Source trailReferences

    A reflection on my Master’s experience

    As at Apr 2026

    On being a beginner again.

    What a year of reframing taught me about change, personalisation, and innovation.

    Gerald Ajam

    SMU MSc Entrepreneurship & Innovation (MEI)

    This is not a summary of a programme. It is a set of working insights from returning to formal study after a long career in education, and from the productive disorientation that followed.

    Three threads run through it: a personal reflection on becoming a beginner again; three working insights on learning, change and innovation; and what they imply for education and the public sector. The three ideas are offered as tools for thinking about educational change, not as conclusions.

    §00

    ON BEING A BEGINNER AGAIN

    Programme overview

    A crash course in business reality.

    The mix that shaped the thinking.

    Content

    Electives

    Experiential

    The mix gave sustained exposure to venture and commercial logic: how products are built, tested, funded and taken to market, with repeated encounters with experimentation, product thinking and scale.

    Most valuably, it offered a productive contrast with the assumptions and demands of education systems. It made visible what is usually taken for granted in public-sector work.

    Context

    A humanities-trained educator with no business experience.

    I walked into the programme with the mind of a true beginner.

    20

    Years of experience across teaching, school leadership, curriculum design, resource development, edtech.

    2

    Areas of familiarity — Geography and English. Some dabbling in design thinking and learning space design.

    3/29

    Singaporeans in the cohort. Oldest student in the room.

    Reframing, not retraining. Twenty years across teaching, school leadership, curriculum resource design, edtech projects and personalised learning left me with a longstanding concern: how promising ideas translate into actual practice, given public-sector realities, educational aims and system constraints.

    Trained in Geography and English Language, with some exposure to design thinking and innovation, I had little prior appreciation of business, entrepreneurship or venture logic. That made the programme both unfamiliar and revealing.

    Entering as the oldest student in a cohort of 29, and one of only three Singaporeans, the experience was professionally useful precisely because it was personally disorienting. The unfamiliar setting made certain assumptions newly visible. Being a beginner again sharpened not only what I learned, but what I noticed about learning, change and innovation, and about the limits of my own prior mental models.

    Initial thoughts

    What I learnt

    TB;DL — all three turned out to be both true but incomplete.

    01Money is easy to make It is possible to build things that attract attention, users or revenue. The mechanics of value creation are learnable and often more accessible than assumed. But worthwhile change and purposeful work are much harder to achieve than a growing revenue line might suggest.
    02The private sector knows best The private sector often appears faster, sharper and more decisive. Its feedback loops are clearer, its incentives more legible, its tolerance for failure more explicit. But its operating logic cannot simply be borrowed into education or the public sector without significant loss.
    03Innovate like a startup Speed, experimentation and iteration matter. The bias toward action over endless deliberation is useful in institutions prone to inertia. But public-sector innovation also needs continuity, trust and institutional memory, which startup logic does not naturally cultivate.

    Three flippant takeaways, each with some truth in it. What follows are the three more developed ideas that grew out of them, refined through the discomfort of being a student again.

    §01

    IDEA 1 · LESS OF THE TOOL

    Idea 1

    Less of the tool. More of the change in practice.

    Any innovation or change enters a crowded ecology — routines, judgments, workarounds, accountabilities, and fatigue.

    New tools do not arrive in empty spaces. They enter environments already shaped by routines, workarounds, professional judgement, relational trust, accumulated fatigue and the quiet weight of the familiar. This is true in schools, in hospitals and in any organisation where people have found ways to get work done.

    "Is this a good tool?" matters, but it is the wrong first question. Technical quality does not reliably predict adoption or impact. The more penetrating questions: what practice is this trying to change, and why is that practice worth changing? What existing behaviour is it displacing, and how entrenched is that behaviour? Why would people adopt this under real conditions, when attention is scarce? What burden does it remove, and what burden does it introduce?

    A tool succeeds when the new practice becomes easier to trust, easier to fit into work and easier to repeat than the old one. Success is not a feature of the tool alone. It is a feature of the relationship between the tool and the existing ecology. Schools are dense with tacit judgement, time pressure and relational work, so a technically strong tool may still struggle if it feels slower, harder to interpret or less usable under pressure than the informal habit it is meant to replace.

    Myth 1

    If the product is good enough, adoption will follow.

    Adoption does not follow quality alone. It follows behavioural fit.

    Old lens

    Product quality drives uptake.

    New lens

    Replacement cost, trust, workflow fit, and emotional load drive uptake.

    What often defeats a promising initiative is not poor design in principle but strong incumbents in practice: the current routine, familiar and already trusted; the professional confidence to do it differently; a legacy spreadsheet that still works well enough; cognitive overload ("one more thing I cannot think about right now"); and the easier coordination that runs through informal channels and existing relationships.

    Quality in principle is not the same as usefulness in practice. A more sophisticated system may still lose if it is harder to access, harder to interpret or more disruptive to workflow than the workaround it was meant to replace.

    Adoption is not only a design problem. It is a workflow problem, a trust problem and a sensemaking problem. Practitioners must find the new tool legible, fitting and reliable before they will reach for it when the stakes are real. So before asking whether people are using the new system, ask why they are still reaching for the old one. The answer is usually more instructive than any usage metric. What does the user reach for first when the stakes are real and time is short? That is the true competitor: not another product, but a trusted habit.

    Thinking exercise

    Name the real competitor.

    What is this new initiative really competing against?

    Scenario A

    Teachers are not rejecting data. They are choosing the source of information they trust most in the moment, and that source is usually their own professional knowledge, built through direct contact with students. If a system does not clearly improve professional judgement, if it adds interpretation burden rather than reducing it, it may never become part of actual practice, whatever its technical quality.

    What the dashboard is really competing against: professional confidence, the teacher's accumulated judgement about individual students; timeliness, the ability to notice and act in the moment without a login; interpretability, knowing what the data means and what to do next; and emotional assurance, trusting one's own reading of a student's state.

    The design implication is not to replace teacher judgement in the abstract. It is to support judgement in ways that are timely, interpretable and trusted enough to be used when it matters.

    Thinking exercise

    Name the real competitor.

    What is this new initiative really competing against?

    Scenario B

    Organisations may believe they have implemented a solution because the system exists and usage metrics look reasonable. But the real work may still flow elsewhere, through the shadow workflow that nobody formally maintains and few people acknowledge. The result is a damaging false impression of adoption: the official process map shows one thing, while the pathway through which work actually gets done is entirely different.

    Unofficial systems persist because they are familiar and already embedded in daily habit; fast, with responses from trusted colleagues; relationally trusted, so people know exactly who to ask; and free of any learning curve, login or new interface.

    To change practice, we must understand the shadow workflow, not just the formal one. The map and the territory are rarely identical.

    Takeaway 1

    The real product is not the platform. It is the changed routine.

    Painting a clear picture of the future is only part of the task. We also need to show people how to get there — by demonstrating how to shift behaviours, routines, and providing a basis for people to trust us.

    FIG 01 — When stakes are real, the trusted habit wins

    What did people give up? What made the old alternative emotionally safer? What did the new initiative require people to relinquish, in autonomy, familiarity or professional confidence? What trust had to be earned before use under pressure became possible?

    Redesign the work, not just the interface. Strong implementation means lowering switching costs, reducing interpretation burden and making the new routine easier to enact repeatedly. Improving the product's visual design or feature set is not the same job.

    A one-off trial is not adoption. The goal is to help people move from intention to habit: to reach for the new practice when the stakes are real, not only when conditions are ideal. Painting a compelling future is not enough. We also need to design the migration from old habit to new practice, and that migration is usually the hardest part of any change effort.

    §02

    IDEA 2 · LESS CUSTOMISATION

    Idea 2

    Less customisation. More agency.

    Personalisation is not mainly about matching. It is about rebalancing who gets to notice, decide, initiate, and shape.

    Personalisation is most often framed as better matching: the right level, the right pace, the right next task, the right pathway through a curriculum. The implicit promise is that a better-fitted experience will produce better outcomes.

    A working distinction. Customisation: the system adapts to the learner, adjusting content, pace and pathway from performance data. The learner gets a better-fitted experience, but the locus of decision stays with the system. Agency: the learner becomes more able to act meaningfully within learning, to notice what they need, make reasoned choices, reflect on whether those choices worked, and adjust.

    The deeper question is who gains more power to notice, decide, initiate and shape the learning process. Tailoring can improve access and efficiency. Agency changes what kind of learner is being formed, and that is the more enduring educational question. A system may know the learner very well while still leaving the learner passive. Better tailoring does not automatically cultivate stronger self-direction.

    Myth 2

    The more tailored the system, the more personalized the learning.

    A perfectly tailored experience can still leave the learner passive, dependent and overly served.

    Old lens

    Better matching = better personalization

    New lens

    More authorship, judgement and self-direction = deeper personalisation

    Customisation can improve access. Agency changes what kind of learner is being nurtured. Commercial definitions of personalisation might not best serve educational aims.

    A highly personalised system can still decide too much on the learner's behalf, removing the need for judgement; smooth away the productive struggle that builds capability; over-direct the learner, optimising efficient progress without strengthening self-direction; and create dependency, learners who perform well inside the system but cannot navigate without it.

    Efficient completion is not the same as strong learning. Support can become over-support. A system can remove difficulty without developing capability, and a learner can move through tasks quickly while becoming less, not more, able to act independently. The question is not only whether learning is easier or more efficient. It is whether the learner is becoming stronger, more reflective and more capable of directing their own learning, today and in the absence of the system tomorrow.

    Thinking exercise

    What is personalization?

    Which would you rather see more of in your classroom?

    A

    B

    In A, the output has improved. The student's writing has not. In B, the learner is drawn back into substantive thinking: the AI names specific weaknesses in the argument, offers several next moves, and the student chooses one and explains why.

    Three probes for any AI-supported learning activity. Where is the real thinking happening, with the learner or with the system? What learner capability is being strengthened through this interaction? If the AI disappeared tomorrow, what would the student still be able to do?

    Not all AI support is equal. Some forms of support replace thinking. Others scaffold better thinking. The distinction is educationally consequential, and rarely made explicit in procurement or implementation decisions.

    Thinking exercise

    What is effective personalization?

    What is the purpose of personalisation – engagement or empowerment?

    C

    D

    How much agency should learners have? Is choice alone sufficient?

    Scenario C has real value. The system reduces frustration, improves task fit and appears to raise both engagement and completion. But who is driving the path? Are students navigating a learning path, or moving through one authored entirely for them? Is the system optimising support, or optimising compliance? Is this personalisation, or sophisticated allocation? Could the learner function without the system's direction?

    Scenario D shows what dashboards do well: they increase visibility of progress and gaps, improve metacognitive awareness, make it easier to spot where effort is needed and reduce the opacity that can make learning feel arbitrary. What may still be missing is the harder part: interpreting what matters most among the visible data, prioritising what to address first and why, articulating one's own choices, reflecting on whether the chosen action worked, and revising plans on evidence rather than following recommendations. Stronger design would add goal setting, rationale prompts, structured reflection cycles, teacher dialogue about the choices made, and real opportunities to revise a plan rather than follow a curated queue. Seeing one's own data is not the same as making meaningful decisions about one's learning.

    Takeaway 2

    The best personalisation does not serve learners more accurately — it helps them steer better.

    The disparity highlights the challenge of edtech evolving or being adapted from commercial "best practices" — rather than it being purpose built around educational aims, assumptions, and standards.

    Commercial personalisation aims for subservience and reliance.

    FIG 02 — Fit is not the axis that matters most

    Who is making the consequential decision? Not simply what task appears next, but who is exercising the judgement that shapes the learning path. Is the learner being enabled to make a meaningful move, or being served efficiently through one?

    Are we developing judgement or improving task flow? Smooth, efficient task completion is a reasonable goal, but not a sufficient one. The deeper purpose is to cultivate the learner's capacity for independent, reflective, purposeful action.

    The teacher is not made less important by better personalisation technology. The teacher becomes more important, as a designer of conditions for agency, productive struggle, reflection and meaningful challenge. The stronger view of personalisation: the aim is not only to improve fit between system and learner. It is to cultivate more capable learners, people who can notice what they need, choose wisely and learn from the outcomes of their choices. Commercial definitions are too thin for educational purposes. Better matching is useful. Greater agency is the deeper educational goal.

    §03

    IDEA 3 · LESS SCALING FAST

    Idea 3

    Less scaling fast. More compounding of institutional learning.

    Progress is not only what we launch — it is what we learn that changes how we proceed & sharpens our decision making.

    What entrepreneurship gets right: the instinct toward movement, experimentation and iteration. The bias against paralytic deliberation, the willingness to try things before conditions are perfect, and the discipline of testing assumptions early are productive habits that institutions often lack. Movement matters, because inertia is a real and costly force. Experimentation matters, because learning requires trying. Iteration matters, because the first version is rarely the best one.

    What institutions additionally need: learning while moving. Not just momentum, but memory. Not just activity, but improved judgement. The strongest innovation work in institutional contexts does more than produce a pilot or a proof of concept.

    It reduces uncertainty about what actually works and why; sharpens judgement for future decisions; surfaces hidden assumptions before they become expensive commitments; builds reusable capability that survives personnel changes; and improves the quality of the next decision, not just the current one. In education and the public sector, trust is slow to build and easy to lose. Decisions accumulate and compound. Learning must survive turnover, handovers and shifting political priorities.

    Myth 3

    A pilot is successful if it proves we should scale.

    A pilot is successful if it reduces meaningful uncertainty – even if the answer is not yet, not like this, not for this group of users.

    Old lens

    Successful pilot = proof for scaling

    New lens

    Well designed pilot = better decision making for the next judgement call

    Having many shiny, polished pilots (or POC, POV, MVP, MPP, MLP) is not necessarily evidence of a disciplined, innovative organisation.

    A good pilot may legitimately conclude: not yet, conditions are not right for this to work at scale; not like this, the design needs revision before it can travel; not for these users, the population matters more than anticipated; not under these conditions, the context shaped the outcome more than the intervention; not with this trust model, adoption needs relationship-building that was not accounted for.

    A polished showcase is not the same as institutional learning. Positive momentum can hide weak evidence. Pilots should be designed around real learning questions, not implementation activity that generates favourable-looking metrics. The purpose of piloting is not to secure permission to expand. It is to make better strategic decisions, including the decision not to expand, or to expand differently.

    Thinking exercise

    What of these do we want?

    What did these attempts, experiments, pilots accomplish?

    A

    B

    C

    There is no simple choice. It depends on what we want and where in the idea cycle we are.

    A: what was actually learned? What assumptions were tested? What evidence would have caused a pause? B: less impressive on paper, considerably more valuable as institutional learning. C: the pause is not failure. It is the pilot working as intended.

    Mixed results can be more valuable than uniformly positive anecdotes, if they improve the quality of the next decision. The stronger pilot is not always the most impressive-looking one.

    Takeaway 3

    True innovation is disciplined experimentation, with the aim of building institutional memory.

    Its deeper value lies in how well it sharpens judgment, improves the next decision, and leaves the system wiser than before.

    The most valuable innovation work compounds institutional wisdom, it is the hard work of trying, learning and remembering.

    FIG 03 — Visibility and learning are different currencies

    Four questions every initiative should answer. What uncertainty is this project meant to reduce? Name the specific question the work is designed to answer, before beginning. What assumptions are actually being tested? Surface the beliefs that, if wrong, would cause a change of course. What will we learn that changes the next decision? Define in advance what a useful finding looks like, including an unwelcome one. Are we scaling evidence, or scaling hope? Distinguish what the data shows from what we wish it showed.

    Without institutional memory, teams repeat activity but do not compound wisdom. Pilots become episodic rather than cumulative. Innovation stays visible and busy, but not necessarily developmental. Each new initiative starts from roughly the same place as the last.

    Disciplined experimentation requires explicit learning questions articulated before the work begins; careful capture of what actually happened, not just what was planned; honest interpretation of what the findings mean, including ambiguous ones; and translation of findings into changed decisions, revised criteria or reusable capability. Innovation should not be judged mainly by visibility, speed or scale. Its deeper value lies in whether it leaves the system wiser than before, and better placed to act well next time.

    §04

    ENTREPRENEURSHIP ↔ EDUCATION

    Entrepreneurship ↔ Education

    Thinking about change, personalization and innovation.

    Lessons that entrepreneurship might offer to education.

    1

    Not if it works well or looks promising but whether it changes practice.

    2

    Not if it is tailored or engaging but whether it grows agency.

    3

    Not if it scales or moves quickly but whether it makes the next decision wiser.

    1From tools to changed practice The key question is not whether something is promising in principle. It is whether it changes what people actually do, under real conditions, in real time, when the stakes are genuine. Judge innovation not by whether it looks promising or generates enthusiasm, but by whether it changes what people do when attention is scarce.
    2From customisation to agency The key question is not whether learning is more tailored or efficiently matched. It is whether the learner becomes more capable of acting with ownership, direction and informed self-awareness. Judge personalisation by whether it helps form more agentic learners, people more capable of directing, reflecting on and owning their own learning.
    3From speed and scale to institutional learning The key question is not whether something launched quickly or reached large numbers. It is whether it made the next decision wiser and left the institution more capable of judgement. Judge innovation by whether it sharpens the next decision and leaves the institution with better judgement than it had before.

    All three ideas push in the same direction: away from what looks impressive on the surface, toward the deeper conditions that make change worthwhile and durable.

    The through-line is formation: what practices are formed, what kinds of learners are formed, and what kinds of institutional judgement are cultivated over time.

    Final thoughts

    What I actually learnt

    The real challenge is not to become more startup-like. It is to become more purposeful in the change we pursue, more educational in what we value, and more disciplined in how we learn from what we do.

    The initial slogans are attractive because they carry some truth. But the truth is always further away from the slogan, and more demanding to act on.

    Thank you for your attention & I’m always happy to chat…

    Questions, disagreements and counter-examples all welcome.

    SMU MSc ENTREPRENEURSHIP & INNOVATION (MEI)

    Source trail

    References

    Further readings from the handouts, on practice, implementation and institutional learning, and on agency, personalised learning and AI in education. APA 7th edition; links point to the DOI or publisher page where one was verified.

    1. Biesta, G. J. J. (2017). The rediscovery of teaching. Routledge.
    2. Bryk, A. S., Gomez, L. M., Grunow, A., & LeMahieu, P. G. (2015). Learning to improve: How America's schools can get better at getting better. Harvard Education Press.
    3. Coburn, C. E. (2001). Collective sensemaking about reading: How teachers mediate reading policy in their professional communities. Educational Evaluation and Policy Analysis, 23(2), 145–170. https://doi.org/10.3102/01623737023002145
    4. Feldman, M. S., & Pentland, B. T. (2003). Reconceptualizing organizational routines as a source of flexibility and change. Administrative Science Quarterly, 48(1), 94–118. https://doi.org/10.2307/3556620
    5. Fixsen, D. L., Naoom, S. F., Blase, K. A., Friedman, R. M., & Wallace, F. (2005). Implementation research: A synthesis of the literature. National Implementation Research Network.
    6. Garvin, D. A., Edmondson, A. C., & Gino, F. (2008). Is yours a learning organization? Harvard Business Review. https://hbr.org/2008/03/is-yours-a-learning-organization
    7. Kapur, M. (2015). Learning from productive failure. Learning: Research and Practice, 1(1), 51–65. https://doi.org/10.1080/23735082.2015.1002195
    8. McGrath, R. G., & MacMillan, I. C. (1995). Discovery-driven planning. Harvard Business Review. https://hbr.org/1995/07/discovery-driven-planning
    9. OECD. (2019). Student agency for 2030 (OECD Future of Education and Skills 2030 concept note). OECD Publishing. https://www.oecd.org/content/dam/oecd/en/about/projects/edu/education-2040/concept-notes/Student_Agency_for_2030_concept_note.pdf
    10. OECD. (2020). The public sector innovation lifecycle. OECD Publishing. https://www.oecd.org/en/publications/the-public-sector-innovation-lifecycle_0d1bf7e7-en.html
    11. Pisano, G. P. (2019). The hard truth about innovative cultures. Harvard Business Review. https://hbr.org/2019/01/the-hard-truth-about-innovative-cultures
    12. Ryan, R. M., & Deci, E. L. (2017). Self-determination theory: Basic psychological needs in motivation, development, and wellness. Guilford Press.
    13. Selwyn, N. (2019). Should robots replace teachers? AI and the future of education. Polity.
    14. Thomke, S. H. (2003). Experimentation matters: Unlocking the potential of new technologies for innovation. Harvard Business School Press.
    15. Tyre, M. J., & Orlikowski, W. J. (1994). Windows of opportunity: Temporal patterns of technological adaptation in organizations. Organization Science, 5(1), 98–118. https://doi.org/10.1287/orsc.5.1.98
    16. Watters, A. (2021). Teaching machines: The history of personalized learning. MIT Press.
    17. Zhao, Y. (2012). World class learners: Educating creative and entrepreneurial students. Corwin.