the process.
Sending through Zernio. The publishing skill connects through MCP, uploads the media, creates each post in the requested mode and writes the ids back into the campaign. An error in one item does not cancel the others.
The safety rule that governs everything: nothing is sent without an explicit order in the conversation. A finished plan is not an order, and approved content is not an order to publish. When in doubt, the safest mode wins: draft before scheduling, scheduling before immediate publication. Scheduling requires a time I state; publishing now requires a literal request.
What changed in Wama's content process
Before August 2026, the process was me. I went six months without posting, and demand for new projects dried up, something I told people on social media myself when I came back. The problem was never a lack of topics. It was the cost of turning a topic into seven versions and publishing each one in the right place, at the right time, without breaking the character limit.
Between August 25 and September 9, 2026, the pipeline sent 33 posts through Zernio, on six networks (LinkedIn, Threads, X, Instagram, Reddit and Pinterest), from five campaigns. Thirty-two were published. One failed on X because the copy went over 280 characters, and the reason came back readable in the API response, with the limit and the count. There are nine connected accounts, including the Wama page, my profile and my personal networks.
What automation did not change is who decides. Each campaign stops at a checkpoint before becoming a post, and the copy for each network comes from a skill with its voice, limits and prohibitions on record. It is the same standard I apply to Claude Code in development: the tool handles the repetitive work, the person answers for the decision. What really changed was consistency, which is the one thing a service brand cannot buy later.
How much does Zernio cost?
Zernio charges per connected account, in tiers: the first two are free, the third through the tenth cost US$ 6 per month each, the 11th through the 100th cost US$ 3, and above that US$ 1. Unlimited scheduling, publishing, analytics, inbox, webhooks and the full API are included in every tier. Ten accounts add up to US$ 48 per month, according to the official pricing page, checked on September 9, 2026.
Connected accounts | Price per account, per month | Cumulative example |
|---|---|---|
1 to 2 | free, no card | 2 accounts: US$ 0 |
3 to 10 | US$ 6 | 10 accounts: US$ 48 |
11 to 100 | US$ 3 | 100 accounts: US$ 318 |
101 or more | US$ 1 | 1,000 accounts: US$ 1,218 |
On Wama's account, with nine networks connected, the September statement showed US$ 12.80 gross through the 9th and US$ 2.20 net, because the initial free usage credit absorbed the first week. Three tweets through the X API cost 1.5 cents each, a charge X passes on per call, as the Zernio documentation explains, and that Zernio itemizes line by line on the statement. That is less than one hour of anyone doing this by hand.
To run your own numbers, the cheapest path is to connect two networks for free and measure for a week: open Zernio, turn on the MCP server in the Claude you already use and send the first post as a draft.
Three routines that run on their own today
Automation without a routine is a demo. The three below run at Wama in September 2026, each with a clear line between what the machine does and what stays with me.
Campaign for every network
A master piece of content becomes a campaign and goes out adapted for each approved network, with media in the right format. The most recent example is the LinkedIn post with six 6K renders of the Stanley cup, made from the AI ad video that became an article: the images were uploaded through the Zernio API and the post was published directly at 6:53 p.m. on September 7, with the id saved in the campaign. What stayed with me: approving the copy, choosing the photos and deciding the sending mode.
Daily blog as drafts
Every day at 7:05 a.m., a routine researches Wama's 30 commercial keywords, compares them with the entire archive (Framer CMS, local files and sitemap), picks the day's topic by a recorded score, writes the article with the same skill that produces this text, generates the cover and saves the item in the blog collection as a draft. It never publishes. There have been three articles since September 7: how much a landing page costs, professional website creation and corporate website creation. Publishing on the website is still manual.
Blog repost on LinkedIn
Monday, Wednesday and Thursday at 9 a.m., the next article in the queue becomes a post on Wama's LinkedIn page, written by the copy skill, humanized and published directly through Zernio, with the id recorded in the queue. There are 26 articles in the queue; the first went out on September 9, 2026. It is the only series that publishes without per-post approval, because approval was given for the whole series, and it is the exception that proves the rule.
The pattern across all three: the routine produces, the tool receives it in a safe state, and the decision to go live stays with a person, except where I chose to delegate by series.
Mistakes automation does not fix
Each item below happened, or almost happened, here. A good tool returns a readable error; it does not return judgment.
Copy too long for the network. A post for X went out with 1,459 characters and failed. Zernio replied with the 280 limit and a note that each link counts as 23. The mechanical length check exists to catch this before sending.
The same copy on every network. Pasting the LinkedIn text into Threads is the fastest way to look like a bot. Each network has its own skill, voice and limit.
Publishing without approval. The default is a draft. Direct publishing only exists where I authorized an entire series, and it is written down as an exception.
A pile of hashtags. On Instagram, the house limit is five per caption. Automation that dumps fifteen only automates the mistake.
Scheduling without a set time. A time invented by the tool is the wrong time. Scheduling requires a time stated by whoever decides.
Measuring likes. The X skills optimize for replies, conversation and saves, because that is what carries weight. Likes are the easiest number to inflate and the one that brings the fewest clients.
What does this change for people who hire Wama?
For clients, post automation with Zernio and Claude changes consistency, not the promise. Wama publishes every day because the cost of publishing dropped, and consistency is what sustains a brand over time. A corporate website and a social media presence are branding, value built slowly, with no overnight return. Anyone who promises otherwise is selling something else.
The same process always points to the same place: the website. The article is born on the blog, becomes a post, and the post takes the reader to the right page. That is why the website remains the center of the work, and it is what Wama delivers as a Framer agency, with pricing laid out in how much a professional website costs. More examples of AI in the process are in agencies already using AI to build websites.
Frequently asked questions
Does Zernio work with Claude?
It does. Zernio provides a hosted MCP server at mcp.zernio.com, which Claude Desktop and Claude Code connect to with an API key. After that, the agent lists accounts, creates posts, schedules and reads analytics through tool calls. At Wama, that is how the publishing skill sends each campaign, without opening a dashboard.
Do I need to know how to code to automate posts with Zernio?
Not to get started. The MCP server is set up once, with the account key, and from then on requests are in natural language: schedule a LinkedIn post for tomorrow at 9 a.m., for example. Coding helps when you want fixed routines, like reposting the blog three times a week, which runs as a scheduled task.
Does Zernio have a free plan?
Yes. The first two connected accounts are free and do not require a card, with scheduling, publishing and analytics included. From the third account on, billing is per account per month, in decreasing tiers. The exception is X, which charges per API call and therefore requires payment from the first post. See the details on Zernio.
Does the AI publish on its own or does someone approve?
It depends on the routine, and the default rule is to approve. A campaign only reaches Zernio after a human checkpoint, and the safe mode is a draft. The only series that publishes directly is the blog repost on Wama's LinkedIn page, approved as a series. Immediate publishing requires a literal request; a scheduling time is never invented.
Which networks does Zernio publish to?
According to the official page, 16 platforms: Instagram, TikTok, YouTube, X, LinkedIn, Facebook, Threads, Pinterest, Reddit, Bluesky, WhatsApp, Telegram, Discord, Snapchat, Google Business and Slack, plus seven ad networks. At Wama, nine accounts are connected, and six networks have already received posts through the pipeline: LinkedIn, Threads, X, Instagram, Reddit and Pinterest.
Good automation removes the mechanical work and leaves the decision on the table. At Wama, that is how content goes out every day, and that is how I design and develop websites end to end, from design to launch. If your company needs a website that can sustain that consistency, see Wama's work.
