high
bar
low
bar

Fast and simple group polling with realtime reporting

Create a poll with your own high and low labels. Share the link. Watch your group's responses come in.

Create a Poll No account required.

How it works

1

Create

Name your poll. Set the high label and the low label. Choose who can see results.

2

Share

Send the voter link to your team, your clients, or your group. No account needed to respond.

3

See results

Watch responses flow in. The overlap zone reveals the consensus. Export to CSV any time.

Why a range

A yes/no forces an artificial choice and throws away the most useful signal: how sure anyone actually was. Even a Likert or 1 to 5 scale still pins each person to a single point; a range lets one voter mark a span, and the gap between their low and high is their uncertainty. A high/low range keeps the nuance, so a tight result and a divided room never look the same.

Capture confidence, not just a choice

Every voter drags two handles to set an upper and lower bound. The width of that band is the data a binary poll throws away: certainty, hesitation, and risk tolerance in one gesture.

See the alignment gap

A clean average can hide a room where half voted high and half voted low. The results page stacks every band and plots the distribution, so convergence and a two-camp split read at a glance.

Three links, total control

No accounts, no dashboards to learn, no multi-page setup. Creating a poll hands you exactly three links: one to vote, one to share results, one to manage. That is the whole product.

For agentic systems

Built for the era of AI swarms

The autonomous Delphi method for LLMs.

An orchestrator spins up a poll, fans the voter link out to a swarm of sub-agents, and each one reports an independent range instead of a flat yes/no. Poll the consensus endpoint, read the variance, and run another round until the agents converge. Standardized over MCP, so any compatible agent can discover and drive it.

01 Spin up a poll with the MCP create_poll tool
02 Fan the voter link out to independent sub-agents
03 Read the consensus endpoint: mean range, variance, convergence
04 Below threshold? Refine the prompt and run another round

Same primitive, whether the voters are people or models. The MCP server and JSON consensus API expose create, vote, and read-consensus as first-class tools.