ZyprHub
A shared thread for people and AI

Everyone prompts into one thread. The merge is the only meeting you need.

ZyprHub lets a small group write with AI together — forking when they disagree, merging automatically when they don't, and asking a human only when it's a real call, not a formatting choice.

How it works

Three moves, repeated as often as the work needs.

01

Prompt together

Everyone involved writes into the same live thread instead of separate chats — the group's intent builds in one place, visibly.

02

Fork when you diverge

Two people pulling the thread in different directions isn't a conflict to hide — it's a fork, kept side by side until it's resolved.

03

Merge — automatic, or approved

No real overlap, it merges quietly. A genuine disagreement, it proposes a reconciliation and waits for a human to say yes.

The platform

Every seat can be a person or an agent. Every merge follows your rules.

Agnostic

Bring your own AI

Connect whichever model you already use — Claude, GPT, Gemini, anything with an API. ZyprHub doesn't replace it, it's the room everyone on your team prompts into, no matter whose AI they showed up with.

Participants

Humans or agents, same seat

Any participant in a thread can be a person or an AI agent of your choosing. Three teammates and two agents can prompt live into the same thread, side by side, no different from an all-human room.

Tiered

Merge protocols, by tier

Assign whichever AI you trust to do the reconciling, and choose how it merges — from a light-touch auto-merge to a fully arbitrated review. Which protocols you can pick from follows your plan's tier.

RBAC

Role-based access

Every participant has a role. Roles decide who can prompt, who can see what, and whose input carries more weight when a merge is a real judgment call — a lead's fork can outrank an intern's without anyone refereeing it by hand.

Lineage

Full prompt lineage

Every prompt, fork and merge is kept as an ordered, attributed record — a visible log of exactly how the document came to say what it says, for as long as your team needs it.

Real-time

The live document

Watch the target document update as merges land, not just review it at the end. The thread and the document stay in the same room the whole time.

Why this, why now

The merge is where it matters. Everything else is table stakes.

It reads intent, not text

Two people can phrase the same idea completely differently. Recognizing that they agree — instead of flagging a false conflict — takes understanding meaning, not diffing characters.

It never overrides a real choice

When two directions are genuinely incompatible, the right move is to say so plainly and hand the decision back — never to quietly average two things nobody actually asked for.

No new tool to learn

Bring whichever AI you already use. ZyprHub doesn't replace it — it's the room several of you prompt into together.

Nothing to set up

No accounts, no install. A shared link, a prompt each, one merged result — small enough to try before anyone has to be convinced.

Want to try it with your team?

ZyprHub is early — a small, working demo, not a finished product. If the idea of prompting together instead of separately is useful to you, I'd like to hear about it.