About Jybe
Jybe is a structured discussion platform built to make hard conversations easier. We help people separate what is true (facts), what matters(values), and what we should do (policies) so communities can see where they agree, where they disagree, and why.
Our mission
Jybe exists to foster healthier online dialogue by combining AI-powered fact checking with transparent structure. The goal: surface shared ground, highlight real points of contention, and guide people toward actionable next steps instead of endless back-and-forth.
What makes Jybe different
- Structured claims: Every post and reply can be broken down into Facts, Values, Policies, and Unstructured notes so meaning stays clear.
- Built-in fact checks: AI-assisted verdicts (True, False, Mixed, or Uncertain) come with rationale and citations, plus suggested rewrites when a claim needs work.
- Consensus first: Live consensus bars and dashboards summarize support, opposition, and agreement patterns across participants so you can see the room at a glance.
- Transparent attribution: Badges on posts, replies, and individual claims show who contributed what while keeping telemetry private to the backend.
- Safety and clarity: Guardrails encourage engaging with verified facts, warn when responding to uncertain claims, and keep accessibility and readability front and center.
How to use Jybe
1) Browse the feed
From the top navigation, choose Feed (or click the Jybe logo) to see the latest structured discussions. Scroll through post previews that show excerpts plus quick Fact / Value / Policy counts. Pagination links let you jump to specific slices you can share.
2) Filter and search
Use the filters above the feed to zero in on what matters:
- Category — pick a topic like climate or transit to focus the list.
- Author — start typing a handle to see suggestions and filter to a voice.
- Search — type keywords; results update automatically as you type.
Active filters appear as chips you can remove anytime to reset the feed.
3) Read a structured post
Open any post to see the original text plus its structured breakdown. Tabs and sections separate factual claims, value statements, and policy proposals so you can review each on its own terms.
4) Compose your own post
- Click Compose in the header and draft your statement.
- Live extraction suggests candidate statements grouped into Fact, Value, Policy, and Uncategorized buckets. You can move items between buckets if the guess is off.
- For Fact suggestions, hit Check claim to run a fact check. Verdicts include confidence, rationale, and suggested rewrites for False or Uncertain claims.
- Accept supporting facts to attach them to your post; publish when you're ready so the structured content appears in the feed and consensus views.
5) Check a single factual claim
On the featured demo post, the Ad-hoc factual claim panel lets you paste a single sentence and run a quick fact check. You'll see a verdict and suggested rewrites you can accept as supporting facts.
6) Tag with categories
When categories are enabled, the composer offers a Categories panel where you can search for existing labels or create new ones. Add up to five to help others discover and filter your post; remove chips to clear them on the next save.
7) Reply with structure
- Click Write a reply at the bottom of a post.
- Replies use the same Fact / Value / Policy buckets and live extraction to keep the conversation coherent.
- Fact claims can be checked and accepted as supporting facts. Some demo configs require you to Jybe (agree), abstain, or counter-claim every true fact before submitting.
- Accepted supporting facts appear as chips and show up in the “Recent Replies” list.
8) See consensus and attribution
Consensus bars and dashboards summarize how people respond to each fact, value, and policy. Hover or click author badges (when the popover is enabled) to compare agreement patterns. Attribution badges keep contributions clear without exposing private telemetry.
9) Stay unstuck
If something looks off—like an empty feed or missing featured post—try refreshing the page or signing out and back in. Operators can also reseed demo data or toggle advanced features like inline annotations and live AI pipelines as needed.

