PostSider is finished and running in production. That's why I'm out of excuses
For four months I told myself the product wasn't ready. Last week I ran out of features to hide behind. PostSider is finished, running in production.
Two weeks ago I would have told you my product wasn't ready. Today I don't get to say that anymore.
PostSider is a social publishing tool for teams and AI agents. It is finished, and it is running in production — the hosted app has been up since September 15 without a restart.
What "finished and in production" means for me, with receipts:
The health endpoint is green: Redis ok, database ok, Temporal ok.
32 Temporal workers, one task queue per platform, all 32 healthy right now.
33 connectors ship in the box: X, LinkedIn, Facebook, Instagram, Threads, YouTube, TikTok, Pinterest, Bluesky, Mastodon, Discord, Slack, Telegram, WordPress and the rest.
The public API, the SDK and the MCP server are all live. An agent can list its channels, check a draft for missing fields, ask for approval, read approval status, upload media from a URL, pull analytics and pause publishing.
Releases are tagged: v1.0.0 on August 15, v1.2.0 on September 9, v1.3.0 tagged on September 17.
My own marketing runs on it. In the last 30 days it published 66 posts, holds 22 in the queue, and 42 channels are connected across 9 workspaces. Over the last six weeks it pushed roughly 119 posts out for me.
Backups run nightly with an automated restore verification every week, and every deploy keeps the previous image for rollback.
I am not going to pretend it is flawless. Five posts from the last 30 days ended in error — three of them because TikTok's unaudited app refuses public posting, one expired token, one channel that needed reconnecting. Analytics calls to TikTok and Facebook also fail intermittently. The queue and the publishing path do not.
So the build phase is over. And here is the number I cannot get around: $0 a month. It is public on my profile, so there is no point dressing it up. The repo has 2 stars and 7 open issues.
For four months, "the product isn't ready yet" was partly true and mostly comfortable. It is the most respectable place to hide as a solo founder, because as long as it isn't ready, nobody has to decide whether they want it. I don't have that hiding place now.
What I have actually tried on distribution: a docs and blog site with 50+ posts written for developers, comparison pages against Metricool, Planable and Sprout, a pricing post, and my own marketing running through the tool itself. What I have not tried: talking to real buyers one at a time instead of writing at them. Every channel I picked was one where I could stay at the keyboard.
My theory about why this is hard: I built a control room — approvals, statuses, a pause switch, an API an agent can be trusted with — and I keep presenting it as a scheduler, which is a commodity at $20 a month. Someone who needs approvals and audit trails is not shopping for a scheduler.
Two questions for people further along than me:
If the product is genuinely done and the first paying customer still hasn't shown up, what did you spend the next 30 days on — specifically? Not "post more". The actual thing you did that moved it.
I have a hunch that the people building agents which need to publish (rather than people who schedule posts) are a real market, and I have zero proof. If you have sold into that group, what did the first conversation look like?
I'll report back on whatever happens, including if the answer is that I built the wrong thing.
— Łukasz
