From Paper-Based Monitoring to AI-Assisted Alignment Feedback: Design and Development of LP Monitor for Lesson Plan Supervision

Abstract

Lesson plans are intended to guide instruction and provide a basis for professional reflection and instructional supervision, and their instructional value depends on the coherence among the intended learning competency, objectives, learning experiences, and assessment rather than on the mere presence of these components in a document (Biggs, 1996; Martone & Sireci, 2009). However, the processes used to submit, monitor, and review lesson plans may themselves create administrative and instructional difficulties. Paper-based monitoring requires recurrent printing, physical document movement, and manual tracking, and it frequently delays supervisory feedback until after a lesson has already been taught, which reduces much of that feedback's formative value (Hattie & Timperley, 2007; Shute, 2008). High submission volumes further limit the depth of structural analysis a supervisor can perform, often reducing review to compliance checking rather than instructional coaching. This design and development research (DDR) study documents the problem analysis, co-design, and formative tryout of LP Monitor, a web-based lesson-plan supervisory system built to digitize submission and monitoring while providing automated, rubric-based feedback on instructional alignment, consistent with current Philippine guidance on lesson planning (Department of Education, 2026b) and on human-centered use of artificial intelligence in basic education (Department of Education, 2026a). Problem identification was conducted through a focus group discussion held during a joint Learning Action Cell session with 18 teachers from the Languages (English and Filipino) and Science departments at San Agustin National High School, Bohol, Philippines; the session was documented through minutes independently recorded by Master Teachers from the requesting departments, and thematic coding was performed by a single coder. The same 18 teachers, who were not under the author's direct supervisory authority at any stage of the study, took part in a subsequent co-design session that shaped the system's requirements and interface, including the integration of a large language model that evaluates three alignment relationships anchored on an author-developed Scaffolding Objectives construct (Enad, 2026) informed by scaffolding theory (van de Pol et al., 2010) and by recent work applying large language models to constructive-alignment tasks (Pereira et al., 2024): Competency-to-Scaffolding Objective, Scaffolding Objective-to-Learning Experience, and Scaffolding Objective-to-Assessment. During a four-week formative tryout in August 2026, the same cohort submitted 91 lesson plans through the system, of which 50 received an automated recommendation for revision and 41 an approval recommendation; teachers used the automated feedback to revise and resubmit plans at their own discretion. The study is exploratory and descriptive: no control group, pre-registered hypothesis test, or pre-tryout baseline was used, and the automated alignment engine and the underlying Scaffolding Objectives construct have not undergone external psychometric validation. The findings describe LP Monitor's developmental logic and preliminary operational patterns rather than definitive claims of effectiveness, and the paper closes with a research agenda for psychometric validation, controlled comparison, and longitudinal follow-up.