RESEARCH · RESEARCH · #1535
Akka runs spec-driven AI porting experiment across 65 open-source projects
Akka tested a spec-driven workflow for AI-assisted software porting across 65 open-source projects, using a delivery harness that cycled through discovery, specification, porting, benchmarking, and improvement. The initial tranche took 99.3 hours and consumed 9.41 billion tokens; Akka reported lines-of-code or performance improvements in 57 of the 65 ports and used Claude with Akka Specify for implementation, with Sonnet averaging 61 minutes per port versus 120 minutes for Opus (Opus used ~40% fewer tokens).
KEY POINTS
- Akka tested a spec-driven workflow for AI-assisted software porting across 65 open-source projects, using a delivery harness that cycled through discovery, specification, porting, benchmarking, and improvement.
- The initial tranche took 99.3 hours and consumed 9.41 billion tokens; Akka reported lines-of-code or performance improvements in 57 of the 65 ports and used Claude with Akka Specify for implementation, with Sonnet averaging 61 minutes per port versus 120 minutes for Opus (Opus used ~40% fewer tokens).
- This experiment provides empirical data on the cost, speed, and effectiveness trade-offs of spec-driven AI-assisted porting across many projects, informing tool design, model selection, and validation practices.
WHY IT MATTERS
This experiment provides empirical data on the cost, speed, and effectiveness trade-offs of spec-driven AI-assisted porting across many projects, informing tool design, model selection, and validation practices.