OneStone
Trust, privacy, and legibility

Useful insight should show where it came from.

OneStone keeps community choices, behavioral signals, AI assistance, moderator judgment, and audience context legible instead of blending them into one mysterious score.

Signals keep their names

Likes show appreciation. One movable endorsement identifies the argument a participant believes most deserves to travel forward. Replies, cross-position support, and documented view shifts remain observable community actions; AI assessments are labeled separately.

Aggregate by default

Education, professional experience, age bands, location, and demographic tags are opt-in. Demographic and location summaries require 10 eligible respondents for topics with one or two positions. For topics with three or more positions, the minimum is five respondents per position.

AI assistance with receipts

AI flags include a category, confidence, explanation, and source links when available. Users can appeal. Human moderators can override. A flag is context—not a verdict.

No political microtargeting

Individual demographic records are not sold or exposed to candidates. Any insights product should use explicit consent, minimum cohort sizes, audit logs, and contractual limits on downstream use.

Public reasoning can travel without private content leaking

Private and institutional repeats may include a balanced summary from a linked public topic. The bridge broadens the argument set while keeping closed-room membership and content confidential.