MEMO · TO readers evaluating a workshop or a speaker · RE published writing from the practice
Insights
Writing from the people behind Praxora Lab
Every piece here is published on this site, in full, with the framework or argument intact rather than a teaser pointing elsewhere. Where a piece first appeared on a speaker's own site, that original is cited as a reference at the foot of the article, not as the only place to read it.
01 / 05 Strategy
The leadership-level decisions that determine whether AI changes how an organisation works, or just adds a tool to how it already works.
Most organisations already use AI somewhere. Independent research keeps landing on the same shape: adoption is nearly universal, and financial return from it is rare. The gap is not a technology gap. Close to seventy percent of the value AI produces comes from redesigning how a company works, its workflows, roles and decision rights, not from the model underneath it. That redesign is a leadership decision, not a procurement decision, and it cannot be delegated to whichever team happens to be piloting a tool.
In the modern AI era, strategy work is what separates an organisation running dozens of disconnected pilots from one compounding a real advantage. It sets the outcome a deployment is meant to produce before a vendor is chosen, decides which functions get redesigned rather than merely augmented, and puts a board or executive committee on the hook for the result. Every workshop and framework in this section is built for that layer of the decision, not the tool layer beneath it.
-
The productivity gap: why companies must fix their systems before scaling AI
Most organisations use AI frequently but few have scaled it company-wide, and the gap comes down to architectural debt, not appetite.
-
Is your business ready for AI? Start with these five questions
A diagnostic framework for identifying whether a business is actually positioned to operationalise AI tools, not just experiment with them. The same framework the Executive Programme is built on.
-
Why large enterprises are deploying AI at scale but measuring it incorrectly
Common measurement frameworks target the wrong layer of the technology stack, missing the business outcomes that actually matter.
-
How leaders can actually use AI without losing what makes them leaders
One leader pastes a strategy question into a chatbot and goes with the answer unchecked. Another refuses to touch AI at all and delegates it to IT. Both are making the same mistake about what AI-augmented leadership actually requires.
02 / 05 Capacity Building
Turning a leadership decision into something a workforce can actually execute, day to day, inside their own workflow.
A strategy decided at the top does not execute itself. Close to half of employees say a lack of training is the reason AI adoption stalls in their organisation, and close to half say they have received little or none. Generic AI literacy courses do not close that gap, because the skill that matters is knowing how to use a model correctly inside one specific workflow, with the judgment to catch it when it is wrong, rather than knowing what it can do in the abstract.
In the modern AI era, capacity building is the discipline that makes a strategy real at the point of execution: role-based, workflow-embedded training instead of generic tool tutorials, with clear checkpoints for what a team can now do that it could not do before. Skipped or rushed, it produces exactly what independent research keeps finding, premium licenses deployed with no measurable change in output. Done properly, it is the difference between a strategy that stays a slide and one that shows up in the numbers.
-
Implementing AI feels like walking a tightrope, and you're not the only one wobbling
Four independent studies put AI project failure somewhere between thirty and ninety-five percent depending on how you count it. The wobble a team feels on its first real pilot is the statistically normal experience, not a sign the project is failing, and there's a specific set of moves that gets a team through it faster.
-
Everyone's buying governance training. I keep wondering who's going to do the work.
Governance maturity is sprinting ahead of workforce capability, and the gap between them is being mistaken for progress. A case for funding the capability half before the paperwork provides false confidence.
-
It was never my curriculum doing the heavy lifting
What actually predicts whether AI training survives past week one: not the curriculum, but whether a colleague is on hand to answer the small, specific questions that come up in week three.
-
An awkward confession about one-day workshops
The most commercially convenient training format is also the one least likely to change behaviour on its own. A case for treating the one-day workshop as a diagnostic, not the intervention.
-
The time a student caught me trusting a boxplot
The moment a tool flags something, most people stop questioning it. A teaching story about why a flag is a candidate for judgement, not a verdict, for boxplots and for AI systems alike.
-
An AI plan and staff training are not the same thing
Why leadership-level AI strategy and workforce-level AI training are two separate investments that have to run at the same time, not one after the other. The argument behind Praxora Lab's two approaches.
-
A thousand for-loop sermons later, I finally noticed what I was actually teaching
A classroom pattern about resisting vectorized code turns out to predict, almost exactly, how the same learners resist delegating work to AI, and where the caution should actually be aimed.
-
Working with AI agents: what it does to your people, and how to get it right
Agents are now assigning work as often as they receive it. The teams getting real value from that shift aren't the ones with the newest agent, they're the ones who designed the collaboration around it deliberately.
-
A structured approach to AI adoption for Singapore SMEs
For smaller organisations, AI adoption outcomes depend more on discipline than on which tool gets chosen.
-
"Looks good to me": three words I'm still trying to unlearn
As more of the workforce shifts from generating AI output to judging it, the missing skill is writing down what success means before you've seen the answer.
-
Half my AIGP class didn't want to be there. That taught me more than the syllabus did.
AI governance certification is having a moment across Singapore and Malaysia. A frontline account of why it sticks for learners with a live case, and slides off everyone else.
-
The first time someone told me about shadow AI, I reported it. I wouldn't now.
Why treating unsanctioned staff AI use as a discipline problem teaches people to stop disclosing it, and what changes when it's read instead as a training needs analysis delivered for free.
-
Why change management for AI adoption keeps failing, and what actually works instead
Licences activated, training completed, dashboards green, and nothing about how people actually work has changed. The fix isn't more training. It's admitting AI adoption doesn't behave like the change your change management was built for.
-
I taught agentic AI in the wrong order, and I suspect I'm not the only one
A candid account of teaching multi-agent orchestration to a room that hadn't yet learned to trust one AI with one task, and the single-agent exercise that fixed it.
-
Everyone in my class can prompt now. Checking the answer is another matter.
Prompting has become easy enough that checking the output quietly became optional. A field note on why capacity building now needs to train the habit of catching a plausible-but-wrong answer, not more prompt literacy.
-
Change management ROI: what skipping the people side actually costs
There's no universal dollar figure for change management ROI. What's consistent across the evidence is the gap between organisations that fund the human side of an AI rollout properly, and the ones paying for the shortfall in turnover, absenteeism and burnout instead.
-
Checkbox adoption: what happens when you mandate AI without managing the change
Seventy-four percent of companies have yet to show tangible value from their AI investments. BCG's own diagnosis is that the gap is people and process, not technology, and most mandatory rollouts fund that ratio backwards.
-
Psychological safety: the secret engine of AI adoption
Training's done, licences are active, and adoption is stalling anyway. The reason usually isn't the technology or the budget. It's that employees don't feel safe enough to actually use what they were given.
03 / 05 Human Capital Development
What AI changes about judgment, decision quality and career capability inside a workforce, once the initial training is over.
Capacity building gets a team to competent use of a tool. Human capital development is the longer question underneath it: what happens to judgment, decision quality and career growth once a workforce works alongside AI every day, not just in the first month. An employee who defers to AI output without evaluating it has been automated around, not augmented. An organisation that cannot say whether its people are making better decisions with AI, or simply approving AI decisions faster, has no real answer to the question its board will eventually ask.
In the modern AI era, this is the discipline behind measuring AI's return on the people who use it, not only the cost it replaces. Return on Employee, the framework several Praxora Lab practitioners apply, treats hours reclaimed, decision quality and cognitive bandwidth returned to higher-value work as the real yield of an AI deployment, set against a workforce built for cost-reduction business cases. It is the category most often skipped in an AI rollout, and the one whose absence shows up first in employee sentiment, then in turnover, then in the board deck nobody wants to present.
-
92 million jobs will disappear by 2030. Arguing about it won't save yours.
The argument about AI and jobs has settled into two camps repeating the same two numbers at each other. Both are correct at once, and neither tells you what happens to your own role.
-
Why experience alone no longer guarantees higher pay
The old deal was simple: put in the years, make fewer mistakes, get paid more. AI now runs that same pattern-matching in seconds, for a fraction of the cost, and what replaces the old deal is not the technical course everyone is rushing to take.
-
Accelerating the AI upskilling pathway: a strategic framework for early-career and practising engineers
Formal curricula move on approval cycles measured in years while frontier AI moves weekly, so a qualification alone now means arriving on the job with skills already out of date. Closing that gap is a self-directed responsibility, for graduates and practising engineers alike.
-
The entry-level job market broke. Here's the opening it left behind
Recent graduates are unemployed at 5.7 percent against roughly 4.2 for the workforce overall, and the loss is real: AI has absorbed much of the ticket-queue work that used to train junior judgement. But the same shift has opened an unusually early window for graduates willing to build rather than apply.
-
The four dashboards every Chief AI Officer must operate
As the role shifts from championing AI capability to demonstrating ROI and resilience, these are the four things a CAIO needs to see, including whether a workforce is genuinely augmented or just quietly deferring to the machine.
-
Why Gen Z is sabotaging AI rollouts, and what it tells you about every generation
Forty-four percent of Gen Z workers admit to actively undermining their company's AI rollout. The generational framing obscures the real driver: career risk, and whether the rollout feels like it's being done to them rather than with them.
-
The human advantage in the age of AI
AI can replicate what a person knows faster than they can acquire it. What it cannot replicate, judgement, trust, attitude and the capacity to care, is becoming the primary source of competitive advantage for professionals and the organisations that employ them.
-
AI is not coming for jobs. It is coming for tasks. How senior leaders can exploit that shift
White-collar displacement is accelerating at every seniority level, including the C-suite. The professionals staying indispensable are repositioning as the human layer that governs, interprets and directs AI-augmented work, not competing with it on execution speed.
-
The fear of AI replacing jobs is real, it's just not the fear you think
Ninety-two million jobs displaced globally by 2030. Nearly 55,000 US workers told AI took theirs in a single year. But the fear driving Silent Resistance in your workforce goes beyond unemployment. It is about whether the expertise someone spent twenty years building still counts for anything.
-
The silent cost of AI at work: how AI erodes psychological safety
Adoption looks strong on the dashboard. In the hallways, people stopped asking questions. Research now confirms what that quiet actually is: AI adoption measurably erodes the psychological safety teams need to learn, and it's a mechanism leaders can interrupt.
-
AI shame: why half your workforce is hiding their AI use, and what it's costing you
Forty-nine percent of employees have concealed their AI use at work. So have fifty-three percent of the C-suite telling everyone else to adopt it. The barrier is the fear of being judged as less competent for using the tool at all, not a skills gap.
04 / 05 Governance
The oversight, audit and accountability structures that let an organisation trust an AI system enough to actually rely on it.
AI systems fail differently to traditional software. A crashed application is visible immediately, an AI system that has quietly drifted, hallucinated, or been manipulated through a poisoned input can keep producing confident, wrong output for weeks before anyone notices. Regulators in Singapore, the EU and elsewhere have converged on the same baseline expectation: an organisation must be able to demonstrate, not merely assert, that a human remains accountable for what an AI system does, with a traceable record and a tested way to intervene.
In the modern AI era, governance is a design decision made before the deployment rather than the paperwork that follows it, the same way a building's structural engineering is decided before the interior design. Every framework in this section, delegation tiers, risk scoring, vendor contract terms, oversight architecture, exists to answer one question a board, a regulator or a customer will eventually ask directly: who is accountable when this system is wrong, and how would you know before they told you.
-
Your vendor can't explain the model either. That's the part people miss.
There's a version of the AI procurement conversation where documentation closes the transparency question. It doesn't, because most of what makes a modern model opaque was never the vendor's to hand over.
-
Parenting a superintelligent child: what values are we actually passing down?
Elon Musk called SpaceX and xAI staff the 'parents' of Grok. Two AI governance failures from the past eighteen months map almost exactly onto sixty years of parenting research, and only one of the four quadrants produces a system that keeps its values once nobody is watching.
-
Foundations of dependable agentic AI
Why engineering reliability into agentic systems depends on bounded task specifications and trajectory-level observability in production, not on how capable the underlying model is.
-
What your AI vendor contract is probably missing
Standard SaaS contract templates miss protections that matter specifically for AI vendor relationships, and the contract itself, not regulation, is currently an organisation's primary protection.
-
Where to begin: identifying tasks for quality, traceable, auditable AI agents
The TRACE framework that structures how to evaluate whether a task is right for autonomous agent deployment, and how much oversight it needs.
-
From human-in-the-loop to AI-on-the-loop: redesigning oversight architectures
How oversight structures need to change as AI systems take on more decision-making without a person approving every step, and why that shift is a design choice regulators already permit.
-
Beyond the pilot: a risk governance framework for scalable AI deployment
A governance framework for the step most AI programmes skip: moving from a working pilot to infrastructure that can be trusted at scale, built around a four-pillar risk model and a scoring method borrowed from industrial engineering.
05 / 05 Operations & Systems
The plant-floor, warehouse and engineering detail that decides whether an automation or AI project pays back, written from the site walk rather than the vendor's slide deck.
A strategy deck and a governance framework both assume the physical and operational layer underneath them works. It often does not. A warehouse management system that never reconciles with the ERP, a safety-rated control loop with no defined boundary for where a model's advice is allowed to end, a robotics case built on flexibility rather than repetition, all fail for reasons that have nothing to do with AI capability and everything to do with engineering practice that was skipped or rushed.
In the modern AI era, this is the layer most strategy and governance writing never reaches: instrumentation, certification, integration and the plant-floor economics that decide whether a project pays back before the product line changes underneath it. Written from twenty-six years running logistics, food technology and manufacturing digitalisation projects on the ground, this section is the operational detail that gets skipped when automation and AI get discussed in the abstract.
-
Why the AI layer has to sit outside the safety-rated control loop
A machine's safety function runs on a certified, bounded response time. A model's inference time is neither certified nor bounded. Most of what gets called an AI integration challenge on the plant floor is really that mismatch, misdiagnosed as a data or model problem.
-
Robotics ROI on the SME floor: where it's real, and where it's oversold
A cobot handles the four-thousandth repetition exactly like the first. That is the entire case for buying one, and it stops being the case the moment the task changes more often than the cycle repeats.
-
The Scope 3 number your customer needs is sitting in your OT data
From the 2026 financial year, Singapore's largest listed companies must disclose Scope 3 emissions, and a supplier with no activity-based data to offer becomes the industry-average estimate their customer is stuck reporting.
-
Your WMS and your ERP probably don't agree with each other
A warehouse management system rarely closes the gap it was bought to close. The gap usually sits at the handoff into the ERP, where someone still reconciles the two systems by hand, and that reconciliation step is where the real cost of a warehouse project sits.
-
Machine vision inspection pays back on some lines and stalls on others
The variable that decides the return is not the sophistication of the model. It is whether the defect you are asking a camera to catch has a stable visual signature at all.
