What’s $JELLY?
$JELLY is an open-source prediction market protocol built for the multi-chain world. While traditional prediction platforms lock users into regional rules and single ecosystems.
https://t.co/nCD2l574Qb
👇👇👇👇
Shaping agents on-chain - JellyOS
Ready to fight the bears & equipped for the @FIFAWorldCup
Official framework and “OS layer” for the Jelly ecosystem set
•Single canonical starting point for building anything Jelly-native
•Coordination brain for agents, bots, on-chain and off-chain components
•Extensible interface so other Jelly repos can plug in cleanly
•Designed around a multi-armed, multi-agent “octopus” architecture
•Actively evolving core rather than a static spec
•Intended to standardize patterns across all Jelly projects and tooling
We’ve got a big month ahead!
As the FIFA World Cup approaches, we have officially opened manual predictions (without Agents), something the community has been waiting on for a long time.
Over the past few weeks, we have seen agents take over the space and drive new features forward. Alongside this shift, we have continued building and scaling Claude wrapper, Jelly CLI, and our agents API (currently invite-only), positioning Jelly at the center of this prediction ai race.
We are already seeing volume increase as traders begin predicting using agents and through our beta testers. This is only the beginning, we expect both prediction activity and token volume to accelerate significantly in the coming weeks.
Jelly on BSC will soon be live for predictions use, expanding access and opening the door for even broader participation.
Our growth has been exponential. We have adapted quickly and are positioning ourselves to become the hub for agent-driven prediction markets and trading on BSC, and beyond. With a strong team and a solid floor, we are continuing to build with conviction.
Our long-term vision is clear: non-holders will need to acquire JELLY to access and compute through Jelly-powered tools. This will create sustained buy pressure as usage scales. However, this will be for professional agents and the current agents you see are all open-source. This means the community has something almost as powerful as the agent we’ll release so the open-source agent can also be valued highly.
Since launch, over 430,000 users have already passed through the Jelly platform. With the World Cup around the corner, we expect that number to at least double.
Thank you to everyone who believes in AI, in Jelly, and in the vision we are building. This is just the beginning.
📊 Market Update
Bloodbath alert! Crypto heavyweights are taking a beating, with BTC and ETH down around 6.8%. SOL and ADA are getting crushed, plummeting 8.66% and 12.13% respectively. Meanwhile, XRP and BNB are "leading" the pack with relatively modes… https://t.co/RpTfe0FJRD
We’re building Jelly in the open: an agent-first, community-owned prediction market stack with no paywalls and no black boxes. Transparent, composable, and here for the long haul. Come build with us! @BNBCHAIN
BINANCE FOUNDER CZ ABOUT AI AGENTS:
CZ says AI agents are going to transact 1,000 times more transactions than humans can do.
"And AI is going to use crypto."
This is not “ask AI what to do.”
This is AI that can do it.
We moved from single-model dependency to a model-agnostic system. JellyOS connects to
OpenRouter, OpenAI, Anthropic, or even local models, and dynamically rotates between them. Rate limits, failures, or cost constraints are handled automatically.
We moved from static prompts to adaptive swarm intelligence. Complex queries are decomposed, executed in parallel, and recombined into a single decision.
We did not set out to build another wrapper.
JellyOS started as a simple interface, a thin layer over Claude.
JellyOS/agent 700+ downloads since launch
A faster way to query models, experiment with prompts, and explore trading ideas. It worked, but it was limited. Every decision still depended on a single model call. No memory, no execution layer, no real autonomy
At its core, JellyOS is an agent runtime:
•An AgentRunner that loops, reasons, and executes tools
•A SwarmRouter that splits complex problems into parallel sub-agents
•A SignalEngine that turns raw data into actionable trades
•A FeedManager pulling from 20+ live market and on-chain sources
•A Wallet + Vault system that actually holds and protects value
Jelly Agents are a fleet of AI agents designed to analyse prediction markets manage vault positions across multiple chains and settle automatically. Each agent is a focussed module that can be deployed independently or combined into automated workflows. This includes options to create your own predictions or let agents handle the work for you.
https://t.co/iL2EOdHJxi
Huge thank you to everyone who used, broke, and improved Jelly Claude.
Now we invite you to help shape JellyOS.
🔗 JellyOS: https://t.co/UUgAszotN6
🌐 Website: https://t.co/Skk7Af3WfC
💬 Telegram: https://t.co/l6F9vh3Vz9
Let’s keep pushing open‑source agents forward – for web2, web3, and whatever comes next.
Over the last months we pushed Jelly Claude from a simple agent launcher into a full AI prediction & trading system – multi‑chain wallets, 38+ skills, 47+ agents, Telegram control, DeFi, DEX routing, prediction markets, on‑chain analytics and more.
Agentic really works.
What Jelly Claude achieved for us and the community:
•Unified EVM & Spl wallets from first run
•Plug‑and‑play skills for DeFi, NFTs, prediction markets, analytics
•Agent templates for traders, scanners, portfolio managers & on‑chain explorers
•A real production testbed for agentic execution in web3
That work doesn’t go away – it’s the foundation of JellyOS.
Today we’re announcing the next chapter:
🔁 We’re moving away from the Claude framework
🧠 And into JellyOS – the first fully on‑chain, open‑source prediction agent CLI for running, scheduling, trading, scanning & much more.
One CLI. Multi‑agent swarm. 20+ chains. Built for serious ops.
Agentic is the way.
As agents move from “toy examples” to always‑on systems running portfolios, strategies, infra & ops, the design space explodes:
•Autonomous scanners and risk guards
•Cross‑chain arbitrage & prediction bots
•On‑chain + off‑chain data fusion
•Scheduling, monitoring, and self‑healing systems
We’re just getting started.