Consensus
“Raft consensus for group chat dinner plans. Every ‘where should we eat’ triggers a leader election. Followers replicate the log. Sarah walks into a tunnel — heartbeat lost — term++— and you end up at a Thai place nobody voted for.”
The Problem
Six people. One group chat. One question: “where should we eat.” This is not a coordination problem. It is a distributed systems problem, and your friend group is running zero consensus protocols. Nobody is the leader. Everybody is a candidate. Consensus is never achieved. You eat cereal standing up at 9:47pm.
The Solution
We embed a Raft implementation directly into your group chat. Exactly one person is elected leader per outing. Their message becomes the source of truth. All follower messages are appended to the log but do not commituntil the leader acks. When the leader misses a heartbeat — usually by opening TikTok mid-plan — we trigger a graceful re-election instead of the current process, which is passive-aggressive silence.
Beta Metrics
Six months of production data across 1,412 friend groups. Bars are the share of plans that actually happened.
Pricing
TAM
The Ask
$4M seed. Half goes into porting Raft to TypeScript (there are already six npm packages that do this, don’t worry about it). The other half funds the Byzantine GeneralSKU, which requires a legal review that our lawyer keeps rescheduling because he’s waiting on his own friend group to pick brunch.
Series A pitch:“We’re etcd, but the cluster is a group chat and the leader has a therapist.”