Persona-as-Configuration: Generative Stakeholder Reporting for Agricultural Floods
2026-07-20 • Software Engineering
Software EngineeringArtificial IntelligenceHuman-Computer Interaction
AI summaryⓘ
The authors looked at systems that use sensors and rules to detect flooding in farm fields, which create clear decision records. They wanted to add large language models (LLMs) to make reports for different types of users, but faced a challenge because LLMs work unpredictably while the system needs to stay consistent and traceable. To solve this, they designed a pattern where the LLM only reads existing data and does not change it, and user-specific reports are made using fixed templates instead of on-the-fly changes. They tested this idea on a flood detection system and found experts liked how it keeps the parts separate and reliable, with plans to test it with farmers soon.
cyber-physical systemsdeterministic edge inferencelarge language modelsdecision logsunidirectional consumptionpersona-as-configurationJSONstanding-water detectionISO/IEC 25010dashboard
Authors
Oliver Aleksander Larsen, Tiziano Santilli, Francesco Daghero, Mahyar T. Moghaddam
Abstract
Cyber-physical systems built on deterministic edge inference, such as on-vehicle flood detection for agricultural fields, produce structured decision logs that must be interpreted differently by heterogeneous stakeholders. Pairing such systems with large language models (LLMs) to generate stakeholder-specific reports introduces a tension: the generative layer is non-deterministic, while the edge plane must remain replayable and auditable. We propose an architectural pattern resting on two invariants: unidirectional consumption, in which the generative layer is a strict read-only consumer of the deterministic plane and never writes back, and persona-as-configuration, in which stakeholder adaptation is a versioned prompt-template artifact rather than runtime improvisation. We instantiate the pattern as a context-aware dashboard layer over the JSON decision logs of a previously published edge-based standing-water detection system, and analyse how the integration boundary admits standard generative-reliability mitigations as configuration- or middleware-level extension points. A structured expert review rated the pattern favourably across five ISO/IEC 25010-aligned quality dimensions, with strongest agreement on separation of concerns. End-user evaluation with agricultural stakeholders is planned for future work.