Eric Siu walks through a growth loop for short-form video that uses Overlap to auto-clip moments from long-form content, then tests variations with Trial Reels to find what works. The system targets four short-form candidates daily, hands off to team members as "agent managers," and compounds reach by re-publishing high-performing clips.
This one seems pretty good for marketing. Overlap is a great tool I use, great at finding clips, and it has an API, it has an MCP as well, and it'll also caption the clips, it'll find the right moments for mid-form and long-form and short-form as well, and that's good. And by the way, if you imagine you just publish a webinar to it, it'll do a good job of finding the right moments ultimately, and you can just continue to tweak it, right? And that's good. The GrokBot itself, like running these trial reels, it's kind of just doing it itself, and honestly, I probably should have just told it to use the Overlap API to run this. But I'm actually curious to see what it comes up with. So Short Form Scaler, the way this one works, at least I want to have 4 short forms published per day, or at least 4 candidates, and then I can approve. And that way, like, I'm not just letting it kind of do its thing. Again, the cool thing is once you start getting it going, you can hand it off to someone on your team and everyone on your team should become an agent manager. You might get it working, but you can't manage these things. You need to hand these off to other people and that's how it works. And so the Short Form Scaler will kind of get more short forms out. The Trial Reels one is important because it will test out different overlays and it will test out different editing as well. And then it's just going to keep testing. I don't know, let's say I want to Test 4 a day. Trial reels are great because you're testing against people that don't follow you. And if it performs well on those, if you made like a little tweak and then one does like 10,000 views in trial reels, the other ones all do like 100 views, then you know from the 10,000 one, you can go ahead and make it a little tighter, publish a net new one, and it'll get a lot more views for you. And that's how you can start to really compound your views.
