Google protocol-buffers and Apache Thrift are two industrial-strength projects with fairly similar scope. The primary difference is that these projects use a custom IDL language to generate both clients and servers. They are also used for for serializing data to store, or else in distributed systems that are not updated in lock-step, as we assume in Gluon; such scenarios introduce the new requirement of being able to extend service contracts without breaking existing code. Both protocol-buffers and Thrift have a versioning facility to address this.
Either project could be used instead of Gluon, however:
protocol-buffers lacks TypeScript support and Thrift has it as third-party beta
neither has F# support, though C# is probably working (so, no Unions, option, etc)
the IDL approach is a bit more tailored for truly multiple-language projects; inferring schema from F# code is more convenient if that is the principal language
These projects should definitely be considered if requirements arise of interacting with more languages such as Python or serializing semi-structured data. It should be possible to extend future versions of Gluon to, for example, generate Thrift service descriptors and services from existing F# specifications.