with zarie, we wanted to build jarvis for everyone. an omnipresent assistant that quietly kept your life together. when it didn't take off and openclaw went viral a month later doing "ai assistant" in a very different way, i couldn't help but ask: what did we miss?
zarie was closed source and opinionated. we built the whole thing and exposed it through telegram. you used it, gave feedback, and we shipped improvements. but it wasn't a hair‑on‑fire problem for most people, and the ones for whom it was had an obvious alternative: hire a real assistant.
openclaw was open source. you had to set it up yourself, tinker to add features. but that was the magic. it sat in a goldilocks zone, hard enough to feel like a real project, easy enough that you didn't need to build an assistant from scratch. and because people were building on it, they contributed back. the project got better with every user, something zarie's closed model never allowed for.
the big difference wasn't "open source vs closed source" or "devs vs consumers". it was about ownership and identity. openclaw made people feel like iron man, wiring up their own suit. zarie asked you to be a user of someone else's assistant. it's the ikea effect applied to software: people value what they help build.