s24

Completely autonomous AI agent that trades on Solana Blockchain

Comment 1

Our YouTube video DOES NOT HAVE AUDIO. Use this link for the demo: https://www.loom.com/share/868e7e30057d40fd972976755f06baa3

Inspiration

Your agent can already create tools, write code, do research, find patterns, and generate insights.
It can produce real value faster than most humans. But one bottleneck still remains: monetization.

Today, agents can create. Humans still have to sell. Humans still wire payments, set pricing, and route execution.

That gap is where we focused s24.
We built a system that moves AI closer to economic autonomy by connecting intelligence to execution in a controlled way: market data in, decisions through risk gates, actions on a blockchain-aligned stack, and full operational visibility.
For this project, we set up paper trading on Solana Devnet.

What it does

s24 is an autonomous trading control plane for Solana that:

  • ingests live market data from Jupiter
  • runs AI-assisted trade cycles through OpenClaw
  • enforces strict backend risk rules before any execution
  • supports paper mode and solana devnet execution mode
  • logs all activity to PostgreSQL
  • provides a dashboard for monitoring, control, and safety

How we built it

  • OpenClaw, AWS Bedrock, EC2 with VPS for agent orchestration
  • Node/Express trading API for execution logic and risk enforcement
  • PostgreSQL/RDS schema for orders, fills, positions, snapshots, and risk events
  • Jupiter quote adapter for live tick ingestion
  • Next.js dashboard for operations, controls, and session visibility
  • AWS scripts and templates for deploy, bootstrap, watchdog, and validation

Challenges we ran into

  • keeping AI orchestration flexible while making safety deterministic
  • syncing API contracts with frontend models during rapid iteration
  • integrating and coordinating infra, backend, and UI changes in parallel

Accomplishments that we're proud of

  • end-to-end working loop from live data to paper execution
  • strong server-side safety model with kill switch + anomaly checks
  • operational tooling for deploy, health checks, and service recovery
  • clean dashboard flow for trading, history, and risk controls

What we learned

  • prompt intelligence is not enough without hard backend guardrails
  • autonomous systems need observability as much as strategy
  • contract-first API design saves time across teams
  • blockchain-style transparency (auditable events/state) improves trust

What's next for s24

  • expand risk + strategy layer and performance analytics
  • deepen Solana devnet execution and transaction lifecycle tracking
  • add multi-user auth and role-based controls

Built With

Share this project:

Updates