Skip to Main Content

Elsewhere by Roszhan

Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

Elsewhere by Roszhan — Demo video

Elsewhere: see the life you didn't choose. Every big choice leaves a life behind: the startup or the stable job, nursing school or keeping the paycheck, a construction job in the Gulf or staying near family in Kerala. Most people decide on gut feel and a pros-and-cons list, and only a lucky few have an advisor to think it through with. Chatbots will tell a confident story, but they blur facts, your own assumptions and guesswork into one smooth paragraph. Elsewhere turns one decision into two simulated futures you can talk to, stress-test and fact-check. You say or type the choice. An intake agent asks at most 3 questions, and only the ones that would change the result. Then Muse Wonder orchestrates 10 Muse Spark specialist agents in parallel (career, money, lifestyle, risk and values for each path) and builds both futures from today to 5 years out. What you can do with them: - Know what to trust. Every value is tagged FACT (a web search ran and returned a source), ASSUMPTION (you said it) or INFERENCE (the model reasoned it). - Talk to both future selves, one at a time or in debate mode. A branch guard stops either one from knowing the other's world. - Throw a curveball like "startup fails at month 14". Only what's downstream recomputes, and it lands as two newspaper front pages with a causal chain for every change. - See what decides it. A chart ranks which assumption swings the result most, and dragging a slider updates the money math instantly. - Find the third door. Wonder checks whether A-or-B is a false choice and proposes a hybrid path you can simulate in one click. - Roll 1,000 random lives (layoffs, medical bills, raises) in about a second, with no model call. - Bring people in with an ask-friends voting link and a printable counselor handout. It's built for people who rarely get an advisor. The demos include a worker weighing a Gulf job against staying in Kerala, a nursing assistant weighing nursing-school loans, and a single parent choosing between rideshare and warehouse work. The design rule: models reason, and code owns ids, provenance, money math and causality. That's what keeps 10 agents consistent and every number traceable. Live demo (demo mode, no key needed): https://elsewhere-futures.vercel.app

Team