Building automations manually takes way too long

I completely agree. Building automations manually can quickly become a bottleneck, especially as workflows grow more complex. Looking forward to seeing how your approach evolves and saves time for builders. Thanks for sharing!

The core insight is genuinely sharp: "you forget the tiny steps when you describe a task, so just do it and let the system watch." That's a real observation about why automation-building fails, and "screen record yourself, get an n8n flow" is a clean, concrete promise. The demo-by-doing approach sidesteps the exact thing that makes Zapier/Make setup miserable. Good wedge.

The honest challenge, since this is a launch and the AI Tools crowd will probe it: the hard part of your product is the part the pitch skips over, and buyers will sense it. Watching clicks and turning them into a working n8n flow is a genuinely difficult inference problem. A screen recording shows what you clicked, but not why, not the conditional logic (if this field is empty, skip), not the error handling, not the intent behind the action. The gap between "records my clicks" and "produces a flow that actually works on the messy real cases" is where tools like this usually break, and it's the thing a skeptical buyer will test first. So the trust question isn't "does it record," it's "how often does the generated flow actually run without me fixing it."

Which points at the positioning decision: if the flows need heavy manual cleanup, you're a starting-point generator (still useful, but a different, smaller promise). If they run clean out of the box, that's the whole magic and it should be the entire headline with proof. Those are different products with different buyers, and which one you actually are determines everything about how you sell it. Lead with a real before/after: "here's a 5-minute recording, here's the flow it produced, here's it running untouched." Show the untouched part, because that's the claim people won't believe until they see it.

What's your actual hit rate, what share of generated flows run without editing? That number is your product, and it decides whether you're selling magic or a head start.

This resonates with me. I had the exact same problem — every time I tried to automate a task, I'd forget to document the small steps and the automation would break on edge cases. That's actually what motivated me to build a collection of 30 Python automation scripts for common tasks like file management, web scraping, and data processing. Instead of trying to capture every step of a complex workflow, I found it more practical to have ready-to-use building blocks that you can chain together. Your screen recording approach is interesting though — have you considered letting users annotate the recording with decision points or conditional logic? That could bridge the gap between recording and replay.

Congrats on the launch. I spent a few minutes looking through the product and I like that you're solving a problem almost every growing team eventually runs into—manual automation setup becomes a bottleneck long before people realize it.

One thing that stood out is that the homepage explains what the product does quite well, but the first screen could do a bit more to communicate who it's for. A visitor should immediately recognize themselves without having to read further.

I also noticed a few places where the messaging could connect features to business outcomes more directly. People don't just want faster automation—they want fewer repetitive tasks, fewer mistakes, and more time for higher-value work. Making those outcomes more explicit could strengthen the first impression.

I found a few other messaging opportunities as well. If it's useful, I do a detailed homepage audit for $10 where I document every conversion friction point and include practical rewrite suggestions.

The interesting part is capturing the hidden steps people forget when designing automations.

Curious how you handle edge cases after recording — like decisions, exceptions, or steps that depend on changing data?