This is brilliant. Pairing this with codemode would be sick. Even cooler: co-locating the speculative function calls with the inference itself. Dynamic Workers + Worker AI on the same box. AI smart placement.
Introducing Speculative Programmatic Tool Calling (sPTC)!
A general class of technique for speculating on tool calls during code generation in a harness and queuing them early to overlap with token generation + REPL execution time.
Blog: alexzhang13.github.io/blog/2026/spec…
