1900 — Julesia

"An automaton may have its
'own mind' …
as if it had intelligence."

Nikola Tesla · 1900

"Their advent will create a revolution."

Nikola Tesla · 1919

History is proving, 126 years later, he was right.

1770

The Mechanical Turk

Long before any machine could genuinely think, the world was ready to believe one could. Wolfgang von Kempelen unveils "The Turk" — a chess-playing automaton that defeats challengers across Europe for decades, including, reportedly, Napoleon and Benjamin Franklin. It's a masterful hoax: a human chess master concealed inside the cabinet. The desire for a thinking machine predates the technology by nearly two centuries.

1843

Ada Lovelace and the First Algorithm

Working with Charles Babbage's never-built "Analytical Engine," Ada Lovelace writes what's now recognized as the first published computer algorithm. Her notes go further than Babbage himself — speculating the machine could one day manipulate symbols, even compose music, given the right instructions.

1898 — 1900

Nikola Tesla's Prophecy

Tesla demonstrates the world's first radio-controlled boat, calling it a "borrowed mind." In a June 1900 Century Magazine essay, he goes further, predicting an automaton with its "own mind" — independent of any operator, "capable of distinguishing between what it ought and what it ought not to do" — a startlingly precise description of an autonomous AI agent, written 126 years early.

1950 — 1956

The Idea Is Born

Alan Turing asks a deceptively simple question: "Can machines think?" Six years later, a small group of researchers meets at Dartmouth College and gives the field its name — Artificial Intelligence. Nobody agrees on what it means yet. Everyone agrees it matters.

1958

The Perceptron

Frank Rosenblatt builds the Perceptron, the first artificial neural network — a rough model of a single biological neuron, wired to classify simple images. It's decades early and badly limited by the hardware and math of its time, but it's the root of every neural network that follows, including the ones running today's AI.

1960s — 1970s

Early Ambition

ELIZA, one of the first chatbots, convinces people it understands them — it doesn't, but it's close enough to unsettle its own creator. Expert systems promise to encode human knowledge into rules. Optimism runs far ahead of what the hardware can actually deliver.

1970

The Uncanny Valley

Roboticist Masahiro Mori names an effect he's noticed: a face that's almost human but not quite doesn't charm people, it unsettles them. Called the "uncanny valley," it becomes the founding problem every AI avatar and digital human effort spends the next five decades trying to engineer its way past.

1971

The First Microprocessor

Intel packs an entire processor onto a single chip — the 4004. It's the birth of "micro-" as a prefix for computing, four years before a small startup borrows the word for its own name: Micro-Soft. Every chip that follows, including the ones running today's AI models, descends from this one.

Mid-1970s & Late 1980s

The AI Winters

Funding dries up — twice. The promises of the 1960s hadn't materialized on schedule, and the money moves elsewhere. AI doesn't disappear; it goes quiet, kept alive by researchers who keep believing even when nobody else does.

1980 — 1987

Expert Systems Go Commercial

Between the two AI winters, corporations bet big on "expert systems" — programs that encode a human specialist's know-how into thousands of if-then rules. Companies spend heavily building them into everyday decision-making. It's the field's first real commercial boom — and, when the systems prove too brittle and expensive to maintain, part of what triggers the second winter that follows.

1986

Backpropagation Breaks Through

Geoffrey Hinton, David Rumelhart, and Ronald Williams popularize backpropagation — an algorithm that finally lets multi-layer neural networks learn from their own mistakes. It's the missing piece that turns Rosenblatt's single-neuron Perceptron into something that can actually be trained at scale. It takes another two and a half decades and a lot more computing power before the payoff arrives.

1975 — 1986

Microsoft Builds the Platform

Bill Gates and Paul Allen found Micro-Soft in April 1975 to write software for an early hobbyist computer, the Altair 8800. In 1981, their MS-DOS ships alongside the first IBM PC — the deal that turns a small language company into the operating system running the personal computer age. Windows follows in 1985, and Microsoft goes public in 1986. None of it is "AI" yet, but it's the platform nearly every wave of software since, AI included, ends up running on.

August 1981

The PC Arrives

IBM releases its first Personal Computer, the IBM PC. It isn't the first personal computer ever built, but it's the one that makes "a computer on your desk" a mainstream business reality — and it sets the standard architecture the industry builds on for decades. Computing stops being something only institutions own.

1984

A Different Kind of Launch

Apple's Macintosh ad doesn't sell a computer — it sells an idea: technology stops belonging to institutions and starts belonging to individuals. It's not officially an "AI" moment, but it's the moment personal computing becomes personal. Every wave of intelligent technology since has chased that same promise.

1997

Deep Blue Beats Kasparov

IBM's Deep Blue beats world chess champion Garry Kasparov — the first time a machine beats the best human at a game long considered a proof of human intellect. Meanwhile, the internet is quietly generating the one thing AI needed all along: data, at a scale nobody had ever had before.

1999

Sony's AIBO

Sony releases AIBO, a robotic dog with real machine learning behavior. The first 3,000 units sell out online in twenty minutes. It's the first time an AI-driven product reaches ordinary living rooms — not research labs, not chess tournaments.

2011

Watson Wins Jeopardy!

IBM's Watson defeats Jeopardy!'s two greatest all-time champions on live television — parsing puns, wordplay, and open-ended trivia under real time pressure. Ken Jennings, in his own Final Jeopardy answer: "I for one welcome our new computer overlords."

1993 — 2006

Nvidia and the Rise of the GPU

Nvidia is founded to build graphics chips for video games — but their massively parallel design turns out to be almost perfectly shaped for neural network math nobody was thinking about yet. In 2006, Nvidia's CUDA platform opens that hardware up to general-purpose computing. It's the quiet hardware story underneath nearly everything AI does from here on.

2012

The Deep Learning Breakthrough

A neural network called AlexNet dramatically outperforms every prior approach at recognizing images, and the field pivots almost overnight. Deep learning goes from academic curiosity to the dominant approach to building intelligent systems — a shift still shaping everything built today.

2014

Generative Adversarial Networks

Ian Goodfellow invents GANs — two neural networks locked in a contest, one generating fake images, the other trying to catch them, each getting better by trying to beat the other. It's the first time AI can convincingly create something that looks real rather than just recognize or classify. Every deepfake and AI-generated image since traces back to this idea.

December 2015

OpenAI Is Founded

Sam Altman and Ilya Sutskever launch OpenAI as a nonprofit research lab, with Elon Musk, Peter Thiel, and Reid Hoffman among the backers pledging $1 billion to keep AI's most powerful capabilities out of any one company's hands — notably, three members of the so-called "PayPal Mafia" betting on AI the way they'd once bet on payments. The mission starts as explicitly anti-monopoly. Seven years later, the org built from it releases ChatGPT.

2015 — 2016

The Fourth Industrial Revolution

World Economic Forum founder Klaus Schwab popularizes the term "Fourth Industrial Revolution" — the idea that A.I., robotics, and ubiquitous connected technology are merging the physical, digital, and biological worlds the way steam, electricity, and computing once did before it. It's the large-scale version of exactly what Tesla saw coming in 1900: intelligence, embedded everywhere.

2016

AlphaGo's Move 37

Google DeepMind's AlphaGo defeats Go legend Lee Sedol 4-1. In game two, it plays a move so unconventional commentators — and Lee Sedol himself — think it's a mistake. It isn't. The moment AI shows something closer to intuition than raw calculation.

2016 — 2017

Sophia the Robot

Hanson Robotics unveils Sophia, a humanoid robot capable of 60+ facial expressions. In 2017, Saudi Arabia grants Sophia citizenship — the first robot in history to receive legal citizenship, and a genuinely controversial one: critics note Sophia gained rights many human residents didn't have.

2017

"Deepfake" Is Coined

The word "deepfake" appears for the first time, naming AI-generated videos that convincingly swap one face onto another body. It's controversial from day one — but the same underlying technique, generating a realistic human face and voice from data, becomes the root of every AI avatar built since, including the live, interactive ones in use today.

June 2017

"Attention Is All You Need"

Google researchers publish a paper introducing the Transformer architecture, throwing out the old word-by-word way machines read text. Its self-attention mechanism lets a model weigh every word against every other word at once, and it can be trained in parallel on GPUs instead of one word at a time — collapsing training times from months to days. It becomes the structural foundation under nearly every major generative model that follows, Julesia included.

2018 — Present

The GPT Era

OpenAI builds on the Transformer to create Generative Pre-trained Transformers, scaling from GPT-1 through the multi-modal, reasoning-capable systems in use today. What started as a paper about reading sentences becomes the engine behind an entire generation of AI products.

2020

AlphaFold Solves Protein Folding

DeepMind's AlphaFold cracks a problem biologists had chased for 50 years — predicting a protein's 3D shape from its amino acid sequence alone. It isn't a chatbot or a game-player; it's AI doing real scientific work, accelerating drug discovery and disease research worldwide. The result is significant enough to win an actual Nobel Prize in Chemistry in 2024.

2021

Anthropic Is Founded

A group of researchers, several of them formerly of OpenAI, start Anthropic with AI safety built into the company's founding premise rather than bolted on after. Their model, Claude, goes on to become the intelligence Julesia itself is built on — the direct line from this moment to the one at the bottom of this timeline.

2022

Text Becomes Pictures

DALL-E 2 and Stable Diffusion put "type a sentence, get an image" into ordinary people's hands for the first time. It's the moment generative AI stops being confined to text and starts making visual media — the same category of work Julesia's video and content tools now build on.

November 2022

ChatGPT and the Moment the Public Noticed

OpenAI, led by Sam Altman, releases ChatGPT to the public — 100 million users in two months, the fastest consumer app adoption in history at the time. Every prior milestone here was witnessed by researchers or television audiences. This is the first time ordinary people could sit down and talk to AI themselves.

March 2023

GPT-4 and the Multimodal Jump

OpenAI releases GPT-4, a sharp leap in reasoning ability that can also read images, not just text. It sets off the race that follows: every major lab pushes toward models that see, hear, and reason together, not just talk — the direct runway into the agentic tools built today.

November 2023

Governments Start Coordinating

The UK hosts the first international AI Safety Summit at Bletchley Park. 28 countries, including the US and China, sign the Bletchley Declaration — the first time governments agree, on paper, that the most powerful AI models need shared oversight. It's a declaration of intent, not a binding law, but it's the moment "AI safety" becomes a subject of formal diplomacy rather than just an industry conversation.

March 2023 — 2024

"Agentic" Gets Its Name

Open-source projects AutoGPT and BabyAGI go viral in March 2023, wiring GPT-4 into a loop — plan a step, take it, check the result, decide the next one — with almost no human in between. It's clumsy and unreliable, but it's the first time the public sees the shape of the idea. The word for it already existed: psychologist Albert Bandura coined "agentic" back in 1986 to describe people who act with intention rather than just react. In 2024, Andrew Ng borrows it for AI, popularizing "agentic AI" in his keynotes as the industry's name for exactly what AutoGPT had just shown the world.

2024

The EU AI Act

The European Union passes the first binding AI law from a major government body — tiering AI systems by risk and banning some uses outright, like real-time biometric surveillance in public spaces (with narrow exceptions). Where Bletchley was a declaration, this is enforceable regulation. Other governments start drafting their own versions soon after.

2024

Avatars Go Live

AI avatars stop being pre-recorded video and start being live: real-time, conversational, responding in the moment rather than being rendered ahead of time. Emotional range, natural lip-sync, and instant generation all arrive around the same year, turning "digital human" from a novelty demo into a real product category — the one Julesia builds live avatars inside of today.

2025

The U.S. Race for AI Dominance

The Trump administration reverses the prior White House's AI safety-focused executive order and releases "America's AI Action Plan," an aggressive push to fast-track AI infrastructure and cut regulation, framed explicitly around outcompeting China. A follow-up order in December moves to override state-level AI laws with a single federal standard. For the first time, AI leadership is treated as an openly declared national strategic priority, not just an industry trend.

2024 — 2026

Hollywood's AI Reckoning

Within two years, at least six films each claim some version of "the first AI-generated feature" — DreadClub: Vampire's Verdict (first to stream, July 2024), Post Truth (first wide theatrical release, July 2025), and Dreams of Violets (first at a major festival, Tribeca, June 2026) among them. None of the claims fully agree, because the industry hadn't agreed on what "AI-generated" even means yet. The most controversial: Dreams of Violets dramatizes a real massacre of Iranian civilians using AI-generated actors — critics called it "AI slop," or worse. The technology arrived before the ethics did.

2026

Musk's Terafab

Elon Musk announces SpaceX and Tesla will invest $16.8 billion to build "Terafab" in Texas — a planned 100-million-square-foot chip factory aimed at becoming the largest building on Earth. Chips stop being a background component and become an openly declared race of their own, with the same names already scattered across this timeline now racing to control the hardware underneath everything else. Ground just broke — whether it's built at this scale, on this timeline, remains to be seen.

Today

The Agentic Era

AI stops just answering questions and starts doing the work — planning, executing, checking its own results, working alongside the people who direct it. Remarkably close to Tesla's own 1900 language, 126 years later. This is the frontier the next chapter gets built on.

2026

Julesia Enters the Story

Born from one foundational belief: nothing is impossible — and for the first time, the tools actually exist to prove it. Nikola Tesla saw this coming 126 years early, describing an automaton with "its own mind" back in 1900. As the Fourth Industrial Revolution arrives, Julesia is founded: Jules + IA, Intelligent Applications. Not a company that watches the agentic era happen, but one that builds inside it — software, systems, and agentic applications built to move the world at the speed of a microchip.

Sourced in part from the Agentic History research timeline, a chronological archive of AI agent development.

"1984 didn't just launch a computer. It launched a mindset — that technology should work for the individual, not the other way around. That's the same belief behind every Intelligent Application Julesia builds."

— Julie, Founder of Julesia

We're not just watching this history.
We're building the next chapter.

Julesia turns 75 years of AI progress into intelligent applications and agents you can actually use — built for how you work, not how a lab thinks you should.

See what we've built