That's a good question. They've built an amazing system. Google and ex-Google engineers love the code intelligence/search systems that Google has, and those are

That's a good question. They've built an amazing system. Google and ex-Google engineers love the code intelligence/search systems that Google has, and those are powered by Kythe and related systems inside Google. At a product level, many ex-Googlers miss those systems in their new companies and fill the gap by setting up Sourcegraph internally. Tech-wise, Kythe is optimized for offline analysis, and we use servers that speak LSP (https://github.com/Microsoft/language-server-protocol https://github.com/Microsoft/language-server-protocol), which is built for online analysis and use while you're editing code. That is what I see as the main difference. We would love for the approaches to converge, or for the representation/configuration to be compatible.