LAIVEcase - Multiagentic AI orchestrated Simulation & Evalutation

The Multi-Agent Conversational Simulations didactic format represents a cutting-edge modality for active practice of skills within learning environments. Examples of application of these methods are two projects, one for Sales simulations (Saltech client) and another about Soft skills simulations developed for the client Bottega Veneta. In the LAIVEcase software, the goal was to enable instructional designers to build case-based didactic simulations in a simple, no-code environment, shifting the paradigm from formal training toward situated and experiential learning, thereby fostering durable and effective knowledge acquisition. Shifting the focus means offering realistic practice scenarios in which learners can actively experiment strategies, actions, and decisions similar to the complexity of real-world professional contexts and shaping them into behavioral responses that activate knowledge and competencies required to perform correct actions in the story. This approach draws on Situated Learning theory (Lave & Wenger), which enlights that knowledge cannot be separated from the context in which it is generated and applied. This imply that learning means participating in authentic practices, not merely absorbing abstract concepts. More recent research on Agency-Based Learning (Matusov et al., 2016) further confirms that repeated, spaced exposure to realistic operational scenarios, combined with structured feedback, produces significantly greater competency transfer compared to content-driven instructional methods. Within this framework, the multi-agent AI simulation has been conceived as a fully-fledged instructional format: SCORM-compliant, integrable into blended learning pathways, and supported by dedicated authoring environments designed specifically for instructional designers. A distinctive feature of the simulation is its ability to run in two distinct ways, adaptable to the learner's position within the learning journey or to specific instructional objectives. The simulation can be experienced as: 1) a guided learning mode, supported by a Mentor who operates through a continuous, progressive KPI-based assessment system to facilitate explicitly scaffolded interactions during the activity, so learners can monitor the impact of their actions on the narrative in real time and draw on the mentor's explicit guidance; or 2) an assessment mode, in which an analytical performance report is available at the end of the activity, to enable a full re-reading of the enacted behaviour to interpret weaknesses and best practices. At the individual level, assessment and the provision of continuous, structured feedback enable participants to develop metacognitive awareness: learners come to understand the cause-and-effect relationship between action and consequence, gain perspective on others' viewpoints, and are able to evaluate alternative strategies. This is the core mechanism of the reflective cycle theorized by Schön through the concepts of reflection-in-action and reflection-on-action. At the organizational level, the systematic collection of assessment data makes it possible to measure the gap between expected and demonstrated competencies, identify recurring behavioral patterns across participant cohorts, and progressively build an evidence base to inform talent development decisions and training priorities.