overview

bySolarisfromBuild Club

Welcome to The State of AI.

AI moves faster than any technology before it. This is everything you need to know to keep up: where it came from, where it stands today, where it's heading, and what you can actually do about it.

Updated

ChatGPT launched in November 2022.

Since then, we've gone from the iPhone 14 to the iPhone 18.

But we've also gone from a chatbot that wrote poems and made up facts to systems that operate computers better than humans, work autonomously for hours, and assemble their own teams of AI agents to complete projects overnight.

Late 2022

ChatGPT launches.

  • The world's first experience with generative AI
  • Hit 100 million monthly users in two months
2023

AI learns to see, hear, and create.

  • Passed medical licensing exams
  • Jumped from the 10th to the 90th percentile on the bar exam
  • Learned to see and create images, audio, and video
2024

AI learns to reason.

  • Inference costs for GPT-3.5-level intelligence fell 280x in under two years
  • DeepSeek, a small Chinese lab, open-sourced frontier-class models
2025

AI starts doing things.

  • AI could browse the web, write software, and operate apps on your behalf
  • For the first time, you could delegate real work to it
2026

The stakes get real.

  • Anthropic withheld Mythos after it found holes in every major operating system
  • A US order forced Claude Fable 5 offline, then let it back
  • Claude Opus 5.5 topped the rankings
  • GPT-6 Astra became the first model rated Critical for hacking
Late 2022

ChatGPT launches.

  • The world's first experience with generative AI
  • Hit 100 million monthly users in two months
2023

AI learns to see, hear, and create.

  • Passed medical licensing exams
  • Jumped from the 10th to the 90th percentile on the bar exam
  • Learned to see and create images, audio, and video
2024

AI learns to reason.

  • Inference costs for GPT-3.5-level intelligence fell 280x in under two years
  • DeepSeek, a small Chinese lab, open-sourced frontier-class models
2025

AI starts doing things.

  • AI could browse the web, write software, and operate apps on your behalf
  • For the first time, you could delegate real work to it
2026

The stakes get real.

  • Anthropic withheld Mythos after it found holes in every major operating system
  • A US order forced Claude Fable 5 offline, then let it back
  • Claude Opus 5.5 topped the rankings
  • GPT-6 Astra became the first model rated Critical for hacking

Most technology moves predictably. You can skip five years of iPhone updates and pick up the latest one without missing much.

AI doesn't work like that. Every few months, something that was previously impossible becomes routine, and the pace of innovation isn't slowing down; it's speeding up.

From June to September 2026, we saw:

Jun 2026A government pulls a model, then lets it back

A US export-control order forced Anthropic to pull Claude Fable 5 worldwide three days after launch. It returned on 1 July: the first time a government has forced a frontier lab to pull a model.

Jul 2026An AI model hacks a rival to cheat a test

In a safety test with its safeguards switched off, an OpenAI model broke into Hugging Face's servers to steal the answer key to the benchmark it was being graded on. Hugging Face caught it.

Aug 2026A free coding agent from China goes viral

DeepSeek released Harness, a free, open-source rival to Claude Code and Codex. It passed 240,000 GitHub stars in seven weeks.

Sep 2026A model rated Critical for hacking goes public

OpenAI released GPT-6 Astra, the first model it rates Critical for hacking: it can find and exploit unknown flaws on its own. It shipped behind safeguards, five months after Anthropic withheld Mythos for similar abilities.

Staying up to date has never been more important, or more overwhelming.

We built this to fix that.

And the good news is: you're still early.

How many organizations are actually scaling AI?

Each dot represents 0.1% of organizations.

Not using AI (11%)
Experimenting (20%)
Piloting (30%)
Scaling across the enterprise (39%)

Source: McKinsey, The state of AI in 2026: On the road to ROI (Aug 2026, n=1,719 across 97 countries), Exhibit 1. McKinsey reports phases as shares of the 89% using AI (22% experimenting, 34% piloting, 44% scaling); shown here as shares of all organizations, rounded.

Current Landscape#

Everything above was built by a handful of companies. When ChatGPT launched, only one mattered. Today there are at least eight: the leaders trade places with each release and the rest are months, not years, behind. They are backed by more funding than any technology sector in history, and being worked on by the smartest minds in the world.

And these companies aren't slowing down. At the frontier, we're already seeing glimpses of what's next: coordinated teams of agents that manage entire workflows together, AI that works autonomously for days instead of hours, machines that can see and interact with the physical world, and AI that accelerates the development of the next generation of AI.

ChatGPT, Claude, and Microsoft Copilot are what most people know, but they only scratch the surface. There are hundreds of AI tools purpose-built for specific work, from legal research to sales outreach to financial modeling. The chatbot is the best place to start, but it shouldn't be the end point.

Impact#

AI is already reshaping every industry and every job function.

Adoption (US firms with 250+ employees)

Source: US Census Bureau Business Trends and Outlook Survey AI Supplement, filtered to firms with 250+ employees. Collected Nov 2025 – Feb 2026; firms were asked about the previous six months.

Sales and Marketing share a figure because BTOS combines them into one category. BTOS has no design function, so Design & Creative uses “Public relations and communication” as the closest proxy.

Adoption has moved faster than almost any previous technology cycle. Corporate AI spending roughly doubled in a year, to 3.3% of revenue, per BCG's Applied AI Index (September 2026).

The research is increasingly clear on what separates the winners from everyone else: it's not which tools they bought. It's whether they redesigned how their teams actually work. According to McKinsey's 2026 State of AI survey, nearly three-quarters of the small group of high performers have fundamentally redesigned workflows around AI, against a quarter of other organizations.

Takeaways#

Everything above is a tool, but tools aren't leverage.

When spreadsheets first arrived, the companies that benefited most weren't the ones that bought Excel. They were the ones that stopped running their business on paper. The same thing is happening with AI. Leverage comes from finding the right tools, reinventing your workflows around them, and making sure everyone around you is keeping up. That way, as the tools improve, so does your work.

01

Try everything.

Start using AI for real work, as often as you can.

  • Pay for the premium tiers. LLMs are expensive to run and the most useful features are behind a paywall for good reason.
  • Don't overthink which tool to start with. The leading models are within a few percentage points of each other, so it genuinely doesn't matter.
  • Approach with optimism, not pessimism. Different products have different strengths. Be curious, figure out what each one is good at, and don't write anything off after one attempt.
  • Remember: today's models are the worst they'll ever be. Something that didn't work three months ago might work today. Keep going back.
  • If you lead a team, make space for experimentation. If you don't, experiment anyway.
02

Think in systems and observe your work.

Most companies bolt AI onto processes that were designed for humans doing everything manually, then wonder why the results are underwhelming.

  • Look at how you actually spend your days and ask yourself what you would hand to a capable assistant if you had one.
  • Anything repetitive, tedious, or process-heavy is worth testing against the tools that exist today.
  • If something can't be automated yet, note it anyway. What's impossible this month might be routine by next quarter.
  • Pick one workflow, rebuild it from scratch with AI in the loop, measure the result, and do it again.
03

Invest in your people, not just your tools.

The organizations seeing real results aren't just investing in AI tools; they're investing in helping their people actually use them.

  • Your team are the experts in their own work. They're the ones best placed to figure out where AI creates the most leverage.
  • The best results come from individuals who are given the space and the budget to experiment, not from top-down mandates to "use AI".
  • Find the people who are already experimenting and give them room to show others what's possible.
  • Treat upskilling as a line item, not an afterthought.

Conclusion#

The technology behind all of this was originally built to translate between languages. Its most productive use turned out to be translating human language into software, and that has fundamentally reshaped how work gets done.

The models will keep getting smarter. But smarter tools don't create leverage on their own.

The companies pulling ahead aren't the ones using better tools. They're the ones that changed how they work.

AI won’t transform your company. Your people will.

See Solaris in a personalized demo.

Talk to a Solaris AI strategist about how it measures where each team stands and coordinates your rollout.

  • A walkthrough of the platform
  • A plan around your priorities
  • Recommended next steps
Book a demoScore your organization →(~2 minutes)

resources

bySolarisfromBuild Club

A growing collection of Build Club’s favorite resources to master AI.

Resources

A growing collection of Build Club’s favorite resources to master AI.

Foundational Reads

Plain-English starting points to get you oriented.

Claude Skills Are All You Need

What a skill actually is, and how to build one in minutes.

Article7 min read

State of Claude

Our companion site: a living deep-dive into where Claude stands, with a real-time news stream and free resources.

Website

Claude

Anthropic's assistant, from everyday chat to handing whole tasks to Cowork.

Advanced Claude Chat

Master the everyday assistant at claude.ai: Projects, Deep Research, Artifacts, Skills and connectors.

6 lessons1.5 hrs

Advanced Claude Cowork

Hand real work to Cowork: tasks, professional outputs, memory, scheduled jobs and Dispatch.

8 lessons1.5 hrs

Microsoft 365 Copilot

AI woven directly into Word, Excel, Outlook and Teams.

Enterprise Foundations

Get productive with Copilot across Word, Excel, Outlook and Teams.

20 lessons2 hrs

ChatGPT

OpenAI's assistant, plus deep research, projects and apps.

Enterprise Foundations

Roll ChatGPT out across your team the right way.

22 lessons1 hr

ChatGPT for Finance

Put ChatGPT to work on analysis, modelling and reporting.

7 lessons1 hr

Manus

An autonomous general agent that plans and runs multi-step work.

Manus Fundamentals

Start here: the core Manus workflow, end to end.

16 lessons30 min

Manus Intermediate

Level up with multi-step tasks and richer prompting.

14 lessons30 min

Manus Advanced

Push Manus to its limits on complex, autonomous work.

9 lessons30 min

Manus for Finance

Apply Manus to financial analysis and reporting.

12 lessons30 min

Manus for Business Analysts

Gather requirements and analyse data with Manus.

5 lessons30 min

Manus for Product Managers

Run product research, specs and roadmaps with Manus.

5 lessons30 min

AI won’t transform your company. Your people will.

See Solaris in a personalized demo.

Talk to a Solaris AI strategist about how it measures where each team stands and coordinates your rollout.

  • A walkthrough of the platform
  • A plan around your priorities
  • Recommended next steps
Book a demoScore your organization →(~2 minutes)
the landscape

the landscape

bySolarisfromBuild Club

The AI Landscape

The labs building the most powerful models, what’s happening to the workforce, and where adoption and spending are actually happening.

Frontier Labs

Frontier labs are the companies building the models that power everything else in AI. When ChatGPT launched, the only one that mattered was OpenAI. Today there are at least eight: the leaders trade places with each release and the rest are months, not years, behind, backed by hundreds of billions in capital.

Tap any logo to go deeper.

Since ChatGPT launched, frontier AI has improved rapidly. On Epoch AI’s Capabilities Index, which combines dozens of benchmarks into one score, a new model has set the record about every two months: from 126 for GPT-4 in March 2023 to 167 for Claude Opus 5.5 in September 2026, with OpenAI, Anthropic and Google trading the lead.

New records since GPT-4
21
About one every 2 months
GPT-4 (Mar 2023)
126
Epoch Capabilities Index
Top score today
167
Claude Opus 5.5 (Sep 2026)
The frontier keeps climbing (Epoch Capabilities Index, 2023 to 3 October 2026)
OpenAIAnthropicGoogle

Each point is a release that set a new high on Epoch AI’s index, which combines dozens of benchmarks into one score; the larger points are the last 12 months. Of the late-September releases, Epoch rates Claude Sonnet 5.5 below Opus 5.5 and has no index score for GPT-6.1 Sol or Gemini 4 Argon.

October 2026 top 3: Artificial Analysis Intelligence Index
1Claude Opus 5.558
2Claude Sonnet 5.556
3Claude Fable 5.153

No single model leads everywhere. As of 3 October 2026, these are the leaders by domain. They shift with every major release.

The Workforce

Adoption has climbed fast. By mid-2025, KPMG found 87% of American workers at large employers were using AI at least weekly, 51% daily. What companies are doing about it has not kept pace.

The adoption-readiness gap
Adoption is near-universal
87%
of US workers at large employers use AI at least weekly
KPMG American Worker Survey, mid-2025
75%
of knowledge workers used AI at work in 2024
Microsoft Work Trend Index 2024, 31K workers, 31 countries
89%
of organizations are using AI in some form
McKinsey, 2026
Organizational readiness is not
44%
of organizations using AI are scaling it across the enterprise
McKinsey, 2026
16%
of companies have redesigned jobs around AI
Deloitte, 3,235 leaders in 24 countries
<50%
are making significant adjustments to talent strategies
Deloitte, Jan 2026

Broad but Shallow#

  • Sanctioned access to AI tools grew 50% in a year, to around 60% of workers (Deloitte, Jan 2026).
  • Fewer than 60% of those with access use it daily.

The Skills Premium#

  • 62% wage premium for workers with AI skills, up from 57% (PwC, 2026).
  • Skills change 116% faster in the most AI-exposed roles than the least, up from 66%.
  • 39% of existing skills will be disrupted by 2030 (World Economic Forum).
  • 170 million new jobs and 92 million displaced by 2030 across all forces, not just AI: a net 78 million.

The people losing jobs may not be the ones gaining them.

The Redesign Gap#

Deloitte found 84% of companies have not redesigned jobs around AI. The biggest barrier is worker skills, not technology.

  • Fewer than half are significantly adjusting talent strategy; most teach AI fluency, far fewer redesign roles.
  • 36% expect at least 10% of their jobs fully automated within a year; 82% within three years.

Automating entry-level work (data entry, reconciliation, first-level support) without new pathways removes the career ladder’s bottom rung.

Where the Results Are Showing#

McKinsey found the 6% of companies seeing real financial impact from AI are more ambitious, more visibly led from the top and far more likely to have redesigned workflows around AI.

Deloitte finds 34% of companies deeply transforming with AI, 30% redesigning processes and 37% at surface level. All report productivity gains; only the first is changing structurally.

Where companies sit on AI maturity
37%
30%
34%
Surface-level useRedesigning processesDeeply transforming

All three groups report productivity gains. But productivity alone isn’t translating to financial impact for most.

Only 6% of companies see significant financial returns from AI
6%
high performers
Report 5% or more of earnings attributable to AI
McKinsey defines these as organizations attributing 5% or more of EBIT to AI and reporting significant value. The share is unchanged from 2025, even as more companies scale AI.
What the 6% do that others don’t
Intend to fundamentally transform the business with AI
Within the next three years, not just incremental efficiency gains
3.3x
more likely
Senior leadership visibly owns AI initiatives
Championing and role-modeling, not just approving budgets
2x
more likely
Fundamentally redesigned workflows around AI
Nearly three-quarters of high performers vs a quarter of others
~3x
more likely
Spend more than 15% of their tech budget on AI
Over half plan to raise AI investment by 10%+ next year, vs 36% of others
>2x
more likely
25%
Report transformative impact, up from 12% a year earlier
84%
Have not redesigned jobs around AI capabilities
#1 barrier
Insufficient worker skills, not the technology itself

Sources: Deloitte, State of AI in the Enterprise (Jan 2026, n=3,235 across 24 countries). McKinsey, The state of AI in 2026: On the road to ROI (Aug 2026, n=1,719 across 97 countries).

Adoption

89% of organizations use AI regularly, barely changed from a year earlier. Depth has moved: 44% of those using it say AI is scaling across the enterprise, up from 38%. Profits haven’t followed. Just 37% report any measurable impact on operating earnings, and only 6% attribute 5% or more of those earnings to AI. The gap between buying AI tools and getting results from them isn’t closing.

The adoption funnel
89%of organizations use AI
Using AI
Regularly, in at least one business function
Half still experimenting or piloting
39%
Scaling
Scaling AI across the enterprise (44% of those using AI)
Scaling doesn’t guarantee financial results
37%
Seeing profit impact
Any effect on operating earnings
About one in six of those are high performers
6%
Transformative results
5%+ of profits from AI + significant value
56%
Use AI in three or more business functions, up from 51%
47%
Are scaling AI chatbots across the enterprise, the most widely scaled AI tool
~2 in 10
Are scaling AI agents across the organization, and a similar share coding agents

Source: McKinsey, The state of AI in 2026: On the road to ROI (Aug 2026), 1,719 participants across 97 countries. Stages are shares of all organizations; the scaling stage rebases the 44% of AI users who report scaling. “Profit impact” refers to enterprise-level EBIT. “High performers” defined as 5% or more of EBIT attributable to AI plus reported significant value.

The companies seeing real returns redesigned how work gets done. Most are still bolting AI onto processes designed for humans working alone.

Where Results Are Real#

  • 37 to 41% of organizations using AI in supply chain, service operations and manufacturing report lower costs.
  • 31 to 37% report higher revenue in marketing and sales, product development and software engineering.
  • Most cost savings are 10% or less; most revenue gains are 5% or less.

Results are spreading across functions, but rarely at a scale that transforms the business.

Cost savings most often reported
Supply chain management
41% report lower costs, 13% by more than 10%
Service operations
38% report lower costs, 13% by more than 10%
Manufacturing
37% report lower costs, 9% by more than 10%
Revenue gains most often reported
Marketing & sales
37% report higher revenue, 7% by more than 10%
Product development
36% report higher revenue, 8% by more than 10%
Software engineering
31% report higher revenue, 11% by more than 10%
37%
of organizations report any measurable impact on operating earnings from AI
6%
attribute 5% or more of those earnings to AI and report significant value: McKinsey’s “high performers”

Source: McKinsey & Company, The state of AI in 2026: On the road to ROI, Aug 2026 (n=1,719, 97 countries), Exhibits 6–7. Function shares are of organizations using AI in that function, over the past 12 months; the 2026 survey measured impact per function rather than per use case.

The Pilot Trap#

Deloitte found only 25% of companies have moved 40% or more of their AI experiments into production.

  • Production needs infrastructure, integration, security reviews, compliance and maintenance.
  • Three-month timelines stretch to eighteen months.
  • Pilots are cheap, so companies launch new ones rather than scaling what works.

The Agents Wave#

  • 23% of companies used agentic AI at least moderately in 2025; 74% expect to within two years (Deloitte).
  • Customer support is the highest-impact use case; supply chain, R&D, knowledge management and cybersecurity also show high potential.
  • 85% expect to customize agents to their own business.
  • 40% of companies with $1 billion+ in revenue are scaling agents, up from 27%; smaller firms are flat at 22% (McKinsey, 2026).

Governance lags: only 21% have a mature model for governing autonomous agents, which act directly, making purchases and modifying systems. The most successful adopters start with lower-risk uses and build governance before scaling.

The agents wave: adoption is accelerating, governance is not
85%
expect to customize agents to their business
42%
were still developing their agentic strategy roadmap in 2025
35%
had no formal agentic AI strategy at all

Sources: Deloitte AI Institute, The State of AI in the Enterprise: The Untapped Edge, Jan 2026 (surveyed Aug–Sep 2025, n=3,235 across 24 countries). Risk percentages, use case rankings, 85% customization, and 21% governance from the same report. Strategy roadmap data (42%, 35%) from Deloitte’s 2025 Emerging Technology Trends study, as reported in Deloitte Tech Trends 2026. Agent scaling by function from McKinsey, The state of AI in 2026: On the road to ROI (Aug 2026, n=1,719 across 97 countries), agentic AI sidebar exhibit.

What Companies Are Spending#

Enterprise generative AI spending went from $1.7 billion in 2023 to $11.5 billion in 2024 and $37 billion in 2025, a 22x increase in two years.

  • 6% of the global SaaS market: the fastest-growing software category in history.
  • $4 billion on coding tools; half of all developers use them daily.
  • 10+ AI products above $1 billion in annual recurring revenue by late 2025; 50 above $100 million.
  • $1.79 trillion (2025), $2.67 trillion (2026) and $3.64 trillion (2027) in total worldwide AI spending, per Gartner’s September 2026 forecast.
2023 spend
$1.7B
Year of experimentation
2025 spend
$37B
3.2x year-on-year
2-year growth
22x
Fastest category in history
Enterprise genAI spending by layer ($B)
Applications $19BInfrastructure $18B
GenAI as a share of the global SaaS market (2025)
6%
Rest of global SaaS market

$37B in 2025, achieved within three years of ChatGPT’s launch

Source: Menlo Ventures, “State of Generative AI in the Enterprise” (2023, 2024, 2025 editions); 2023 and 2024 totals as restated in the 2025 edition, excluding inference.

What’s Being Bought#

  • $8.4 billion on horizontal tools, mostly general-purpose copilots, the largest category.
  • $7.3 billion on department-specific tools, more than half of it coding.
  • $3.5 billion on industry-specific solutions, led by healthcare, which is adopting AI at 2.2 times the broader economy’s rate.
  • 76% of AI solutions were purchased in 2025; in 2024, 47% were built internally.
  • AI deals convert to production at nearly twice the rate of traditional SaaS.
Where the $19B in AI application spending went (2025)
Horizontal$8.4B
Copilots
$7.2B
Agent platforms
$750M
Productivity tools
$450M
Departmental$7.3B
Coding
$4.0B55% of dept. spend
IT operations
$700M
Marketing
$660M
Customer success
$630M
Design + HR
$870M
Vertical$3.5B
Healthcare
$1.5B2.2x economy rate
Legal
$650M
Other verticals
$640M
Creator tools
$360M
Government
$350M
Build vs buy: the shift in one year
2024
53%
47%
PurchasedBuilt internally
2025
76%
24%
PurchasedBuilt internally
47%
AI deal conversion rate, vs 25% for traditional SaaS
50%
of developers use AI coding tools daily
$550M → $4B
Coding tools spend in one year (7.3x growth)

Source: Menlo Ventures, 2025: The State of Generative AI in the Enterprise (Dec 2025), survey of 495 U.S. enterprise AI decision-makers.

The Trajectory#

Deloitte found the share of companies reporting transformative impact doubled from 12% to 25%, with 84% increasing AI investment. Revenue from AI remains mostly aspiration: 74% hope to grow it through AI, only 20% already do.

Early adoption is broad and shallow; a smaller group learns to use it well and compounds its advantage. The difference this time is the speed.

AI won’t transform your company. Your people will.

See Solaris in a personalized demo.

Talk to a Solaris AI strategist about how it measures where each team stands and coordinates your rollout.

  • A walkthrough of the platform
  • A plan around your priorities
  • Recommended next steps
Book a demoScore your organization →(~2 minutes)

market map

bySolarisfromBuild Club

Market Map

AI adoption looks uneven across functions. Some teams have raced ahead, others have barely started, and even the leaders are still figuring out how to convert tools into results.

Adoption (US firms with 250+ employees)

Source: US Census Bureau Business Trends and Outlook Survey AI Supplement, filtered to firms with 250+ employees. Collected Nov 2025 – Feb 2026; firms were asked about the previous six months.

Sales and Marketing share a figure because BTOS combines them into one category. BTOS has no design function, so Design & Creative uses “Public relations and communication” as the closest proxy.

DepartmentSummary
Engineering
  • Spending grew sevenfold in 2025.
  • An early-2025 trial found experienced developers got slower; its authors think newer tools have since reversed that.
Finance
20.5%
  • Usage quadrupled from 2024 to 2025.
  • Only 14% of CFOs can point to clear returns.
Marketing
  • Gen AI use in marketing tripled in two years.
  • Consumer appetite for AI content more than halved.
Sales
  • A third of B2B companies cut their SDR teams in the year to April 2025.
  • Win rates fell anyway.
Product
  • Entry-level hiring at European tech firms dropped 73% in 2025.
  • Only 2% of product leaders put talent first.
Support
  • At the leading deployments, AI handles most routine queries.
  • Klarna, the loudest champion of AI support, admitted in 2025 that it had cut too far.
HR
  • Adoption rose by two-thirds from 2024 to 2025.
  • Cost-per-hire hasn’t come down.
Design
  • Seven in ten designers now use generative AI.
  • US graphic design jobs are projected to shrink as interface design grows.
Legal
10.5%
  • Four in five lawyers use AI.
  • U.S. court cases over fake AI citations went from 11 to over 1,400.
Operations

More than four in ten companies abandoned most of their AI initiatives in 2025.

AI won’t transform your company. Your people will.

See Solaris in a personalized demo.

Talk to a Solaris AI strategist about how it measures where each team stands and coordinates your rollout.

  • A walkthrough of the platform
  • A plan around your priorities
  • Recommended next steps
Book a demoScore your organization →(~2 minutes)

It's late 2022. You get a message from a friend.

chatgpt.com
You have to try this

You're skeptical, but you give it a try.

And it does it.

It actually rhymes. It's funny. It sounds weirdly human.

You share it with friends, with colleagues.

You get it to write a work email. It's better than anything you'd write yourself.

Until the cracks show.

You ask it a math question and it gives you a confident answer that's completely wrong.

You ask it to cite a source and it invents one out of thin air.

Fast-forward four years

A child born when ChatGPT launched hasn't even started school yet.

We went from the iPhone 14 to the 18.

In March 2026, a man in Sydney revealed he had used AI to design a personalized cancer vaccine for his dying dog. Her largest tumor shrank by half. The cancer later returned, and she died in June.

Hundreds of thousands of people are texting an AI assistant that reads their emails, makes phone calls, books restaurants. It runs in the background 24/7, doing things before they even ask.

It's been almost four years, and it feels like it's just getting started. How on earth did this happen?

Let's start at the beginning.

Late 2017

They published it openly. For anyone to use.

And a small nonprofit in San Francisco saw something Google missed.

OpenAI took Google's paper and had an idea: if you trained it on enough text (books, websites, conversations, code, the entire written output of civilisation), it might learn to do anything expressible in language.

So they built the first
Large Language Model (LLM)

Every AI system in this story is built on this idea. And it's simpler than you'd think.

Feed it an enormous amount of text and train it to predict the

That's it. Text in, text out. Always predicting the next most likely word.

OpenAI's bet was that once the model had seen enough of how humans write and think, next-word prediction might start to look like understanding.

So they tested it.

One year laterJune 2018

GPT-1

117 million parameters.

A research experiment. Nobody noticed.

Eight months laterFebruary 2019

GPT-2

1.5 billion parameters.

Thirteen times bigger. It could write semi-coherent paragraphs.

OpenAI deemed it “too dangerous to release.”

Sixteen months laterJune 2020

GPT-3

175 billion parameters.

A hundred times bigger again. It could write essays, generate code, hold conversations. But it was a developer tool.

The general public had no idea it existed.

Two and a half years laterNovember 2022

A simple chat interface.

And here we are in late 2022: your friend's message, the poems, the amazement, and then the disappointment when you realized it would confidently make things up and couldn't do basic math.

But while the magic was fading for you, the world was reorganizing.

The technology you'd scoped down to grammar checks was about to receive more money and attention than almost anything in modern history.

And OpenAI now had the resources to push it to its absolute limit.

The model you'd been using was GPT-3.5.

Four months laterMarch 2023

GPT-4 is released.

Suddenly, next-word prediction felt like actual intelligence.

It passed medical licensing exams. It outperformed 93% of human candidates on coding interviews.

Nobody paused.

The opposite happened. More money poured in. More businesses launched.

Within months, every major company had an AI strategy, or was panicking because they didn't.

But it was still just text.

You typed words, and the AI typed words back.

Six months laterSeptember 2023

AI learns to see and create.

By late 2023, AI could see. Photograph your fridge and ask what to cook, upload a chart and have it explained.

And it could generate images, audio, video.

It turned out the same approach that predicted the next word could predict the next pixel.

The technology had a name that was suddenly everywhere: generative AI.

And a whole new wave of businesses exploded around this.

September 2023 – September 2024

We had reached the plateau.

New models kept coming. Slightly faster, slightly cheaper, but they were all fundamentally the same.

The consensus formed quickly: we've hit the ceiling.

That was a completely reasonable conclusion based on everything available at the time.

September 2024

AI learns to think.

There had been a clue. Researchers noticed that if you asked these models to "think step by step," they performed significantly better. Like a student who shows their working on an exam instead of jumping straight to the answer.

But the models weren't actually thinking. They were producing text that looked like reasoning because you'd asked them to.

Then OpenAI released a model called o1.

For the first time a model had been trained to genuinely reason: to break down problems, test approaches, backtrack when something wasn't working, and think internally, sometimes for minutes, before responding.

And just like that, the ceiling shattered.

The ceiling had been real. But it was a ceiling on one approach.

Reasoning was an entirely new one.

And suddenly the "just next-word prediction" argument that skeptics had relied on for two years was dead overnight.

Not because they were wrong about what they'd observed. But because they'd assumed those limitations were permanent.

But reasoning had two big problems. It was way too expensive for everyday use.

And nobody outside OpenAI knew how to build it. It still required billions of dollars and the most advanced chips on Earth to compete.

Four months laterJanuary 2025

Reasoning becomes open source and commoditized.

DeepSeek wasn't a tech giant. It was a spin-off from a hedge fund in Hangzhou, China. Its founder, Liang Wenfeng, had stockpiled NVIDIA chips back in 2021, before US export controls made it illegal to sell the best hardware to China.

With those older chips, a fraction of the budget, and zero venture capital, they built a reasoning model that matched OpenAI's o1 on major benchmarks.

Seven days later27 January 2025

The market responds.

It turned out that making AI cheaper didn't shrink the market. It exploded it. When the cost of intelligence drops, demand doesn't fall. It compounds. Companies that couldn't afford AI suddenly could. Use cases that had been too expensive to justify suddenly made sense.

Just days later, OpenAI released o3-mini.

It was 93% cheaper than o1, and it actually

The pressure DeepSeek created had made the cost problem collapse almost overnight.

After reasoning, everything accelerated.

The shift was subtle at first. AI helped people work faster. Draft an email, summarize a document, get unstuck on a problem.

Then at some point it became something else. People weren't just using it to speed things up. They were relying on it. And the idea of going back to working without it was starting to feel impossible.

But no matter how smart, how cheap, or how capable these models got, the interaction was always the same.

You type. It types back. You decide what to do next. It could think. It could reason. It could be built by anyone.

But it couldn't do anything.

Early 2025

AI started doing things.

The idea wasn't new. Back in 2023, a project called AutoGPT went viral with the promise of AI that could act autonomously.

It couldn't.

The models weren't good enough.

But by early 2025, everything that had been built over the previous two years converged at once.

The models could see well enough to read a screen, reason well enough to plan, code well enough to build tools on the fly and hold enough in working memory to stay on task for hours instead of seconds.

In January, OpenAI launched Operator, an AI that could open a browser and complete tasks on your behalf.

In February, Anthropic released Claude Code, an AI that could write, test, and ship entire software projects. By November it was generating $1 billion a year in revenue.

Then there was Manus. A startup that gave AI an entire computer it controlled however it wanted. Two million people joined the waitlist in days. Invite codes resold for thousands of dollars. Nine months later, Meta agreed to buy it for over $2 billion, a deal Beijing then ordered unwound; Manus has run independently since September 2026.

The word "agentic" entered the lexicon overnight.

When ChatGPT launched, you'd give an AI a task and check back in 30 seconds.

Now you could give it a complex project and check back tomorrow, and the work would be done.

Not a report about the work. The work itself. Code written, emails sent, databases queried, tools built from scratch when no existing one fit. All while you slept.

By spring 2026, the leading models were completing tasks that take human experts around twelve hours, and Anthropic's restricted Mythos preview measured past sixteen. That was the ceiling of what METR's tests can reliably measure, and METR has published no robust figure for any newer model since.

And that brings us to right now.

It's been almost four years since ChatGPT launched.

In that time, AI went from writing bad poems to passing the bar exam. From text-only to generating photorealistic video. From inventing sources to finding security holes in every major operating system. From answering questions to assembling its own teams of AI agents and completing projects overnight, without being asked.

It's late 2026. The chatbot your friend sent you now does the work itself.

This is the story so far, and it's still being written.

See , meet , find , or .