From Ambiguity to AI-Ready Requirements: A Business Analysis Framework for Structuring Business Problems for Intelligent Process Automation

Abstract

This paper argues that intelligent process automation fails at specification more often than at implementation. Requirements engineering offers a mature account of ambiguity as a defect to be removed and a mature account of the quality characteristics of a well-formed requirement, but neither transfers cleanly to processes whose automation depends on judgement and on probabilistic components. This paper develops the Ambiguity-to-Specification (A2S) pipeline, a five-gate business analysis framework; problem framing, decision decomposition, data grounding, behavioural specification, and oversight and acceptance, that converts an ambiguously stated business problem into a specification an intelligent automation can be built and governed against. The framework rests on three arguments: that the unit of specification shifts from system behaviour to the decision; that ambiguity is not uniformly a defect but a resource whose resolution should be routed to the gate at which it is cheapest and most informative, leaving a residue that must be specified rather than resolved; and that probabilistic components break the correctness contract on which classical quality criteria rest. A second contribution extends the requirement quality characteristics of ISO/IEC/IEEE 29148 with six AI-readiness characteristics; decidability, data-groundedness, error-tolerance specification, escalation specification, observability and drift sensitivity, each with a test question, an evidencing artefact and a corresponding requirement smell. The framework is illustrated on a hypothetical mid-sized services organisation performing document-heavy casework, and eight propositions are stated for empirical evaluation. The paper is conceptual: no primary data were collected and the framework has not been validated.