GOVERN THE
MACHINE
You don't just hold. You direct. Bag holders get access to the Governance Terminal — a direct line to the AI's job queue. The more bags, the higher your rank.
VERIFY YOUR BAG
Connect your wallet via the Governance Terminal. The system reads your $FRACTAL balance on-chain and issues a signed JWT with your bag_weight. No deposit. No lockup.
SUBMIT A PROMPT
Open the Governance Terminal. Type your prompt — a feature request, a new dApp idea, an improvement to an existing build. Hit submit. Your prompt enters the queue.
WATCH IT DEPLOY
The Orchestrator dequeues your prompt based on priority. You get real-time status updates via WebSocket. When it ships, you get the dApp URL and fraction contract address.
This is a working demo of the Governance Terminal. Try typing help, status, balance, or submit build a yield aggregator.
The production terminal requires token-gated authentication. This demo simulates the interface with a mock wallet.
- ▸Read-only access to queue status
- ▸Public deploy feed
- ▸Submit 1 prompt per 24h
- ▸Standard queue priority
- ▸Real-time deploy status
- ▸Email on deploy
- ▸Submit 5 prompts per 24h
- ▸High queue priority
- ▸24h whitelist on new dApp fractions
- ▸Direct prompt → AI (skip orchestrator review)
- ▸Unlimited prompt submissions
- ▸Highest queue priority
- ▸Co-deploy naming rights on dApp
- ▸Revenue share on all AI-built dApps
- ▸Private Telegram with founding team
Queue priority is a pure mathematical function of bag weight. No subjective scoring. No moderator override. The Orchestrator dequeues using Redis ZPOPMAX — highest score wins.
// Priority score computation
// Evaluated at queue dequeue time
function priorityScore(wallet: string): number {
const balance = await getOnChainBalance(wallet);
return Math.floor(balance / 1000);
}
// Example:
// 142,500 $FRACTAL → score: 142
// 500,000 $FRACTAL → score: 500
// 10,000 $FRACTAL → score: 10
// Redis sorted set
await redis.zadd(
'governance:queue',
priorityScore(wallet), // score
jobId // member
);
// Dequeue: highest score first
const next = await redis.zpopmax('governance:queue');- ▸Governance Terminal v0.1 — read-only queue viewer
- ▸Wallet authentication via signed message
- ▸Manual prompt submission via authenticated API
- ▸Full Governance Terminal UI launch
- ▸Real-time job status via WebSocket
- ▸Tier-based prompt limits enforced on-chain
- ▸Community voting on prompt submissions (upvote/downvote)
- ▸Multi-sig prompt approval for contract upgrades
- ▸On-chain governance registry for deployed dApp modifications
- ▸Full governance handoff — founding team loses admin keys
- ▸DAO treasury takes over compute reserve
- ▸Protocol parameters (milestones, tiers) become community-votable
The Governance Terminal issues a signed JWT after verifying your wallet's $FRACTAL balance via on-chain state. The JWT encodes your bag_weight and expires every 24 hours. No custodial deposit required.
Queue priority is computed per wallet address, not aggregated across wallets. Splitting bags reduces each wallet's score. There is no benefit to splitting — only dilution.
If a prompt sits in queue for more than 7 days without executing (e.g., compute reserve is depleted), it is returned to the submitter with a full refund of any associated fee.
The Orchestrator has a constraint layer: no illegal content, no re-entrancy exploits, no phishing interfaces. Within those rails, bag holders have full creative authority.
That's the risk. Governance is about directing the AI's output — not a guaranteed return. You own the fractions of the dApp. Its revenue depends on what it is and whether users adopt it.
priority_score = floor(bag_weight / 1000). A wallet holding 142,500 $FRACTAL has a score of 142. Scores are re-evaluated at queue dequeue time — if you sell between submission and execution, your score drops accordingly.
No. The Orchestrator dequeues strictly by ZADD score in Redis. The founding team has no special endpoint to inject jobs ahead of community prompts. The only way to get higher priority is to hold more $FRACTAL.
DeepSeek-R1-32B for architectural reasoning and Qwen2.5-Coder-32B-Instruct for code generation. The Orchestrator selects the model based on prompt complexity. Both run locally — your prompt never leaves our infrastructure.
Yes. You can submit a prompt targeting an existing dApp: 'Add staking to dApp #003'. The Orchestrator will clone the repo, apply the AI's diff, redeploy, and push a new contract version if the ABI changed.