Roadmap
Where PulseGrid is, and what is next. We list the credibility gaps openly, because for a network that settles real value the honest gaps matter more than the feature list.
Shipped and live on mainnet
- The full settlement engine on BOT Chain mainnet (chainId 677), settling in canonical USDT, with all four PulseGrid contracts Blockscout-verified.
- Value-layer SLAs: settlement checks data quality (signed in-range readings), not just liveness.
- Permissionless keepers: anyone can drive settlement for an equal tip from a dedicated pool, without weakening the guarantee.
- Gasless devices via a self-hosted, MegaFuel-shaped sponsor relayer (devices spend zero gas).
- Real-hardware device SDK: an importable Go package plus a standalone binary with an encrypted keystore.
- An adversarial Foundry suite (81 passing tests) and an internal, AI-assisted review (see Security).
Product: what is next
- Multi-asset settlement (additional stablecoins via a DEX; the vault is already asset-agnostic).
- Batched settlement to cut per-interval overhead as the fleet grows.
- Richer bonding and reputation: risk-priced bonds and a stronger reputation signal.
- A real anomaly-detection model feeding the advisory layer (kept strictly non-authoritative; it never touches settlement).
Trust and decentralization: the gaps we are closing
These are the items a serious operator or partner should ask about. In rough order of leverage:
- External smart-contract audit. The largest outstanding trust item for real-USDT settlement. The contracts have an internal AI-assisted review and an adversarial test suite today; an independent third-party audit is planned, and the report will be published.
- Bug bounty. A public program (once the audit lands) so researchers are paid to find issues.
- Key custody. Today a single operator key holds the deployer, treasury, relayer, and settler roles. The plan is to move the treasury and contract owner to a multisig and split the hot keys, and to run a second keeper so the permissionless keeper set is demonstrated in practice, not just in code.
- Real-hardware pilot. The browser and CLI paths are on-ramps; running one physical device end-to-end is the next credibility step.
- Availability model. Because BOT Chain has a gas floor, the streaming fleet runs in short funded windows rather than 24/7. We will either fund sustained operation or publish a clear windowed schedule plus a public status page, so availability is never ambiguous.
- Native gasless. BOT Chain advertises an EOA paymaster, but none is live on mainnet or testnet and it does not sponsor blob transactions (logged as BOUNTY_LOG B-008). We run a self-hosted sponsor and will point the device path at a native paymaster the moment one ships. Devices are provably zero-gas either way.
- Domain and channels. A custom domain and public community channels (currently the project runs on a vercel.app subdomain with no socials yet).
Nothing above changes the on-chain guarantee: an operator cannot be paid for undelivered or out-of-range data, and the agent cannot stiff an honest operator, because on-chain policy re-derives every settlement.