THE HIDDEN ROUTING PROBLEM

On Solana, no transaction goes directly to the slot leader.
Every TX takes a slow, observable, multi-hop detour.

✗ DEFAULT ROUTING

Wallet → RPC Node

RPC → Relay Node

Relay → Validator

Val → Leader

~340ms • 4 hops • 3 MEV points

✓ REVM ACO ROUTING

Wallet → Leader TPU

.

~9ms • 1 hop • 0 MEV points

ACO ROUTING ENGINE

01

SCOUT

Reads the leader schedule 1 full epoch ahead.
Resolves TPU addresses, geo-latency, reliability scores.
Maps all food sources before dispatching the colony.

02

PHEROMONE

Probe agents measure latency, success rate, slot accuracy.
High-scoring paths gain pheromone. Low paths evaporate.
Self-healing — dead routes vanish in seconds.

03

COLONY

TX submitted → ACO selects optimal path → Direct QUIC to leader TPU.
1 hop. No relays. No observation points.
Result recorded. Pheromone updated. Loop continues.

MAINNET BENCHMARKS

Benchmarked during peak congestion (>95% slot utilization)

0x FASTER LANDING

0% SUCCESS RATE

~0ms AVG LATENCY

0% SLOT ACCURACY

METRICDEFAULTREVM

Landing latency~340ms~9ms

Network hops41

TX success rate~82%98.7%

MEV exposure3+ points0

Convergence time—<2s

IN THE WILD

💰

DEX ROUTING

Jupiter and pump.fun swaps through default RPC average ~2s with retries.
Transactions routed through ACO land in the current slot at ~9ms.
Same swap. Different infrastructure.

⚡

LATENCY-SENSITIVE OPS

Arbitrage windows close in <200ms. Default routing delivers in ~340ms.
ACO pre-computes the fastest path to every upcoming slot leader.
When the signal fires, the route is already resolved.

🛡

OBSERVATION SURFACE

Default path: 3+ relay nodes. Each one sees pending transactions.
Each hop is an observation point for sandwich extraction.
Direct-to-leader TPU routing: zero intermediate nodes. Zero observation surface.

BUILT ON REAL SCIENCE

Not another whitepaper-only project.

REVM's core algorithm is backed by 34,000+ citations across 30 years of peer-reviewed research.

2 papers published

Textbook published

1 paper published

KEY PAPERS

1996 Dorigo et al. — Ant System (IEEE, 18K+ citations)

1998 Di Caro & Dorigo — AntNet (JAIR, 2.8K+ citations)

2004 Dorigo & Stutzle — ACO Textbook (MIT Press, 12K+ citations)

2003 Sim & Sun — ACO Routing (IEEE, 680+ citations)

$REVM

The pheromone map improves with every probe.
More participants, more data, faster convergence.

1 HOP

DIRECT TO LEADER

~9ms

LANDING LATENCY

0

MEV EXPOSURE

More nodes → more probe data → better pheromone map → faster routes.
The colony scales itself.

ENTER THE COLONY

CA: EX4nLYSaHwBwtbsJwBbWjRUrfyjNYBaD81PPi3kQpump

RE[VM]