MODE
Mode:NormalHypa

NORMAL MODEWELCOME TO to HYPA MODE

Human capability, amplified. AI moved the ceiling on what one person can do. The tools were the easy part.You just changed a default. Most people never do.

The site activates when you do.

Most people will use this to write emails a little faster. A few will rebuild how they work entirely. Both are choices, and one of them compounds.

01  The idea

EVERYONE HAS ACCESS TO AI.VERY FEW HAVE CHANGED HOW THEY OPERATE.

Diagram: a person standing inside three rings; a unit of AI capability moves around each ring, passing behind and in front of them. Figure: the AIGA / U.S. DOT symbol sign, public domain.

Most people use it as a better search box. The opportunity is becoming capable of things you couldn’t do before.

02  Idea → execution

THE DISTANCE BETWEEN IDEA AND EXECUTION IS COLLAPSING.

There used to be a team in between.

Idea → execution
Idea
ExecutionHistorical · weeks to monthsExecutionHypa · hours to days

The historical route: research, learn, find a specialist, brief, wait, revise, build, test — hand-offs between every step. Typically weeks to months.Research · learn · revise · build · test — one seat, with AI. The hand-offs between them almost disappear. Specialists still matter; the distance to them doesn’t. Typically hours to days.

Diagram: two routes from idea to execution. The historical route passes through research, learn, find a specialist, brief, wait, revise, build, test, with hand-offs and waiting between steps. In Hypa mode the route is far shorter: research, learn, revise, build, test happen in one person’s seat with AI, and the hand-off steps almost disappear.
  • PhD experts, in their own field: 69.7 percent on GPQA Diamond.
  • A frontier model, 2025: 87 percent on GPQA Diamond.

IT ALREADY SCORES HIGHERTHAN THE EXPERTS.

Graduate-level questions, written to be un-Googleable. The people who set them hold doctorates in the subject. That gap is not closing — it closed.

GPQA Diamond ·Rein et al., 2023 ·Epoch AI benchmarks

THE TOOLS GOT SMARTER.DID YOUR WORK?

How much work can AI finish entirely on its own?

In 2019 it could handle a task that takes a person a couple of seconds. By 2025: around five hours of human work — succeeding about half the time.

That reach has been doubling roughly every seven months.

Adoption hasn’t kept pace. Only about one in five US workers say AI does some of their work.

Sources: METR time-horizon research, 2025 to 2026; Pew Research Center, September 2025.
03  Capability surface

AI ISN’T THE CAPABILITY.
ARE.

You are still the centre. The edge of what you can do is moving.

Your capability surface — six areas
  • Think
  • Research
  • Create
  • Build
  • Automate
  • Execute

AI moves the edge — unevenly, and without replacing you.

Capability surface
Latent   Within reach0 units within reach · illustrative
Diagram: a field of dots — each one a unit of capability — around a person at the centre, across six areas: think, research, create, build, automate, execute. Without AI only the units close to the centre are within reach. With AI the surface edge moves outward and many more units light up, unevenly — most in research and creation, least in execution.
  1. 01

    Think

    Normal

    Your reasoning is limited by time and attention.

    Hypa

    Work with AI as a thinking partner on decisions, plans and hard problems.

  2. 02

    Research

    Normal

    Hours of searching, skimming and second-guessing.

    Hypa

    Find, interrogate and synthesise information at a different speed.

  3. 03

    Create

    Normal

    Limited by your existing creative skills.

    Hypa

    Turn ideas into writing, imagery, presentations and prototypes.

  4. 04

    Build

    Normal

    Ideas wait for specialists and budgets.

    Hypa

    Prototype products, websites and systems directly.

  5. 05

    Automate

    Normal

    Repetitive work absorbs your week.

    Hypa

    Hand recurring work to systems that run without you.

  6. 06

    Execute

    Normal

    Ambition capped by execution capacity.

    Hypa

    Combine capabilities and ship things that used to need a team.

Tools change. Capability compounds.

04  The Hypa method

CAPABILITY COMPOUNDS.

Each pass through the loop is wider than the last — and the combinations are where the real leverage is.

01 Learn
Understand the capability.
02 Try
Experience it immediately.
03 Build
Create something real.
04 Apply
Integrate it into your work or life.
Diagram: the learning loop — learn, try, build, apply — drawn as three passes. In Normal mode only the first, smallest pass is active. In Hypa mode each further pass is wider, showing capability compounding.

NOTHING HERE MOVESUNTIL YOU DO.

The capability is already on the table. It stays inert until someone reaches for it — and every pass you make leaves the field brighter than you found it.

05  Operator
Hologram · live

Marcus Blenkhorn

Entrepreneur. Builder. Operator.

Marcus has spent around a decade working across entrepreneurship, commerce, logistics, technology, sales, e-commerce, fashion and digital products.

His interest in AI is practical: using technology to remove barriers that used to require specialists, agencies, developers, analysts and large teams.

HYPA MODE is his exploration of a larger question.

How much more capable can one person become when they learn to work with AI properly?
06  Hypa experiments

Editorial experiments in capability.

Each one sets a question, runs it for real and publishes what happened: result, lesson, capability gained. The first four are in planning.

  1. Exp 01

    Can one person build what used to require a team?

    Building · productStage 1 of 6 · Question
  2. Exp 02

    Can AI help someone learn a new specialist capability?

    Learning · skill acquisitionStage 1 of 6 · Question
  3. Exp 03

    How much of a workflow can actually be automated?

    Automation · operationsStage 1 of 6 · Question
  4. Exp 04

    Can one founder go from idea to working prototype without an agency?

    Creating · go-to-marketStage 1 of 6 · Question

CONGRATULATIONS.YOU USED AI TO WRITE AN EMAIL.

We can probably aim higher.

ToThe team
SubjectQuick update

SentDrafted with AI
Hi all — quick note to say the report is on track for Friday. I’ll share a draft tomorrow.
07  The book

One thesis: AI may be one of the largest expansions of individual human capability in modern history.

Serious mainstream non-fiction, written from the inside of the experiments — not a guide to tools.

Follow the book
08  First release

Enter Hypa mode — the first release.

A practical, foundational experience for people who want to change how they work with AI. It runs the six areas, in order. Then a seventh: multiply.

Before

  • I know AI matters.
  • I occasionally use ChatGPT.
  • I don’t know what’s genuinely possible.
  • I feel overwhelmed by tools.

After

  • I understand how to work with AI.
  • I research faster.
  • I create. I build. I automate.
  • I apply it to real problems.

Individual capabilities are useful.
Connected ones create leverage.

The seventh module, multiply, is where the six connect — the point at which one person starts shipping what used to need a team.

Diagram: six clusters of seven capability units — think, research, create, build, automate, execute — connect into one system of sixty-one. Individual capabilities are useful; connected capabilities create leverage.
09  Hypa mode for business

GIVING PEOPLE AI TOOLS ISN’T THE SAME AS GIVING THEM AI CAPABILITY.

Practical capability education for teams — adoption that changes how people operate.

Business enquiries

Licences issued

Five seats, paid for. Nothing about the work has changed.

Capability built

Same five people, same five seats. The work is done differently.

Diagram, two columns. Left, headed licences issued: five people, each holding a licence, all inactive. Right, headed capability built: the same five people holding the same five licences, each now surrounded by units of capability, all active.

In a study of 5,172 support agents, the ones given an AI assistant resolved 15% more issues per hour — 36% more for the least-skilled fifth, and no significant change for the most skilled. The licence was identical. The outcome was not.

And the capability is mostly not being built. Among employed Americans, 38% say training in how to use AI matters to them. 15.9% say their employer offers any.

Brynjolfsson, Li & Raymond · Generative AI at Work · Quarterly Journal of Economics, 2025  ·  Hashim, Kosar & van der Klaauw · Use of Gen AI in the Workplace · Federal Reserve Bank of New York, 2026

TWELVE MONTHS FROM NOWYOU ARE EITHER THE SAMEOR UNRECOGNISABLE.

Those are genuinely the only two on offer. The tools will have doubled twice more either way — the only variable left in the equation is you.

NORMAL IS A MINDSET.NOT A CEILING.

Nobody handed you the way you work — you inherited it, and no one has ever asked you to keep it. Defaults are the easiest thing in the world to change.

10  Join

ENTER
MODE.

Ideas, experiments and practical systems for becoming more capable with AI. Occasional email. No noise.

Channel
Email
Cadence
Occasional
Cost
Free

No spam. Unsubscribe any time. Privacy