Lalit Kumar

Bridging Legacy C++ and Go with AI Agents

Many Go teams are not replacing C++ overnight. They are building new Go services next to legacy C++ systems that still own critical paths, shared data models, and years of operational knowledge. The hard part is usually not syntax. The hard part is coordinating changes across code, configs, telemetry, tests, and runbooks without losing control. This talk shows a practical way to use AI agents and MCP as workflow glue between those two worlds. Using one concrete migration story, it walks through how an agent can map call-sites across a C++ and Go boundary, draft a migration plan, support safe experiments with flags or shadow traffic, propose reviewable changes on the Go side, and help keep observability and documentation in sync. The goal is not to put agents on the hot path or let them rewrite risky low-level code. The goal is to expose each language’s tools consistently, set clear safety boundaries, and use agents where workflow automation actually helps. Attendees will leave with a repeatable migration playbook, a clearer trust model for agent-assisted engineering, and practical ideas they can try in their own mixed-language systems.