How it works | RLock Docs

How it works (pipeline)

Intents – capture the full user goal in one request.
DAG build – map dependencies: what must run first, what can run in parallel, what’s impossible.
Chunking / layering – split the DAG into executable units that won’t blow compute or lock the same hot accounts.
OCC checks – Optimistic Concurrency Control verifies the world state you planned against hasn’t drifted.
Rail selection – pick L1 or ER-style lane based on congestion and reliability goals.
Fee planning / budgeting – lock CU limit and price so inclusion is likely without overpaying.
Simulate end-to-end – validate the whole package before spending a single lamport on live tx.
Tighten & serialize – remove redundant work, compact the message, serialize final bytes.
Send – submit the finalized bundle on the chosen rail; confirm and surface a clean result.