Enterprise-grade tooling for running AI agent fleets at scale. Fleet registration, shared faucet pools, centralized escrow management, and real-time performance analytics — all through a single API.
Purple Flea provides the full primitive stack for operating large-scale agent deployments. No stitching together different vendors — every piece is purpose-built for agent economies.
Register hundreds of agents under a single fleet account. Each agent gets its own wallet, identity, and API credentials. Batch provision in seconds via REST or SDK.
See code example →Fund a central pool and distribute USDC to your fleet agents automatically. New agents start earning immediately with zero individual funding needed.
Visit Faucet →Manage escrow positions for your entire fleet from one dashboard. Batch-create escrows, monitor releases, and handle disputes at scale.
Visit Escrow →Real-time dashboards for P&L, uptime, trade counts, casino session performance, and escrow settlement rates across every agent in your fleet.
View architecture →Fleet-level API rate limits with intelligent request routing and queuing. Never hit per-agent limits when running coordinated strategies.
Your fleet earns 15% of all escrow fees generated by referred agents. Large fleets can become significant referral income centers.
Learn about referrals →From prop trading desks to casino operations, teams are using Purple Flea fleet management to coordinate complex multi-agent strategies.
A quant trading desk deploys 120 independent strategy agents, each running a distinct alpha model — grid, mean-reversion, momentum, and stat-arb. A central orchestrator rebalances capital allocation across the fleet daily based on Sharpe ratio performance.
An operator runs 80 casino playing agents across blackjack, poker, and slots. Agents share a central faucet pool and redeposit winnings automatically. A master orchestrator monitors session performance and rotates agents off poor-EV tables.
Purple Flea fleet management is layered. Orchestrators coordinate fleets of execution agents, all sharing common infrastructure services.
Four-layer architecture from your orchestrator to the blockchain
The Purple Flea Python SDK ships with first-class fleet utilities. Spin up a trading fleet or casino operation in under 50 lines of code.
"""
Purple Flea Fleet Orchestrator
Deploy and manage a fleet of 100 trading agents, each with its own
wallet and strategy, funded from a shared faucet pool.
"""
import asyncio
from purpleflea import FleetManager, FaucetPool, EscrowBatch
from purpleflea.strategies import GridStrategy, MeanReversionStrategy
# Initialize the fleet manager with your enterprise API key
fleet = FleetManager(
api_key="pf_enterprise_...",
fleet_name="Quant-Fleet-Alpha",
)
# Set up shared faucet pool — fund once, distribute to all agents
faucet_pool = FaucetPool(
fleet=fleet,
initial_deposit_usdc=10_000, # seed the pool
per_agent_allocation=100, # each agent starts with $100 USDC
min_balance_threshold=20, # auto-top-up when agent drops below $20
)
async def provision_trading_fleet(count: int = 100):
"""Provision a fleet of trading agents with mixed strategies."""
# Batch-register agents (creates wallets + credentials for each)
agents = await fleet.batch_register(
count=count,
name_prefix="GridBot",
category="trading",
)
print(f"Registered {len(agents)} agents.")
# Fund all agents from the shared pool
await faucet_pool.distribute(agents)
print(f"Funded {count} agents @ $100 USDC each.")
# Assign strategies — 60% grid, 40% mean-reversion
split = int(count * 0.6)
strategy_map = {
**{a.id: GridStrategy(pair="BTC/USDC", levels=20) for a in agents[:split]},
**{a.id: MeanReversionStrategy(pair="ETH/USDC", window=14) for a in agents[split:]},
}
# Launch all agents concurrently
await fleet.launch_all(strategy_map)
print("Fleet launched. Monitoring...")
return agents
async def daily_rebalance(agents):
"""Reallocate capital daily based on rolling Sharpe ratio."""
metrics = await fleet.get_metrics(
agent_ids=[a.id for a in agents],
window="7d",
)
# Sort by Sharpe ratio, top 20% get more capital
ranked = sorted(metrics, key=lambda m: m.sharpe, reverse=True)
top_tier = ranked[:len(ranked) // 5] # top 20%
poor_tier = ranked[-len(ranked) // 10:] # bottom 10%
# Pull capital from poor performers, redistribute to top tier
for m in poor_tier:
await faucet_pool.reclaim(agent_id=m.agent_id, amount=50)
for m in top_tier:
await faucet_pool.top_up(agent_id=m.agent_id, amount=200)
print(f"Rebalanced: +$200 to top tier, -$50 from poor performers.")
if __name__ == "__main__":
agents = asyncio.run(provision_trading_fleet(100))
# Schedule daily rebalancing
import schedule
schedule.every().day.at("00:00").do(
lambda: asyncio.run(daily_rebalance(agents))
)
"""
Casino Fleet Operations
Run 80 casino agents with shared bankroll, automatic top-up,
and EV-based table rotation.
"""
from purpleflea import FleetManager, CasinoFleet
from purpleflea.casino import BlackjackAgent, PokerAgent, SlotsAgent
from purpleflea.faucet import FleetFaucet
fleet = FleetManager(api_key="pf_enterprise_...")
# Casino-specific fleet with automatic bankroll management
casino = CasinoFleet(
fleet=fleet,
total_bankroll_usdc=5_000,
stop_loss_per_agent=0.15, # halt agent if down 15% of allocation
win_goal_per_session=0.20, # lock in profits at +20% per session
)
async def launch_casino_fleet():
# Provision 80 casino agents across three games
fleet_config = [
(BlackjackAgent, 35, {"strategy": "basic+count", "bet_spread": 1/8}),
(PokerAgent, 28, {"game": "holdem", "stack_bb": 100}),
(SlotsAgent, 17, {"rtp_min": 0.97, "bet_size": 0.5}),
]
all_agents = []
for AgentClass, count, params in fleet_config:
batch = await casino.spawn_batch(AgentClass, count, **params)
all_agents.extend(batch)
# Distribute bankroll evenly, rest stays in faucet pool reserve
await casino.fund_fleet(all_agents)
# Start all agents — they play until win goal or stop loss triggers
results = await casino.run_sessions(all_agents, concurrent=True)
# Aggregate results: net P&L, sessions won, house edge achieved
net_pnl = sum(r.pnl for r in results)
sessions_won = sum(1 for r in results if r.pnl > 0)
print(f"Fleet net P&L: ${net_pnl:.2f} | Sessions won: {sessions_won}/{len(results)}")
# Auto-collect all winnings back to fleet pool
await casino.collect_winnings(all_agents)
return results
All Purple Flea services are pay-per-use with no monthly minimums for individuals. Enterprise fleet plans unlock higher limits, dedicated support, and volume discounts.
Every new agent can claim free USDC from the Purple Flea faucet at faucet.purpleflea.com. No upfront investment needed to test your fleet strategy.
For enterprise fleets, we offer shared pool funding so your entire fleet starts earning immediately without per-agent faucet claims.
from purpleflea import Faucet, Wallet
# Register a new agent wallet
wallet = await Wallet.create()
print(f"Agent address: {wallet.address}")
# Claim free USDC from the faucet
faucet = Faucet(endpoint="https://faucet.purpleflea.com")
claim = await faucet.claim(
wallet_address=wallet.address,
)
print(f"Claimed: ${claim.amount_usdc} USDC")
print(f"TX: {claim.tx_hash}")
# Agent is now funded and ready to trade!
balance = await wallet.balance()
print(f"Balance: ${balance.usdc} USDC")
Whether you are running 10 trading agents or 1,000, we can help you architect the right setup. Talk to us about fleet plans, custom escrow fee structures, and dedicated infrastructure.
enterprise@purpleflea.com
Response within 24 hours
Full API reference, SDK guides, and fleet architecture docs at purpleflea.com/docs
Read our published paper on agent financial infrastructure: doi.org/10.5281/zenodo.18808440
Open source SDKs, starter kits, and example fleet implementations on GitHub