Every block explorer shows you logs. Almost none of them tell you what happened. “Transfer(address,address,uint256)” with three hex blobs is technically decoded — and practically useless. Here’s how OpenScan’s AI insight engine goes further.
Layer 1: deterministic decoding
First we do what any explorer should: match event signatures against known ABIs, including every verified contract on the chain. For verified contracts this is exact — we use the contract’s own ABI, so parameter names and types are always right.
Layer 2: behavioral inference
For unverified contracts — the majority on any chain — we infer. Token-like Transfer events with 18-decimal amounts and a matching Approval flow? That’s an ERC-20, and the AI insight engine labels it accordingly, with a confidence signal. Swap-shaped event clusters across two token contracts get recognized as DEX activity. This is inference, not magic: we mark inferred decodes so you can always tell them apart from verified ones.
Layer 3: the summary
The final step is compression: turning a transaction’s full trace — calls, events, token movements — into one sentence. “Swapped 1,200 USDC for XDC on a DEX” replaces a wall of internal transactions. That’s the line you see at the top of a transaction page on xdcscan.io.
All three layers ship in every explorer we operate — XDC Network today, with EVS, Wanchain, and Velas on the same stack — and the decoded data is available over the REST API v2 for builders who want it in their own apps.