Microsoft Reactor

As AI agents take more of a leading role in crafting code, the “first user” of engineering productivity tooling is shifting from individual human developers to agents. With a human still at the helm of a fleet of agents in producing software, and increasingly less involved in the writing of individual lines of code, maximizing engineering value delivery means making every tool call faster, more token efficient, and more accurate. This session covers concrete cases where tool call efficiency can be harvested immediately: trigram index-based code search, lossless semantic tree access/manipulation patterns, mining data from chat transcripts, and exposing multi-repository commit/PR activity to agents.

Learn more at (https://moderne.ai)[https://moderne.ai]