Your Competitor Is Waking Up Smarter Every Morning.

TL;DR: Last week revealed the true trajectory of AI: intelligent systems designed to compound through feedback. Anthropic restored access to Fable 5, SpaceXAI launched Grok 4.5, OpenAI rolled out GPT-5.6 and ChatGPT Work, and Meta, Moonshot (Kimi K2.7-Code), and Zhipu (GLM) advanced agentic and multimodal AI. The common thread is the intelligence loop: sense, act, measure, learn, improve. The future advantage belongs to whoever compounds intelligence fastest.

Last week was the clearest evidence yet that the AI industry has moved from building smarter models to building faster learning systems.

SpaceXAI released Grok 4.5 with an agent-first design, a stronger coding orientation with the recent Cursor acquisition, and aggressive pricing aimed at putting autonomous software development into the hands of millions of developers.

OpenAI pushed the frontier in a different direction with the GPT-5.6 series, a more natural GPT Voice experience (powered by GPT‑Live‑1), and ChatGPT Work, turning AI from a conversational tool into an operating layer for knowledge work.

Meta advanced its Muse Spark 1.1 systems toward multimodal and agentic experiences, while Moonshot’s Kimi K2.7 and Zhipu’s GLM updates pushed the open-model ecosystem toward cheaper, more accessible intelligence.

And, Anthropic’s Fable 5 allows us to push deeper into long-horizon agentic work.

On the surface, these look like competing product announcements. Zoom out, and they reveal a single underlying strategy: every major AI lab is racing to close the loop between sensing, reasoning, acting, measuring, and improving. The model is becoming the engine, but the loop is becoming the machine.

“Learning is not compulsory. Neither is survival.” — W. EDWARDS DEMING

In Groundhog Day, a weatherman named Phil Connors wakes up on February 2nd in Punxsutawney, Pennsylvania, and keeps waking up on February 2nd. He begins as a mediocre man, and by the end of the film he can play jazz piano, carve ice sculptures, quote French poetry, and catch a boy falling out of a tree because he has watched that boy fall a thousand times.

Nobody upgrades his brain along the way. He has exactly the neurons he started with. The universe gives him retries and lets him remember them.

The film is asking a question that now sits on the desk of every executive alive: what becomes of an ordinary intelligence when you give it enough attempts and a memory of each one?

For most of the twentieth century, the notion that a machine could learn from its own experience was treated as a category error. A machine does what it is told, and that is what makes it a machine.

In 1959, an IBM engineer named Arthur Samuel published a paper on a checkers program with a peculiar habit. It played against itself, thousands of times, adjusting the weights it used to judge a position according to which games it won. Samuel was not much of a checkers player himself, and his program soon became better than he was. He gave the phenomenon a name that sounded faintly ridiculous at the time, and called it machine learning.

In December 2017 a DeepMind system called AlphaZero was handed the rules of chess and nothing else. It played itself 44,000,000 times in nine hours, and then it played Stockfish, the strongest chess engine on Earth, one hundred times and lost none of them.

Nine hours is 32,400 seconds. Forty-four million games in 32,400 seconds is roughly 1,360 games every second. The games ran in parallel across a great deal of hardware, but the accumulated experience is still difficult to comprehend. A dedicated human professional playing five serious games a week through their lifetime might accumulate twenty thousand games. AlphaZero burns through an entire human career every fifteen seconds.

The machine starts where Phil Connors starts on his first February 2nd. Each loop changes the next one. He learns from each one.

In the skies over Korea in 1951, the Soviet MiG-15 could out-climb, out-turn and out-accelerate the American F-86 Sabre. On paper the fight was settled before it began, and yet American pilots kept coming home. A fighter pilot named John Boyd spent the next three decades chewing on the anomaly, and his answer had nothing to do with thrust or wing loading.

The American Sabre had a bubble canopy, so its pilot could see in every direction, and it had hydraulic controls. This meant the pilot could break from one maneuver into the next before the opponent had finished processing the last one. Observe, orient, decide, act, and then do the whole thing again while your opponent is still on step two. He named it the OODA loop and built a theory of war on top of it.

The story comes with a caveat that Boyd’s admirers usually leave out. Later historians have deflated the kill ratios he was working from, and the serious modern studies of Sabre-versus-MiG combat do not reach for the OODA loop at all to explain what happened. Boyd may well have been right for the wrong reasons. The principle he pulled out of that cockpit has survived for fifty years. It says something no balance sheet will tell you: the better machine loses to the faster cycle.

A loop can only run as fast as its slowest measurement.

AlphaZero could play 44,000,000 games because chess answers you instantly and cannot lie to you. Checkmate is checkmate. There is no quarter-end, no attribution model, no vice-president explaining that the loss was actually a “strategic win.”

The world beyond the board is far less obliging. Launch a campaign and the revenue arrives next quarter, tangled up with the weather, a competitor’s stumble, and somebody else’s ad algorithm. Hire an executive and you find out whether it worked in three years. Change a culture and the answer may never become clear. Most corporate decisions vanish into fog and are never scored at all, which is precisely why the same expensive mistake can be made twice by the same company without anyone noticing.

The bottleneck is measurement. Nearly every organization that announces it wants to become AI-native has a measurement problem wearing a technology costume.

The useful move is to stop treating “the business” as one undifferentiated thing and start sorting it by how quickly and clearly the world answers back.

Fast feedback: Where the world grades you in seconds, for free, and cannot lie.

Slow feedback: Where the world grades you in months, and mumbles.

Terminal feedback: Where the answer arrives after the decision is already dead.

The horror of Phil Connors’ first hundred loops is that everyone around him wakes up with no memory. He is the only one accumulating anything, while the entire town starts each morning at zero, tells the same joke, spills the same coffee, and steps into the same puddle on the same corner.

That town is your company.

A salesperson learns something true and expensive on a Tuesday call and it dies inside her skull by Friday. An engineer fixes a subtle failure, writes no note, and eighteen months later somebody fixes it again from scratch. A pricing experiment runs for six weeks, yields precisely one bit of information, gets argued about in a meeting, and is forgotten. Your organization wakes up every morning with no memory of its previous iteration, and then wonders why fifty years of experience have not compounded into fifty years of wisdom.

These systems offer a way to end the amnesia, and with it, a path to a superintelligent enterprise. A path to a superintelligent enterprise. Capture the interaction, score the outcome, keep the lesson, run it again tomorrow.

We are already seeing the first signs of this shift. Uber has begun reorganizing engineering around agentic pods, small teams designed to work alongside AI agents rather than treat them as external tools. The technology matters, but the deeper change is organizational. The unit of work is being redesigned around continuous feedback between people and intelligent systems.

I spelt out the seven steps necessary to build a superintelligent enterprise here.

In 1948, Norbert Wiener needed a name for the new science of feedback and control that he had assembled out of anti-aircraft gunnery and neurophysiology. He reached back into Greek for kybernetes and called the field cybernetics. The word refers to the person at the steering wheel.

He could have named it after the engine, the cargo, or the map, and instead he named it after the least glamorous role on the ship: the one whose entire purpose is to notice that the boat has drifted, correct it, notice again, correct again, all night, for as long as the voyage lasts.

The insight of cybernetics was that intelligence emerges from a continuous cycle of sensing, responding, and adapting. For three hundred thousand years, humanity was the only intelligence inside that cycle.

We observed. We acted. We learned. We corrected.

Now we are building systems that can participate in the loop with us.

Every company is entering an era where experience can accumulate instead of disappear, where lessons can persist instead of vanish, and where the accumulated wisdom of an organization can compound continuously.

A thousand customer conversations. A million design iterations. Ten thousand experiments. Each interaction becomes a signal. Each signal becomes learning. And each learning cycle becomes the foundation for the next.

The lessons of yesterday become the inputs of tomorrow.

This is the beginning of the self-learning superintelligent enterprise. Until next time,

Ram — 

Ram Srinivasan

MIT Alum | Author, The Conscious Machine | Global Future of Work and AI Adoption Leader published in Business Insider, Fortune, Harvard Business Review, MIT Executive Viewpoints and more.

—

A Message From Ram:

My mission is to illuminate the path toward humanity's exponential future. If you're a leader, innovator, or changemaker passionate about leveraging breakthrough technologies to create unprecedented positive impact, you're in the right place. If you know others who share this vision, please share these insights. Together, we can accelerate the trajectory of human progress.

Disclaimer:

Ram Srinivasan currently serves as an Innovation Strategist and Transformation Leader, authoring groundbreaking works including "The Conscious Machine" and the upcoming "The Substrate Shift"

All views expressed on "Substrate" and across all digital channels and social media platforms are strictly personal opinions and do not represent the official positions of any organizations or entities I am affiliated with, past or present. The content shared is for informational and inspirational purposes only. These perspectives are my own and should not be construed as professional, legal, financial, technical, or strategic advice. Any decisions made based on this information are solely the responsibility of the reader.

While I strive to ensure accuracy and timeliness in all communications, the rapid pace of technological change means that some information may become outdated. I encourage readers to conduct their own due diligence and seek appropriate professional advice for their specific circumstances.