Hey Product Hunt 👋 When we started building 64ads (an AI-driven platform for generating marketing assets at scale), we genuinely thought publishing would be the trivial part. Like, "we’ll just hit the API at the end and call it done." Turns out we massively underestimated what "just integrate publishing" actually means: - Each platform has its own app review process (some take weeks) - Each has different verification requirements before you can even post - Twitter’s API works completely differently from LinkedIn’s, which is nothing like Instagram’s - Rate limits aren’t just "requests per hour" – they’re nested, conditional, and platform-specific - Orchestrating all of this together so that partial failures don’t cascade? Way more tricky than we expected. We ended up spending w…
Hey Product Hunt 👋 When we started building 64ads (an AI-driven platform for generating marketing assets at scale), we genuinely thought publishing would be the trivial part. Like, "we’ll just hit the API at the end and call it done." Turns out we massively underestimated what "just integrate publishing" actually means: - Each platform has its own app review process (some take weeks) - Each has different verification requirements before you can even post - Twitter’s API works completely differently from LinkedIn’s, which is nothing like Instagram’s - Rate limits aren’t just "requests per hour" – they’re nested, conditional, and platform-specific - Orchestrating all of this together so that partial failures don’t cascade? Way more tricky than we expected. We ended up spending weeks on what was supposed to be a side quest. Our product is about AI content generation - that’s where we wanted our focus to be. Instead, we were stuck dealing with app reviews and debugging why LinkedIn would trim the text in the middle. So, we decided to make it a separate product - Postproxy - and today we’re launching with 7 platforms (Twitter, LinkedIn, Facebook, Instagram, YouTube, Threads, TikTok) and three ways to integrate: REST API, n8n community node, or MCP server for AI agents. Would love to hear if anyone else has felt this pain. What broke your automation? Platform oddities? The gap between "API says ok" and "actually published"? Happy to answer any technical questions about how we handle the messy reality of multi-platform publishing Cheers, Dmitry