Most SaaS marketing strategy advice sounds the same: pick a niche, write comparison pages, run some ads, pray. Then there is Nevo David, the founder of Postiz, who just published a step-by-step X article explaining how a social media scheduling tool reached $2.2M ARR, mostly by being loud on social media and using AI agents to do the heavy lifting.
The article went out on September 7, 2026, pulled more than 100K views in a day, and reads less like a marketing playbook and more like someone opening their notebook. It covers a mindset (“sell an outcome, not an alternative”), and then three concrete tactics: hacking X for 200K to 7M views, running UGC pages that hit millions of views, and TikTok slideshows that go viral without a product in sight. A lot of it runs through ChatGPT Astra and an agentic scheduler.
This post walks through the whole system, keeps the storyline intact, and adds the technical layer the original skipped: how to actually wire ChatGPT, Claude, or any other agent into Postiz so the tactics run on autopilot. You can read the original X article here.
Manage all your social media in one place with Postiz
The numbers behind the story: $186K MRR, $2.2M ARR
Nevo opens with a public Stripe profile rather than a screenshot, which is the kind of receipt you rarely get in SaaS marketing content. The chart tells the story on its own: two years of flat revenue, then a near-vertical climb starting in early 2026.
He is careful to add “might be a little lucky there.” Fair. But the rest of the article makes the case that the luck was engineered: a repeatable system for getting millions of views across channels, with a large share of the publishing handled by Postiz itself. Dogfooding, in other words.
The mantra: four beliefs that shape the whole SaaS marketing strategy
Before tactics, Nevo lays out what he actually believes about selling software. Everything that follows is downstream of these four lines:
A person needs to see you many times before buying from you.
Exponential growth comes from social media and going viral as often as possible.
Being viral is a matter of money and growth hacking, not luck.
Everyone can make a lot of money, but not by doing what everybody else is doing.
Notice what is missing: SEO, paid acquisition, partnerships, outbound. Not because they do not work, but because the strategy is built on repeated exposure at scale, and social is where that is cheapest.
Sell an outcome, not an alternative
This is the part of the article worth printing out. Once you have picked your niche, Nevo argues, there are only two ways to sell: you sell a tool alternative, or you sell a new opportunity.
Postiz as a tool is “an agentic social media scheduler that connects any agent and posts to 30+ channels.” That is a real product, but it has many competitors. Selling it that way turns every conversation into a feature fight, or a build-versus-buy debate, because the buyers are technical. Worse, those buyers already know what they want. They ask ChatGPT for “an AI social media scheduler,” and Postiz shows up as one option among many. That is selling the alternative, and Nevo says you will always hit a ceiling there.
To break past $100K MRR, he says, you have to sell an outcome. Outcomes look like:
How to “hack” X content so you get 200K to 7M views.
How to run UGC pages that get millions of views.
How to get millions of views with TikTok slideshows.
His example: a founder with a B2C mobile app has no idea how to promote it. That founder was never searching for a scheduler. But they stumble into a Postiz article about getting millions of views on TikTok, and now the scheduler is the thing that makes the money. “I created the demand,” Nevo writes. “That’s how you remove yourself from the competition and create a new category for yourself.”
Every tactic below is paired with that outcome framing. The tool never leads; the result does.
The setup: connect an agent to the scheduler first
Nevo’s suggestion before you start is to grab the 7-day Postiz trial and connect it to whichever agent you like: ChatGPT, Claude, Grok Bot, Hermes, OpenClaw, or Cursor. His own pick is ChatGPT, OpenClaw, or Hermes, because they can run ChatGPT 6 Astra, which he calls “next level.”
The connection itself is a single MCP endpoint. Postiz exposes a Model Context Protocol server at mcp.postiz.com, so any MCP-capable client can list your channels, read each platform’s rules, upload media, and schedule posts. In Claude Code it is one line:
claude mcp add postiz --transport http --url https://mcp.postiz.com/mcp/your-api-key
For ChatGPT, Cursor, or any other client, add a streamable HTTP server with the same URL (or use https://mcp.postiz.com/mcp with an Authorization: Bearer your-api-key header). The API key lives in Postiz Settings. Once connected, ask the agent to “list my connected social media accounts.” If it returns your channels, the wiring works.
Prefer the terminal? The Postiz CLI wraps the same public API and works well inside cron jobs and agent skills:
npm install -g postiz
postiz auth:login # device-flow OAuth, no client secret needed
postiz integrations:list # get the id of every connected channel
Two things trip people up. First, every image or video has to go through postiz upload before it is attached to a post; providers like TikTok and Instagram reject external URLs. Second, every scheduled post needs an ISO 8601 date. With that out of the way, here are the three tactics.
Tactic 1: How to “hack” X content for 200K to 7M views
Nevo calls X “super easy” and says it works especially well with agents. The reason is structural. X is an interest-based network: if many people are talking about a specific topic and you make content about that exact topic, X pushes your post to those people and, if you are lucky, into the trending feed.
Right now the hottest topic on X is agentic AI. He knew ChatGPT Astra was launching, so he made it the headline of the article. And because Postiz works with any LLM, the tactic does not depend on which model is trending. You can prepare content in advance and swap the name.
The weekly cadence
Monday: publish a video on X. It must be genuinely valuable, talk about how to make money, and mention Postiz without selling Postiz. He aligns the video’s outcome with the product’s outcome: Postiz exists to make money, so the video helps you make money. “If I nail the video, you will remember Postiz forever.”
Stay “in the launch.” Reply to every single comment, even if the reply is just a heart or a folded-hands emoji. The goal is to signal to X that the content is good.
Keep relaunching all week. Quote-post the video again and again. Launching once does not mean you cannot keep launching until something goes viral. Clip pieces of the original and quote-post those too.
Next week: turn the video into an article. Run the video through ChatGPT Astra (Whisper for the transcript), feed it your previous articles so it matches your voice, then schedule the article through Postiz.
Obsess over the title and cover. Just like YouTube, the title and cover image are all anyone sees before clicking. Add the keywords you want X to target.
Quote-post the article all week too.
Why articles? Two reasons. The X algorithm rewards content people spend a long time reading, because it keeps them on the platform. And articles later rank on Google, because x.com is a very high-authority domain. A note on voice: Nevo points out that a year or two ago AI would never match your writing and would sprinkle in em dashes everywhere. Now, with a few of your old human-written articles as reference, the output is very hard to distinguish from you.
One more thing to note about direct selling: a purely self-promotional video or motion graphic almost never goes viral with the right audience unless you pay influencers to push it. Nevo says he almost never uses influencers. The value-first video is the workaround.
How to schedule X posts and X articles with an agent
Postiz treats an X article as a normal post with a different post_type. The content becomes HTML (headings, paragraphs, lists, links, bold, underline are preserved), the character limit jumps to 100,000, and you can attach a cover image. Through the CLI, the Monday video looks like this:
VIDEO=$(postiz upload monday-video.mp4 | jq -r '.path')
postiz posts:create \
-c "How I get 200k+ views on X every week (and why agents make it easy)" \
-m "$VIDEO" \
-s "2026-09-14T13:00:00Z" \
--settings '{"who_can_reply_post":"everyone"}' \
-i "$X_INTEGRATION_ID"
And the following week’s article, with the cover uploaded first:
COVER=$(postiz upload article-cover.jpg)
COVER_PATH=$(echo "$COVER" | jq -r '.path')
postiz posts:create \
-c "<h2>Why I turn every video into an article</h2><p>...</p>" \
-s "2026-09-21T13:00:00Z" \
--settings "{\"post_type\":\"article\",\"article_title\":\"How I get 200k views on X\",\"article_status\":\"published\",\"article_cover\":{\"path\":\"$COVER_PATH\"}}" \
-i "$X_INTEGRATION_ID"
If you are working through MCP instead, the agent calls integrationList, then integrationSchema for the X platform to read the rules, then schedulePostTool with the same settings. You do not write any of that; you just tell ChatGPT or Claude “turn last Monday’s video into an X article in my voice and schedule it for next Monday at 9am.”
Tactic 2: UGC videos that get millions of views on Instagram and Facebook
Nevo is upfront that this one is a growth hack with a shelf life. “This works right now; I am not sure for how long. You should definitely assume you can get banned.” The tactic is built for Instagram and Facebook, though with a scheduler you cross-post the same videos to YouTube and TikTok for free.
Here is the kind of video the Postiz team has been running:
The structure is always hook + outcome + how-to + CTA. Almost every day the same video, only the hook changes. What they have seen, especially on Facebook, is that each video gets 200 to 500 views until one suddenly gets 200K+. That usually takes about 30 videos, posted once a day. The variations are cheap to produce, and Nevo mentions using agent-media to generate the UGC creator at the beginning of each video.
Look at those two grids. Dozens of videos in the low hundreds, then 286K, 58K, 997K, and 1.2M. That is the distribution the tactic depends on. You are not trying to make every video good. You are buying lottery tickets with a very high expected value, and the daily consistency is the price of the ticket.
The CTA matters more than it looks. “Comment ‘something’ to get ‘something'” produces a flood of comments, and the comments push the video further. Nevo notes this has become a well-known B2B tactic; he keeps spotting friends from X doing it.
Scaling it: creators versus pages
There are two ways to scale. Option one is to hire a creator to post daily, typically $20 to $45 per video. Option two is to create many pages, generate the opening hook with agent-media, and schedule the video to every page in one go. The second option is where the agentic scheduler earns its keep: one instruction, many channels.
The warning that comes with it is important. Meta is one of the strongest platforms at detecting bots, and “bot” does not mean “automation.” It means account behavior. An account that only posts and never engages, never likes, never comments, never follows, will be banned or shadowbanned. So a human has to do the engagement part, or the whole page eventually dies.
How to cross-post one UGC video to every channel
This is the part that is tedious by hand and trivial with the API. One JSON file, one command, four platforms. Every platform gets its own caption and settings, and the same uploaded video:
VIDEO=$(postiz upload ugc-day-12.mp4 | jq -r '.path')
cat > ugc.json << 'JSON'
{
"type": "schedule",
"date": "2026-09-15T15:00:00Z",
"posts": [
{ "integration": { "id": "INSTAGRAM_ID" },
"value": [{ "content": "Comment SCHEDULE and I'll send you the exact setup", "image": ["VIDEO_URL"] }],
"settings": { "__type": "instagram", "post_type": "post" } },
{ "integration": { "id": "FACEBOOK_ID" },
"value": [{ "content": "Comment SCHEDULE and I'll send you the exact setup", "image": ["VIDEO_URL"] }],
"settings": { "__type": "facebook", "post_type": "post" } },
{ "integration": { "id": "YOUTUBE_ID" },
"value": [{ "content": "How I post to 30 channels from one prompt", "image": ["VIDEO_URL"] }],
"settings": { "__type": "youtube", "title": "Post to 30 channels from one prompt", "type": "public" } },
{ "integration": { "id": "TIKTOK_ID" },
"value": [{ "content": "Comment SCHEDULE for the setup #saas #automation", "image": ["VIDEO_URL"] }],
"settings": { "__type": "tiktok", "privacy_level": "PUBLIC_TO_EVERYONE", "duet": true, "stitch": true,
"comment": true, "brand_content_toggle": false, "brand_organic_toggle": false,
"content_posting_method": "DIRECT_POST" } }
]
}
JSON
sed -i "s#VIDEO_URL#$VIDEO#g" ugc.json
postiz posts:create --json ugc.json
To run this across ten pages, loop over ten integration ids. To run it daily, put the loop in a cron job or an agent skill and let the agent write a fresh hook each morning. The engagement (likes, replies, follows) still has to come from a person.
Tactic 3: TikTok slideshows that get millions of views
“TikTok works better for B2C, but can also work for B2B,” Nevo writes, and then spends most of the section on something nobody talks about: account hygiene. Because TikTok serves content based on your current location, a founder outside the US who wants US reach has to do a few unglamorous things first.
Set up the account like a real person
Use a mobile proxy, not a datacenter VPN. NordVPN-style dedicated IPs and Hetzner boxes are datacenter proxies and easy to flag. He suggests a mobile proxy over WireGuard, and turning off GPS.
Warm up the account. A fresh TikTok account shows you brainrot because it is what goes viral. Search your niche, like, comment, follow. Close the app, reopen it, confirm the feed has shifted to your niche. Repeat until it sticks. This is not only to look human; it also tells TikTok who to show your posts to.
Repeat the warm-up before every post. Skip it and you can land in the zero-views jail. If you do, stop posting for a while. It recovers.
Then research. Search your niche before you post anything, find what already works, and copy it. Viral videos can be turned into slideshows.
The slideshow formula
The slideshows that go viral are never about the product. They are about a person, a conflict, and a reveal. “Nobody cares about your features, they care about the moment.”
His example is an AI room redesign app that posted “My landlord said I can’t change anything so I showed her what AI thinks it could look like” and did 234K views. It works because you picture the landlord’s face before you even swipe. The formula:
[Another person] + [conflict or doubt] → showed them → they changed their mind.
Other formats that work: “X things to get Y” with every slide numbered; a problem many people share on slide 1 and a fix on every following slide; “information only I know” on slide 1 and the secrets on the rest.
Slide 1 is the hook, and it is the only thing 95% of viewers will read, so that is where the time goes. Every hook that works has three ingredients: a specific someone (me, my mum, my landlord, my boyfriend, my boss, never “people” or “you can”), a number or a time if you have one, and a tiny story the viewer finishes in their own head.
Producing the slides is the easy part. ChatGPT Astra can generate both the images and the on-slide text, and the scheduler pushes them to TikTok. Once one account works, the same slideshow goes to multiple TikTok accounts.
How to schedule a TikTok slideshow through the API
A TikTok photo post is just a post with several images attached. Two details are easy to miss. content_posting_method must be DIRECT_POST to actually publish (the UPLOAD value drops the media into the app inbox and expires after 24 hours). And autoAddMusic only applies to photo posts, so this is the one place it belongs:
Multiple accounts is the same command with a comma-separated list of integration ids. And if you would rather review each slideshow inside the TikTok app before it goes live, switch to the UPLOAD method deliberately; it is a valid choice for new accounts, just not the default.
What ties the three tactics together
Read the three sections side by side and the same SaaS marketing strategy shows up every time:
Volume over polish. Thirty UGC videos for one hit. Quote-posts all week. Slideshows on multiple accounts. The outlier pays for everything else.
Ride the current topic. Agentic AI on X today, something else next quarter. A scheduler that works with any LLM lets you swap the name without rebuilding the pipeline.
Outcome first, product second. The video is about making money, the slideshow is about the landlord, the article is about views. The tool sits in the background.
Automate publishing, never engagement. Agents write, schedule, cross-post. Humans reply, like, follow, and warm up accounts. Every ban story in the article comes from getting that split wrong.
Nevo closes the original with “let me know if you are going to implement anything here.” Same question applies. If you are going to run one of these, start with the X cadence: it needs zero new accounts, and the video-to-article step is where ChatGPT Astra does the most work for the least effort.
Try the system with Postiz
Everything in this playbook runs on one scheduler connected to one agent. Postiz gives you a 7-day free trial, connects to ChatGPT, Claude, Grok, Hermes, OpenClaw, Cursor, or anything else that speaks MCP, and publishes to 30+ channels including X (posts and long-form articles), Instagram, Facebook, TikTok, YouTube, LinkedIn, and Threads.
Start your free Postiz trial, connect your agent with a single MCP URL, and schedule your Monday video before the week starts. Full setup docs for the MCP server, the CLI, and the public API are on docs.postiz.com.
Discover open source marketing automation. Learn what it is, its benefits over proprietary tools, and how to implement it for full data control and flexibility.
