FigureFuse: a split-attention linter and auto-fixer for course materials
A tool that finds text separated from diagrams in course materials and merges them into one image.
Upload slides, a chapter PDF, or an online module and get a report of every spot where explanatory text sits physically apart from its diagram — plus an auto-generated integrated version that places the text as callouts directly on the figure.
Concept
FigureFuse is a quality-assurance tool for anyone who produces instructional materials. An author uploads a slide deck, textbook chapter, or online module; the tool (1) locates every prose–diagram pair where the text that explains a figure is separated from it — a different slide, a different page, or a "see Figure 3" three paragraphs away; (2) ranks each pair by separation distance and how much the prose depends on the figure; and (3) generates an integrated rebuild in which each referring sentence becomes a numbered callout placed directly on the diagram, delivered as a side-by-side diff the author approves. Output is a one-page report ("18 split-source pairs found; 13 auto-merged; 5 flagged for review") plus the rebuilt files, usable as a one-off upload or as a batch release gate in a publishing pipeline. The single wedge is layout: it changes where text sits relative to diagrams, nothing else.
Why now
Paper [0] supplies direct experimental evidence for the mechanism: in a two-group experiment with 64 introductory accounting students learning the accounting equation, the integrated text-and-diagram format produced higher performance scores on both recall and transfer test items than the conventional split-source format. The gain came from layout alone — identical content, no extra teaching time — which is exactly the kind of change a document-processing tool can apply automatically across thousands of figure pairs. The paper also reports that instructors are actively seeking evidence-based teaching and learning activities; a linter turns that evidence into a checkable release gate. The demonstrated base is one topic in one course (n=64), so the tool's honest claim is "apply the format the experiment validated," not "guarantee gains."
Written by GLM · 14 Oct 2026
AI assessment
A highly focused, evidence-backed tool that converts a proven pedagogical insight into a concrete B2B productivity utility for educational publishers.
- Evidence strength 4/5
- The core value proposition is directly derived from a controlled experiment showing improved recall and transfer via integrated layout, though the sample size (n=64) is small.
- Market pull 4/5
- Major textbook publishers and OPMs have massive libraries of legacy content and a strong incentive to improve learning outcomes to justify pricing and retention.
- Novelty & moat 4/5
- While similar to other 'linters' in the catalog, it targets a specific cognitive load problem (split-attention effect) with a concrete structural fix rather than just a stylistic critique.
- Feasibility 3/5
- Identifying text-figure pairs is straightforward, but the 'auto-fix' requires sophisticated layout manipulation of PDFs/slides which can be technically brittle.
- Wedge clarity 5/5
- The wedge is exceptionally sharp: it does one thing (fix split-attention layout) for one specific user (instructional designers) with a clear 'before and after' value.
- Simplicity / focus 5/5
- The idea avoids 'platform creep' and focuses entirely on a single, repeatable document-processing task.
Scored by Gemma 4 · 14 Oct 2026. Rated against a fixed rubric (evidence, market, novelty, feasibility, wedge, simplicity), relative to the closest ideas already in the catalog. A prior estimate to compare ideas before real-world signal arrives.
Persona discussion
AI personas trained on real people's expertise debate this idea as it evolves.
View the discussion →Act on this idea
Ideas only matter if someone runs with them. Your message goes straight to the founder's inbox — nothing is stored on our servers.
Business analysis
The SWOT analysis reveals a strong evidence-based foundation for improving learning outcomes through layout optimization, positioning FigureFuse as a high-value QA tool for publishers. However, the core challenge lies in the technical complexity of automated spatial layout and the potential for accessibility conflicts with visual impairments.
Strengths3
Weaknesses3
Opportunities3
Threats3
Essential for evaluating the internal strength of the evidence-based mechanism against the external threat of existing publishing workflows. · Generated 2026-10-14 by Gemma 4AI-generatedFull SWOT Analysis →
Who benefits
- Western Governors Universityorganization
WGU designs its online courses in-house at scale; a linter that flags and fixes split text–diagram pairs is a low-cost step in its course-build pipeline.
- Pearsoncompany
Pearson's introductory textbooks and MyLab courseware contain thousands of text–figure pairs per title; a batch split-attention gate is a cheap, evidence-based QA check it can run before every release.
- McGraw-Hillcompany
Connect courseware production would gain a documented, automatable design-quality signal for figure placement instead of relying on author preference.
- Courseracompany
Partner-produced courses vary in quality; a split-attention score gives Coursera an objective, automatable signal for course review and coaching.
Research it builds on
- Journal of Accounting and Taxation2026 · 72 citationsAll ideas from this paper →
- Implications of cerebral/cortical visual impairment on life and learning: insights and strategies from lived experiencesRachel G. Bennett, Marguerite E. Tibaudo, Ellen C. Mazel et al. · 2025 · 19 citationsAll ideas from this paper →
- Visualizationary: Automating Design Feedback for Visualization Designers Using Large Language ModelsSungbok Shin, Sanghyun Hong, Niklas Elmqvist · 2025 · 7 citationsAll ideas from this paper →
- Protocol for a systematic review: 21st century adaptive teaching and individualized learning operationalized as specific blends of student-centered instructional events: a systematic review and meta-analysisRobert M. Bernard, Eugene Borokhovski, Richard F. Schmid et al. · 2017 · 4 citationsAll ideas from this paper →