Scalar retro and post-mortem | Rahul Tarak
A founder's post-mortem on spending 3 years building a "Figma for video editing" reveals why standard startup advice (MVP everything, move fast) destroyed a product that needed deep technical abstractions—and how getting "nerd sniped" by WASM cost them a year of their lives.
Read Original Summary used for search
TLDR
• The fatal mistake: applying generic "MVP everything" advice to a "deep software" product that needed years of technical foundation—MVPing their multiplayer engine cost 1.5 years
• Got nerd-sniped by WASM for local video rendering to optimize unit economics, but it wasn't performant and users didn't care—wasted a year
• Tried to "hack GTM" before product was ready with side projects (Anno.so) and film festivals, which split focus and sold a "bucket with no floor"
• The "infinite canvas" insight: moving away from expected UI (timeline) gave them more user leeway since people didn't have preset expectations
• Root cause: being "wishy washy" about company type—didn't call their shot, so they listened to advice for wrong company types and made poor capital/remote work decisions
In Detail
Rahul Tarak spent three years building Scalar, a professional collaborative video editor (think Figma for video), before shutting it down. The core thesis came from his experience as an amateur filmmaker—editing always required being in-person, and he wanted to enable seamless sync and "soft async" collaboration. The vision was a professional-grade tool you could use 8 hours a day, with a low floor and high ceiling.
The critical mistakes weren't about market or competition—they were about applying the wrong mental model. They got "nerd sniped" by WASM, spending a year trying to render videos locally for better unit economics and free tier experience, but it wasn't performant enough and users didn't care. More fundamentally, they applied standard "MVP everything, move fast" advice to what they now call a "deep software" product. For products like video editors, you can't MVP core technical abstractions—they MVPed their multiplayer engine and it cost them 1.5 years. They also tried to "hack GTM" before the product was ready, launching side projects like Anno.so (a Frame.io clone) and attending film festivals, but this split their focus when they were selling a "bucket with no floor."
The root cause was being "wishy washy" about what type of company they were building. They had conviction about the problem but not their approach to company building, so they listened to advice meant for different company types. This led to poor decisions around capital (too cheap and skittish), remote work (should have been in-person in SF earlier), and product strategy. When they needed a larger seed round, they entered the classic fundraising death spiral where focusing on fundraising killed product momentum. The author's brutal honesty about identity loss ("what was I, if not a founder?") and the specific technical lessons make this valuable for founders building complex technical products.