There's a moment every founder building alone eventually has — where something you built lands in front of a stranger, and you realize you can't explain your way around a bug anymore. It either works, or it doesn't.
Mine happened on a Tuesday afternoon, watching Railway logs scroll past in real time while someone I'd never met clicked "Send for Approval" on a test post inside eWork Social for the first time.
It failed. The table 'public.PostApproval' does not exist in the current database.
I'd built the whole workflow — the Client Approval Portal, the magic-link email, the no-login review page — and shipped it everywhere except the one place that actually mattered: production. Migration ran locally. Never touched Neon.
Fifteen minutes and one SQL script later, it worked. Client got the email. Clicked the link. Reviewed the post. Approved it. No account created, no password, no confusion — exactly as designed, just fifteen minutes later than promised.
That's most of what building alone actually looks like. Not the launch. The fifteen minutes after the launch where you find out what you missed.
The Rejection That Wasn't Really a Rejection
Meta's App Review team sent back a decision on instagram_manage_comments — rejected. My first read was disappointment. My second read, once I actually opened the feedback, was relief: everything else got approved. Nine permissions through. One held back, and only because the screencast didn't show the complete login-to-reply flow clearly enough for a reviewer who'd never seen the product before.
It's a strange kind of lesson — the thing you build and the thing you demonstrate are not the same task, and the second one deserves just as much care as the first. Rerecorded it properly this week. Resubmitted. Waiting again, but waiting differently now — with the confidence that comes from having done it right the second time.
Where the Comments Actually Go Now
While that review sat pending, I built the thing agencies had been asking for without quite naming it: one inbox for every Instagram and Facebook conversation, instead of six tabs and a prayer. The Engagement Hub shipped this week — DMs and comments landing in a single feed, an AI-drafted reply waiting for a human to glance at before anything gets sent.
I was careful about that last part. Not because the AI can't write a good reply — it usually can — but because the moment a brand's voice starts sounding like it was generated by nobody in particular is the moment people stop trusting the brand at all. So the model drafts. A person still decides. That's not a limitation. That's the whole point.
The Page I Didn't Touch
The hardest thing I did this week was nothing. A pile of genuinely good feedback arrived for the Founding Growth Partner page — sharper hero copy, a founder story section, a smarter commission structure that doesn't quietly bankrupt itself at scale. All of it worth doing.
None of it got touched. Because while that feedback sat in my notes, someone real was in the middle of signing up through that exact page. You don't renovate a room while someone's still standing in it.
The improvements will happen. Just not this week, and not while it's working.
The Signup
Somewhere in the middle of all of this — the failed migration, the resubmitted screencast, the inbox that finally unified — a conversation I'd had about the Growth Partner Network stopped being a conversation. Someone signed up.
It's one person. I know exactly how small that number is next to what this needs to become. But it's the first time the words "Founding Growth Partner" pointed at an actual name instead of a hypothetical one, and that distinction is bigger than the number suggests.
Next
There's more waiting in the queue with Meta, more quiet leads worth one more honest follow-up before I let them go, and more of these logs — because I'm starting to think the real story of eWork Social was never going to be the launch. It's going to be every ordinary week between the launch and the first person who stays.
If you're building something similar — an agency, a tool, a community — and you want to see how the Client Approval Portal or the Engagement Hub actually work before you commit to anything, the trial is still free, and still takes about two minutes to start.
— Bernard Oshapi, Founder, eWork Social