Switchboard collapses cross-chain coordination onto a single Solana program, so an agent or
contract acts across many chains through one integration instead of N bridges. That one decision
cascades into every feature below — atomic composability, agent-speed latency, and a topology you
can operate like a database. Here is each capability in depth.
Coordination latency
<400ms
Chains supported
50+
Ops / second
10,000+
Uptime SLA
99.9%
01
One integration, not N bridges
One call fans out to every chain.
An agent, intent-solver, or contract issues a single write and Switchboard propagates it to every supported destination. There is no bespoke bridge, validator set, or message format per corridor to build and maintain. Adding a chain does not add integration code — the coordinator absorbs the fan-out.
✓One SDK call reaches every supported chain; no per-route wiring.
✓The layer autonomous agents need to act across many chains at once.
✓New chains land roughly monthly; Enterprise can sponsor a 6-week integration.
02
Solana-anchored coordinator
One finality clock for every route.
Every supported chain shares a single source of truth: a Solana program that commits source state in the next slot. There is no per-corridor governance, no bespoke validator set per route, and no negotiation between two chains about whose finality wins. (This is a coordination clock — not the Solana price oracle that shares the name.)
✓Source events land on the coordinator in ~400ms (one Solana slot).
✓No per-path DVN, guardian quorum, or gateway contract to configure.
✓Anchor (Rust) programs are auditable the way a Solidity contract is.
03
Sub-400ms end-to-end
Agent-speed cross-chain coordination.
From source commit to destination verification, the median route lands inside 400ms. A competing relayer pool races the destination transaction, so you get p50 ~250ms and p99 under 400ms instead of the multi-second quorum times most message layers inherit. Fast enough for autonomous agents, perps, orderbooks, and real-time game state.
✓p50 ~250ms, p99 <400ms in internal benchmarks.
✓Relayers compete on latency; the fastest valid dispatch wins.
✓Fast enough for agentic payments, perps, orderbooks, and game state.
04
Atomic composability across chains
Cross-chain state as a single call.
Commit once to the coordinator and the write propagates to every destination inside one latency budget, each with a per-route finality policy and a verifiable receipt. For an agent orchestrating a strategy across chains, that turns N brittle integrations into one composable action — one write, many chains, one budget.
✓Commit-once fan-out with per-route finality policy.
✓Verifiable receipts an agent or contract can check.
✓Turns cross-chain orchestration into a single composable call.
05
2-service architecture
A topology you can draw on a napkin.
A Customer API gateway (auth, rate limiting, request proxy) and a Core Engine (oracle, chain ops, billing, coordination) — nothing else. MongoDB or PostgreSQL sits behind it, chosen with one env var. Operating Switchboard is closer to running a database than running a chain.
✓Customer API on :3000, Core Engine on :3001.
✓Persistence is pluggable: MongoDB (default) or PostgreSQL.
✓Self-host the whole stack with `npm run docker:up`.
06
Pluggable verifiers
The strongest available proof, per chain.
Each destination uses the best proof it can support: a light-client receipt where the chain allows it, a BLS threshold signature from the relayer pool otherwise, and zk-proof receipts on chains that verify them natively. You mix and match per route.
✓Light-client, BLS threshold, or zk-proof — configurable per route.
✓Security posture is a per-route decision, not a global compromise.
✓Verified verifiers ship in-box; no ISM to design and audit.
07
~40% gas savings via batching
Amortize calldata across the bundle.
The coordinator commits messages in batches, so destination chains amortize calldata costs across the whole bundle instead of paying per route. On busy corridors that works out to roughly 40% gas reduction versus naive per-chain dispatch.
✓Batch cadence is tunable against your latency budget.
08
Bring-your-own signers
Keep signing material under your custody.
Enterprise teams can plug their own threshold-signing nodes into the relayer pool. Signing keys never leave your infrastructure, so you get managed relayer economics without handing over custody of the material that authorizes cross-chain writes.
✓Plug threshold-signing nodes into the relayer pool.
✓Signing material stays in your own custody.
✓Managed relayer SLA without surrendering key control.