Musk’s algorithm: the antidote to AI pilots that never deliver

Question requirements, delete steps, simplify, speed up, then automate — and watch outcomes improve.

Article link: https://open.substack.com/pub/johanosteyn/p/musks-algorithm-the-antidote-to-ai?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Audio summary: https://youtu.be/JnbNl5yUaeY

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

Walter Isaacson’s biography of Elon Musk is one of the best books I’ve ever read, and I encourage every business leader to read it. Not because you need to agree with Musk’s style, but because the book captures an uncomfortable operational truth: most organisations don’t fail at technology, they fail at process discipline. Isaacson describes what Musk calls “the algorithm”, a five-step approach to improving almost anything: question every requirement, delete any part or process you can, simplify and optimise, accelerate cycle time, and only then automate. That final step is the one most AI programmes get wrong. They start with automation, and end up scaling waste, locking in bad assumptions, and spending money to move faster in the wrong direction.

CONTEXT AND BACKGROUND

We are in a period where “automation” is being rebranded as intelligence. AI copilots, agents, and workflow tools promise to remove admin and accelerate decisions. Yet many companies still struggle to scale AI beyond pilots, even with big budgets and strong intent. The World Economic Forum recently captured this tension at Davos: organisations want transformation, but scaling AI still feels hard in practice.

Part of the problem is that AI is being treated like a shortcut around organisational complexity. But AI is not a broom. It does not tidy your process; it amplifies it. So if your workflow is fragmented, ambiguous, and full of unnecessary gates, AI will not magically make it elegant. It will make it faster, noisier, and harder to control.

Even AI leaders are now saying the quiet part out loud. In a recent Business Insider piece, CEOs from AI companies warned that automation is harder than it looks and that real deployment requires significant engineering and evaluation, not just a clever model.

INSIGHT AND ANALYSIS

This is where Musk’s algorithm becomes useful for ordinary businesses, especially those trying to justify AI spend. It forces the right sequence.

Step one: Question every requirement. In corporate life, requirements multiply like mould. “Compliance needs it.” “Legal insists.” “The committee wants visibility.” Most of these are not wrong, but many are unexamined. If you automate an unexamined requirement, you institutionalise it.

Step two: delete. Deleting feels dangerous because it removes comfort. But deletion is how you stop automation from becoming bureaucracy at scale. You cannot “automate your way” out of a process that should not exist.

Step three: simplify and optimise what remains. AI tools love standard inputs and stable definitions. If the same task has five different versions across departments, automation becomes a mess of exceptions.

Step four: accelerate cycle time only after the first three steps. Speeding up a broken process just produces errors faster and hides them behind dashboards and busy work.

Then, and only then, you automate. This is the core lesson: automation is the reward for discipline, not the substitute for it.

This is also why so many projects disappoint. TechRadar recently warned that more than half of AI projects could fail by 2026, often because of governance and data issues rather than the model itself. Those governance and data issues are frequently symptoms of messy, unowned processes.

IMPLICATIONS

For business leaders, Musk’s algorithm can be turned into a practical “automation gate” for AI spending. Before you buy or build, demand a process map, a list of deleted steps, a simplified target workflow, clear cycle-time baselines, and an agreed owner who carries accountability when things go wrong. If you can’t describe the work cleanly, you’re not ready to automate it.

For procurement teams, this is more than operations. It is risk management. Vendors will sell capability; your job is to buy outcomes with evidence. The AI market is moving towards “serious AI” focused on governed workflows, not flashy demos, as ITPro recently reported in the context of enterprise automation tools.

For South Africa, the stakes are higher because we cannot afford expensive theatre. We need productivity gains that translate into service delivery, competitiveness, and real capacity building. That requires discipline before tools.

CLOSING TAKEAWAY

Musk’s algorithm, as described by Isaacson, is not a tech mantra. It is a sequence for organisational honesty. Most businesses want AI to be the shortcut that saves them from hard choices, but the hard choices are the point: question, delete, simplify, speed up, then automate. If you reverse the order, you don’t transform the organisation, you mechanise its dysfunction. The leaders who win in the AI era will not be those who automate first, but those who make their work worth automating.

Author Bio: Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net