Students felt engaged in real research, not just a chemistry class.
In a large undergraduate chemistry course at Tecnológico de Monterrey, a researcher named Saeed Beigigiboroujeni confronted a question students have always carried silently: why does any of this matter? His response was not to simplify the curriculum but to deepen it — breaking fragments of his own live research on CO₂-based polymers into structured cases that students could genuinely investigate, then pairing that authenticity with AI tutoring assistants capable of providing the personalized guidance that no single instructor can offer at scale. The experiment points toward something larger: that the gap between memorization and understanding in science education may be bridged not by better lectures, but by inviting students into the actual work of science itself.
- A student's blunt question — 'how does this help me?' — exposed a fault line running through organic chemistry education: students were memorizing reactions without grasping why those reactions exist or what they make possible in the world.
- The tension is structural: the pedagogical strategies that produce real understanding demand individualized attention, yet university science courses routinely enroll dozens of students with one instructor and a fixed clock.
- Beigigiboroujeni's answer was to disassemble his own laboratory research into three-stage 'mini-cases' — orientation, design, and analysis — so that undergraduates could engage with authentic scientific problems, real spectroscopic data, and genuine mechanistic reasoning rather than invented exercises.
- Students responded with observable curiosity: they asked questions beyond the syllabus, built mechanistic explanations, and stopped treating the material as something to survive and started treating it as something to understand.
- To prevent success from collapsing under its own weight, the team is deploying AI assistants through Tec's Agent Studio platform — a Socratic reaction tutor, a spectroscopy guide, and a metacognitive coach — designed to extend the instructor's reach without replacing human judgment.
- The trajectory suggests a replicable model: authentic research fragments plus scalable AI mentorship could reshape how STEM subjects are taught in large institutional settings, making personalized scientific inquiry a structural feature rather than a rare privilege.
Fifty minutes into a lecture on epoxy ring-opening reactions, a student raised a hand and asked the question that undoes every chemistry professor eventually: 'How does this help me?' For Saeed Beigigiboroujeni, a researcher at Tecnológico de Monterrey who develops sustainable polymers from carbon dioxide, the question landed differently than it might have for others. He knew the material was genuinely fascinating. He also knew his students were memorizing it without understanding why it existed.
Organic chemistry is not an abstraction — it is the foundation of biomedical materials, sustainable coatings, drug delivery systems, and the plastics in everyday objects. Yet classrooms reduce it to formulas students learn to reproduce without grasping the reasoning behind them. The gap between what chemists actually do and what students are asked to memorize had become impossible to ignore.
Beigigiboroujeni's response was to look at his own laboratory work and ask whether it could be broken into pieces students could solve. His research on light-curing CO₂-based materials involves carbon hybridization, nucleophilicity, resonance, and reaction mechanisms — the very concepts his students were struggling to retain. He adapted fragments of that real research into what he calls mini-research cases: not simulations, but authentic scientific problems pedagogically restructured for undergraduates.
Each case moves through three stages. In the first, students face a genuine scientific challenge — designing a chemical route to convert CO₂ into a sustainable polymer monomer. In the second, they propose their own reaction sequences, choosing reagents, considering green chemistry principles, and outlining purification strategies. In the third, they analyze real spectroscopic data from Beigigiboroujeni's lab, connecting molecular structure to material performance exactly as a working researcher would.
The shift in student behavior was immediate. Engagement became genuine rather than obligatory. Explanations grew mechanistic and detailed. Students were no longer memorizing — they were reasoning.
But success created a new problem: scale. Multiple teams, each pursuing different routes and interpreting different data sets, each needing specific feedback — this exceeds what any single instructor can provide in a large course. The best pedagogical approaches have always demanded personalization that classrooms structurally resist.
The answer came through Agent Studio, a platform within Tec's TECgpt ecosystem that lets teachers build specialized AI assistants without programming expertise. Three assistants are being developed for future iterations of the course: one that guides reaction design through Socratic questioning, one that helps students interpret spectra and connect structural changes to polymer behavior, and one that supports metacognition by helping students identify misconceptions and situate their learning within larger questions about CO₂ valorization.
The goal is not to replace the instructor but to extend what one instructor can do across an entire class simultaneously. Together, authentic mini-research cases and scalable AI mentorship suggest a path toward science education that treats understanding — not information transfer — as the actual objective.
Fifty minutes into explaining how epoxy rings open—curved arrows, electrons dancing between atoms, molecular structures transforming—a student raised a hand with a question that stopped the lecture cold: "Professor, how does this help me?" It was the kind of question that sounds simple until you realize you cannot answer it, not really, not in a way that connects what you are teaching to anything the student will ever need to do. The professor asking himself this question was Saeed Beigigiboroujeni, a researcher at Tecnológico de Monterrey who spends his days developing sustainable polymers from carbon dioxide. He knew the material was fascinating. He also knew his students were memorizing it without understanding why any of it mattered.
Organic chemistry is not an abstraction. It is the foundation of biomedical materials, sustainable polymers, advanced coatings—the chemistry that makes cell phone plastic possible, that coats camera lenses, that delivers medications into the bloodstream. Every day, students use products that exist because of organic chemistry. Yet in the classroom, they learn that "CO₂ reacts with epoxides" without grasping why the reaction happens or what it could do. The gap between what chemists actually do and what students are taught to memorize had become impossible to ignore.
Beigigiboroujeni began asking himself a different question: What if he could bring the experience of real research into the classroom without a million-dollar laboratory budget, without specialized equipment, without putting students at risk? The answer came from examining his own work. In his lab, he develops light-curing materials from CO₂ as a renewable carbon source. The process involves multiple steps rooted in fundamental organic chemistry concepts—carbon hybridization, nucleophilicity, resonance, functional groups, reaction mechanisms. What if he broke this real-world project into small cases that students could solve as teams? That question gave birth to what he calls mini-research: not simulations or invented exercises, but authentic fragments of actual research, pedagogically adapted so undergraduates could engage with them, supported by intelligent assistants that could provide personalized attention.
Each mini-research case unfolds in three stages that mirror the scientific method. First comes orientation: students are posed a simplified but genuine scientific challenge. In the first case, they are asked to design a chemical route to convert CO₂ into a monomer that could be used to make sustainable polymers. They work in teams, researching, drawing structures, asking questions, exploring reactants and mechanisms. The second stage is design—the most creative part. Students propose their own sequence of reactions using real transformations: ring opening, carbonation, urethane formation, esterification. They must choose reagents and catalysts, outline purification strategies, consider green chemistry principles. The third stage is analysis. Here is where authenticity becomes crucial: instead of analyzing made-up data, students work with real spectroscopic results from Beigigiboroujeni's lab. They identify how functional groups react with one another, examine thermal analysis to understand intermolecular interactions like hydrogen bonds, study photopolymerization data to evaluate their hypotheses. They compare materials, explain trends, relate structure to performance—exactly as a researcher would.
The results were observable almost immediately. Students showed genuine engagement, not the dutiful compliance of a required course. They asked follow-up questions beyond what the syllabus demanded. Working with real CO₂-based reactions, they developed deeper understanding of hybridization, resonance, nucleophilic and electrophilic behavior, spectroscopy, polymerization mechanisms. Their explanations became mechanistic and detailed. They were not memorizing anymore; they were understanding.
But a problem emerged as the approach proved successful: how to scale it. Large groups of students, divided into multiple teams each designing different routes and interpreting different data sets, each needing specific feedback—this requires individualized attention that a single instructor cannot provide, especially in massive courses. This is the classic problem of education: the best pedagogical strategies demand personalization, but classrooms contain dozens of students with limited time.
The solution came through Agent Studio, a platform within Tecnológico de Monterrey's TECgpt ecosystem that allows teachers to create specialized intelligent assistants without advanced programming knowledge. The platform functions as an orchestration layer on top of generative models, integrated directly into the institution's educational workflows. Assistants are trained with specific course materials, documents, and resources so they can answer contextualized questions and guide student reasoning in personalized ways. Three specialized assistants are being designed for future iterations of the course: a Reaction Design Tutor that guides thinking with Socratic questions rather than direct answers; a Data Interpretation Assistant that helps students read spectra and connect structural changes to polymer performance; and a Scientific Reflection Guide that strengthens metacognition by helping students plan activities, identify misconceptions, and connect learning to larger questions about CO₂ valorization.
The intention is not to replace the teacher but to amplify the teacher's capacity to provide personalized attention at scale. The assistants offer on-demand support, interactive visualization, guided reasoning, personalized feedback, and structured reflection. When combined with authentic mini-research cases, this approach suggests a path forward for teaching science in large classes while maintaining rigor and addressing the persistent challenge of individual attention. The future of science education, Beigigiboroujeni argues, lies not in transmitting more information but in creating experiences where students conduct real science. This classroom implementation shows it is possible to begin that work today.
Citas Notables
How does this help me?— Student question that prompted the pedagogical redesign
In the time of AI, the important thing is not the final product but the thought process, the cognitive process.— Dr. Saeed Beigigiboroujeni
La Conversación del Hearth Otra perspectiva de la historia
What made you realize that fifty minutes of mechanism explanation wasn't working?
A student asked me how it helped them. I couldn't answer honestly. I was teaching the beauty of the chemistry, but not its purpose.
So you took your own research and broke it into pieces for students. Why does that work when traditional teaching doesn't?
Because students are not passive anymore. They are solving a real problem—how to turn CO₂ into something useful. The data they analyze came from my lab last week, not from a textbook written in 1995.
But doesn't that create more work for you, not less?
Yes, absolutely. The design phase is demanding. But I stopped evaluating whether students got the right answer and started evaluating how they think. That changes everything.
How do you handle a class of two hundred students all designing different reaction routes?
That's where the intelligent assistants come in. I cannot give each student personalized feedback. An AI tutor trained on my course materials can guide their thinking with questions, help them interpret spectra, push them to reflect on their own reasoning.
Isn't there a risk that the AI becomes a crutch, that students stop thinking?
Only if we design it that way. The assistants are trained to ask Socratic questions, not provide answers. They amplify what a good teacher does, but at scale.
What changes in a student when they work with real data instead of textbook problems?
They ask different questions. They notice anomalies. They want to know why the data looks the way it does. That curiosity is what science actually is.