[DigitalToday reporter Chi-gyu Hwang (황치규)] Amazon Web Services (AWS) has added parallel task execution, a requirements analysis engine and a Quick Plan feature that supports rapid planning to its AI software development tool Kiro, SiliconANGLE reported on May 12 local time.

Kiro has so far processed even independent tasks sequentially, one at a time. With Parallel Task Execution support, it now analyses each task's dependency graph and runs tasks that do not share state, endpoints or files at the same time. AWS explained that this can cut the processing time for large development specifications from more than 1 hour to 15 minutes.

The Requirements Analysis Engine finds logical conflicts in requirements before code is written. A large language model (LLM) rewrites ambiguous requirements into verifiable criteria and converts them into formal logic, then checks for contradictions with an SMT (Satisfiability Modulo Theories) solver, a mathematical reasoning engine. For example, if page 1 contains a rule for complete deletion and page 10 contains a rule that assumes soft deletion, it detects this as a mathematical impossibility and informs developers in language that is easy to understand.

The Quick Plan feature is a fast-track approach for projects where scope and conditions are already clear. It asks all necessary questions upfront instead of step-by-step approvals and generates the full stack at once.