AI Model Details

• Type: LSTM RNN, 128 hidden units.

• Training: On-chain, federated learning.

• Input: Message metadata (length, timestamp).

• Output: Shard ID, latency estimate.

Toggle AI suggestions in the dApp settings to customize your experience.

DOXAI ensures your messages are settled reliably on BSC.

Process

• Validation: zk-Rollup batch verified by validators.

• Settlement: Inscribed in BSC’s Merkle trie.

• Finality: 3 seconds post-block inclusion.

Specs

• Batch Size: ~100 messages.

• Gas: ~1M per batch (~10,000/message).

• Failure Rate: <0.01% (network congestion).

BSC congestion can delay clearing—check network status if delivery slows.