Nois

• Name: Nois
• URL: https://nois.network/
• Category: randomness relay network / IBC randomness middleware / drand-to-Cosmos delivery infrastructure
• Tags: cosmos-ecosystem
• Summary: Nois is drand delivery middleware for Cosmos. The randomness source is upstream; Nois runs the relay, verification, payment, and callback plumbing that makes it usable from IBC-connected CosmWasm apps. That is the part worth tracking. The chokepoints are bot liveness, gateway and proxy topology, payment policy, and whether apps are really depending on drand or on a narrower Nois-operated transport layer.
• What it does:
  • Lets CosmWasm applications on IBC-enabled chains request randomness through a local Nois proxy contract and receive callbacks once the requested round is available
  • Uses Nois-chain contracts such as nois-gateway, nois-payment, and nois-drand to queue requests, verify drand beacons, manage balances, and route callbacks
  • Bridges verified randomness back to consumer chains through nois-proxy contracts, so dapps do not need to manage the underlying IBC flow directly
  • Incentivizes offchain drand bots to submit requested drand rounds to Nois, with rewards depending on speed, allowlisting status, and whether a round was actually requested by a dapp
  • Publishes a standard library and toolbox for integrating randomness requests and callback handling into CosmWasm contracts and JavaScript simulations
• Key claims:
  • The docs describe Nois as a Proof of Stake blockchain protocol that brings secure, unbiased, and cost-efficient randomness to developers via IBC.
  • The most important analytical point is that Nois does not claim to generate the randomness itself. The docs explicitly say drand is the source of secure randomness and that Nois verifies and distributes that randomness to consumer chains.
  • The contracts README exposes a useful layered architecture: nois-gateway routes jobs, nois-payment holds per-customer balances and decides whether requests are fundable, nois-drand verifies incoming beacons, nois-proxy handles appchain-side requests and callbacks, and nois-icecube adds an additional incentive reservoir.
  • The drand-bot docs make a key control surface explicit: bots cannot bias the randomness, but they can materially affect liveness and delivery speed. The system needs at least one functioning bot, and reward design favors fast, allowlisted operators whose submissions land early enough in proposer mempools.
  • That means Nois is analytically useful as a downstream relay-and-delivery layer above drand rather than merely a consumer of drand. It turns public randomness into a chain-specific middleware market with payments, queueing, transport, and callback semantics.
  • Nois belongs in the active corpus because it cleanly separates randomness source, bot relay, onchain verification, payment policy, IBC transport, and application callback handling. That decomposition is more useful than filing it as a generic oracle, generic randomness, or generic Cosmos infra project.
  • The strongest comparison questions for later work are: how much delivery power concentrates in fast or allowlisted bot operators, whether payment/burn/community-pool policy changes app incentives, and when a downstream randomness-delivery chain becomes more important than the upstream beacon itself.
• Whitepaper: No classic Nois whitepaper dominated this pass. The clearest primary materials were the official site, docs repository, contracts repository, standard-library README, and organization overview. Those sources were enough to map the delivery architecture and incentive/control surfaces without relying on secondary explainers; see ../whitepapers/nois-primary-sources-2026-05-12.md.
• Sources:
  • https://nois.network/
  • https://docs.nois.network/
  • https://github.com/noislabs
  • https://raw.githubusercontent.com/noislabs/docs/main/src/what_is_nois.md
  • https://raw.githubusercontent.com/noislabs/docs/main/src/how_it_works.md
  • https://raw.githubusercontent.com/noislabs/docs/main/src/drand.md
  • https://raw.githubusercontent.com/noislabs/docs/main/src/drand_bots.md
  • https://raw.githubusercontent.com/noislabs/nois-contracts/main/README.md
  • https://raw.githubusercontent.com/noislabs/nois/main/README.md

Internal linkages

• Keep this note on the strongest few reads: drand, pyth-entropy, and arpa-network.

• The point is not to build a giant randomness peer cloud. The point is to keep source, relay, payment, and callback control surfaces separate.

• Last reviewed: 2026-06-01 UTC