Synthetic Labs
Synthetic Labs is an A/B testing platform for business decisions you can't actually test in the real world, like pricing changes, plan launches, or anything irreversible.
Core trick
Gemini generates a diverse set of persona agents from one sentence, who then debate and vote on the proposed decision.
Description
Synthetic Labs is an A/B testing platform for business decisions you can't actually test in the real world, like pricing changes, plan launches, or anything irreversible. From a single plain-English sentence, it uses Gemini 3.5 Flash to generate a diverse population of agent archetypes who each have their own background and reasoning, then lets them argue, persuade each other, and decide, so you can watch how a real customer base would react before you ship. It solves the core problem that real A/B tests cost months, can't evaluate one-time decisions, and surface failure modes too late, whereas Synthetic Labs returns a defensible recommendation in minutes. In our demo it not only recommended a student plan for Anthropic but flagged a competitor-style free trial as do-not-ship, catching agents coordinating to redeem and cancel before renewal, a failure mode no spreadsheet would surface.
gyanbhambhani/gd-hackathon-syntheticlabs★ 00 forksTypeScriptCreated May 23, 2026 · last push May 24, 2026
README excerpt
# Synthetic Labs **Synthetic Labs is an A/B testing platform for business decisions you can't actually test in the real world** — pricing changes, plan launches, policy shifts, anything irreversible. From a single plain-English sentence, it uses Gemini 2.5 Flash to generate a diverse population of agent archetypes who each have their own background and reasoning, then lets them argue, persuade each other, and decide, so you can watch how a real customer base would react before you ship. It solves the core problem that real A/B tests cost months, can't evaluate one-time decisions, and surface failure modes too late, whereas Synthetic Labs returns a defensible recommendation in minutes. In our demo it not only recommended a student plan for Anthropic but flagged a competitor-style free trial as **do-not-ship**, catching agents coordinating to redeem and cancel before renewal — a failure mode no spreadsheet would surface. --- ## What you get back Every run produces: - An **executive recommendation** — `ship`, `hold`, or `kill` — with the variant pick, a defended rationale, accepted trade-offs, accepted risks, a `do_not_ship` flag for actively harmful alternatives, and a killer quote from the population. - A **headline delta** vs baseline phrased in the recipe's own outcome vocabulary (e.g. "+8 retained_premium vs status quo"). - **Named movers** — the specific synthetic agents whose fate changed between the baseline and the variant, with their backstories and axis values,