Engineering updates, research, and announcements.
Our adversarial multipliers are static. Our path discovery misses routes. Here's what we're fixing.
We debated keeping the engine closed. Here's why we decided against it.
We assumed Wormhole would win on reliability and deBridge on speed. We were wrong on both counts.
Our engine kept picking routes that looked worse on paper. It took us a week to realize it was right.
First version took 14 seconds per query. That was obviously unusable. Here's how we got it under 10 milliseconds.
The transfer went through. No errors. No delays. We still lost $1,800 by picking the wrong bridge.