The pitch is everywhere once you start noticing it: a Japanese engineer watched a kingfisher dive into a pond, copied the beak, and solved his train’s noise problem in one flash of insight. I kept meeting that version while researching sustainable design. In my day job doing clinical research, a figure without a source document does not count, no matter how good it sounds. I pointed that same rule at biomimicry’s most famous story and went looking for what actually happened.

The version you have heard

In the 1990s, Japan’s high-speed trains had a problem. They were fast, but every time one shot out of a tunnel it produced a loud boom, loud enough to draw complaints from people living hundreds of metres away. An engineer named Eiji Nakatsu, who happened to love birdwatching, watched a kingfisher dive and realised the bird’s beak was the answer. He reshaped the front of the train to match it. The boom went away, and the train got faster and used less power as a bonus.

Every claim in that paragraph points at something real. Almost every emphasis in it is off.

What actually happened

Start with the problem, because that is where the myth first bends. The brief was not “make it faster.” Nakatsu has said plainly that speed was the easy part; the hard part was noise, held to one of the strictest limits in the world (Japan’s environmental standard caps trackside noise at 70 dB in residential zones and 75 dB in commercial zones). The whole redesign was a fight against sound.

Then there is the number of birds. Not one. Three.

The tunnel boom was only one of two noise problems, and the kingfisher solved only that one. The bird dives from air into water, a fluid roughly 800 times denser, and slides in almost without a splash because of the long, tapered wedge of its beak. A train punching into the still air inside a tunnel faces a version of the same physics, so the front car became a 15-metre beak-like cone.

But the train had a second scream, from the pantograph, the hinged arm on the roof that draws power from the overhead wire. At speed, air tearing past that arm broke into a churn of small vortices called a Karman vortex street, and that churn was loud. To quiet it, Nakatsu’s team looked at the owl, one of the quietest hunters in the sky: its leading wing feathers carry a comb of tiny serrations that shred airflow into micro-turbulence before it can build into sound. Engineers cut matching serrations into the pantograph. A third bird, the Adelie penguin, lent its smooth spindle body to the strut that holds the pantograph up, cutting drag. Kingfisher for the nose, owl for the noise, penguin for the strut. The single-bird version deletes two thirds of the actual engineering.

The moment of inspiration was a lecture, not a walk

The popular story is a lone genius struck by sudden insight at a pond. The real account is less romantic and more useful. Nakatsu credited the owl idea to a lecture by an aircraft engineer, Seiichi Yajima, who was also a member of the Wild Bird Society, on how much aviation already borrowed from birds. The insight arrived in a conference room, from someone else’s reading.

It did not resolve in an afternoon either. The pantograph work alone took four years. The team borrowed a stuffed owl from a zoo in Osaka and put it in a wind tunnel next to a stuffed turtledove to measure which was quieter, then tested scale models and full-size prototypes. For the nose, they fired projectiles of different shapes into a pipe and ran simulations on a supercomputer built for space research. The kingfisher beak won that contest against every other shape tried. It was the shape that survived the testing, not a hunch that happened to work.

The numbers, with the caveat

Nakatsu’s own account gives three figures for the 500-series that entered service in March 1997: around 30% less air pressure, about 15% less electricity, and roughly 10% higher speed than the train it replaced. I am treating those as the anchor because they come closest to the source.

Watch what happens as the story travels, though. One retelling says “30% less air resistance.” Another folds the saving into “energy.” A separately reported measured run at 270 km/h put the power saving at about 13% against the older series, close to 15% but not identical. None of this makes the achievement smaller. It makes the point that even a well-documented success gets fuzzy in transmission, and that fuzziness is worth naming rather than smoothing over.

The objection worth taking seriously

If supercomputer testing had already identified the ideal nose shape, did the kingfisher actually teach anyone anything, or did the bird just happen to match a solution physics would have reached anyway?

The honest answer: the bird was a shortcut, not a revelation. Evolution had already run millions of years of testing on entering a dense fluid cleanly, and the kingfisher is one of its passing results. Nakatsu did not extract a law of physics the engineers lacked. He used a living creature as a very good first guess, then confirmed it the slow way, with models and data. That is a more modest claim than “nature solved our problem,” and a more useful one: it tells you what biomimicry is good for. Not magic. A shortlist of forms already stress-tested by survival, which you still have to verify yourself.

Why this matters past the train

Once you have seen how much a true story loses in the retelling, “nature-inspired” claims read differently. Most of them are not doing what Nakatsu did. They borrow the prestige of the kingfisher without any of the wind tunnels, the four years, or a number you can check.

The Shinkansen earns the label because there is a measured result underneath it and a documented process behind it: a boom that dropped below a legal limit, a power figure recorded on a real run, a four-year paper trail. Strip those away and you are left with a nice image of a bird, which is exactly what most “inspired by nature” marketing is: the image without the receipt.

If you have a nature-inspired design story you have always taken on faith, send it to daria@thetrinitytribe.com. It goes on the list for this series.

A nonprofit founder is finishing her third grant proposal this month. She is also the website manager, the donor communications person, and the one who actually runs the programme. She has no budget for outside help. She has been told that using artificial intelligence (AI) is cutting corners. A well-funded organization in the same sector is using AI across every one of those functions and calling it operational excellence. The gap between them is not a values gap. It is a resources gap dressed up as a moral one.

The privilege nobody names

Adrianna

The framing most people inherit treats AI adoption as a moral test. Businesses that use AI are cutting corners, displacing workers, accelerating something vaguely harmful. Businesses that refuse AI are principled, human-centered, authentic. This framing feels clean. It contains a structural assumption that rarely gets examined: that opting out is equally available to everyone.

It is not.

OECD (Organisation for Economic Co-operation and Development) research published in 2025 found that 40% of large firms with 250 or more employees were using AI in 2024. Among firms with 50 to 249 employees, that figure drops to 20%. Among small firms with 10 to 49 employees, it falls to 11.9%. That gap does not reflect stronger ethics. It reflects capital, technical staff, and the operational slack to run experiments that might not pay off for six months. Small businesses are often too stretched for that kind of patience.

McKinsey’s 2025 State of AI survey found 88% of organizations use AI regularly in at least one business function, up from 78% the year before. That headline sounds as if the debate is settled. What it obscures is that the 88% skews heavily toward larger, better-resourced organizations with existing infrastructure and dedicated teams. Widespread adoption does not mean equal adoption. Equal adoption would not mean equal returns.

Returns are more concentrated still. PwC’s 2026 AI Performance Study found that just 20% of organizations are capturing nearly three-quarters (74%) of AI’s measurable economic value. The winner-takes-most dynamic is already in place before most small businesses have run a single pilot. The firms gaining the most were already the best resourced.

There is also a double standard worth naming directly. When large companies adopt AI, it is called innovation and operational efficiency. When small businesses do the same, they are often told they are cutting corners or replacing real workers. The same behavior earns different verdicts depending on who is doing it. This asymmetry discourages the businesses that have the most practical reason to engage with AI carefully.

The useful question is not who is pure. It is who bears the cost when the answer is no.

What inefficiency actually costs

Daria

Here is what a real week looks like for a solo founder or small team without AI support. You write every email from scratch. You rebuild proposals by editing old files, usually missing something in the process. Research that takes an AI forty seconds takes you forty minutes, interrupted twice. Admin piles up. Content does not happen because by the time client work is done, there is nothing left.

That invisible work carries a real cost. Not just time. Cognitive load, the mental effort of holding every loose end in your head at once.

A founder spending three hours on tasks that could take thirty minutes is not three hours behind on admin. They are three hours behind on thinking, deciding, and building the thing that makes their business worth running. Small businesses cannot absorb this the way larger firms can. A company with thirty staff treats administrative waste as a line item. A two-person business treats it as a slow, compounding crisis.

There are three realistic paths for a small business that wants to operate without AI.

Stay fully manual and price at premium. Some craft businesses, therapy practices, and high-trust consultancies operate here profitably. It is a real model. But it requires pricing that reflects the true cost of manual labour. Most underfunded founders cannot yet command those rates, particularly early on.

Hire specialists to cover what AI would otherwise handle. That means a copywriter, developer, strategist, project manager, and analyst. Together, those roles cost more per month than most early-stage businesses turn over. “Just hire humans” is advice from a position of financial comfort. It does not reckon with the actual choices a small business faces between payroll, software, rent, and still somehow being visible in the market.

Use AI selectively, with a clear internal policy about where it belongs and where it does not. This is not a compromise. It is the path that gives a small business a realistic shot at staying viable without burning out the people running it.

JPMorgan Chase Institute research found that 82% of small businesses using AI report enhanced operational efficiency, and 77% report improved competitiveness against larger firms. That is not an argument for adopting every available tool. It is an argument for taking the question seriously.

The divide that actually matters

TTT

The real risk with AI in small business is not adoption. It is adoption without criteria.

Generic content that sounds like every other business in the category. Statistics published without review. Automation that undercuts client trust. Privacy shortcuts applied to data that should never have left the building. These failures are real and worth preventing. They are solved by having clear standards for where AI is used and how, not by refusing to engage with the question entirely.

There is a fifth failure mode this framing risks skipping over: environmental cost. Running large language models draws real electricity, at a scale that stopped being trivial a while ago. The IEA’s April 2026 report put global data centre electricity demand at roughly 485 terawatt-hours in 2025, on track to reach 950 terawatt-hours by 2030, with AI-driven demand as the fastest-growing share of that number. For a purpose-driven small business that has spent years building a low-carbon operation, treating a favourite AI tool as a free lunch is exactly the kind of adoption without criteria this piece is arguing against. The honest position is not to pretend that trade-off does not exist. It is to weigh it deliberately: use AI where the operational case is strong enough to justify the energy cost, favour providers who publish real efficiency data over vague offset claims, and stay sceptical of any tool that treats compute as free.

The divide that matters is not AI versus no AI. It is strategic AI versus chaotic AI. Chaotic AI produces more noise, faster. Strategic AI reduces waste, sharpens decisions, and keeps humans in the roles where their judgment actually counts.

For a small business, building that distinction starts with five questions:

1. Where is time actually going, and which of those tasks require human judgment to do well?
2. Which workflows create the most cognitive load for the people running them?
3. Where would automation improve the client experience, and where would it erode it?
4. What data should never enter a third-party tool, regardless of how convenient that tool is?
5. Where is the current operation producing output that nobody actually needs?

Those questions matter more than any subscription. They are also the foundation of AI literacy, which is more useful to a small business than AI fluency. A business that understands what a tool consistently gets wrong, which outputs require mandatory human review before reaching a client, and what it should never handle is in a much stronger position than one with twelve subscriptions and no policy.

A small business that has done this well uses AI for the work that was never a good use of human attention: drafting, formatting, summarizing, preparing options for someone to evaluate, handling routine communication. It keeps humans in every decision requiring taste, accountability, or genuine care. The work gets faster. The standards do not drop. The founders are still standing in two years.

AI should not replace your values. For many small businesses, it may be one of the few tools that gives those values a fighting chance in a market where better-resourced competitors have already made their choice.

If you want to work out where AI actually belongs in your operation, start by mapping where your time is going. We run AI literacy sessions for small and purpose-driven businesses that want to adopt AI with judgment rather than pressure. Book a session, or email us your biggest operational time drain.

When you register a domain, the host offers you a one-click default: a stack that works, ships today, and quietly runs on whatever energy the grid happens to supply. We had spent months building a methodology for sustainable digital design. Standing at our own checkout, we noticed how easy it would be to ignore all of it. The website nobody briefs properly is usually the agency’s own.

The website nobody briefs properly

Agencies are good at briefs for other people. A client arrives, you set targets, you measure against them, and you hold the work to a standard before it ships. Then the agency’s own website goes up on whatever was fastest, because nobody wrote the brief. It is an easy pattern to fall into, and we were one click away from it ourselves.

So we did the obvious, slightly uncomfortable thing. We treated ourselves as a client: a written brief, two numeric targets, and a deadline. If a sustainable-design methodology is real, the first place it should show up is your own infrastructure. A sustainability agency running its front door on a default stack is not a small irony. It is the whole credibility question.

A website’s footprint is a real number

Most conversations about sustainable websites stall because the thing being discussed feels abstract. It is not. A web page has a measurable carbon cost per visit, estimated from the data it transfers, the energy that moves and renders it, and how that energy is generated. You can put a number on it in about thirty seconds.

The average web page produces about 0.36g of CO2 per view, according to the Website Carbon Calculator. Most organisations have never checked theirs. Most agencies have never offered to check a client’s. That gap is both the opportunity and the accountability, because a number you can measure is a number you can be held to. “More sustainable” is a feeling. “Under 0.1g per visit, measured on this date” is a claim. We wanted the second kind.

The brief we set ourselves

We wrote two numbers down before touching a line of code.

The sustainability target was under 0.1g of CO2 per page visit. We picked it against an external benchmark rather than a number that flattered us. The average web page produces about 0.36g, and the most efficient sites the Website Carbon Calculator showcases sit around 0.06 to 0.09g. We wanted to be in that second group, not just below average, so we set the bar at under 0.1g: more than three times cleaner than the typical page, and within reach of the cleanest sites on the web. A looser target would have proved we could beat an average. It would not have proved we could build to the standard we actually care about.

The performance target was a PageSpeed score of 90 or above on both mobile and desktop. Ninety is the line where PageSpeed rates a page “good,” so the target sits at a meaningful threshold rather than an arbitrary one. We required mobile as well as desktop on purpose, because mobile is the harder test, and the easy cheat is to quote the desktop score and stay quiet about mobile.

Two numbers, both pinned to an external standard. That is the part that makes a target honest: you do not get to move it after you see the result.

Hosting was the decision with the widest reach

We made the hosting call first, deliberately. Of every decision in a build, the host is the one that changes the energy behind every single request at once. You can compress every image on a page and still serve it from a data centre running on fossil power. Switch to a host running on renewable supply, and you have cleaned the energy under the whole site without shipping a single byte less. Everything after that optimises what remains.

We chose Krystal. Their data centres run on 100% renewable energy, and their flagship UK facility reports a power usage effectiveness of 1.10, which means almost all of that energy reaches the servers rather than the cooling. They have been a certified B Corp (an independently audited social and environmental standard) since early 2023, and their hosting is independently verified as green by the Green Web Foundation. That independent verification is the part that matters: it is a third-party check, not a self-declared green badge. Renewable supply is also a different thing from a carbon offset bought after the fact. One keeps the emission from happening; the other pays to compensate for it once it has. We wanted the first kind.

What going static removed

The second decision was to build the site static. A static site serves pre-built pages straight from the host. There is no server-side rendering on each request, no database query waiting to happen, and no per-visit processing assembling a page that is identical for everyone anyway.

The reasoning mattered more than the technique. Fewer moving parts means less energy per request, fewer things that can break, and a faster response, because there is nothing to compute before the page starts arriving. The constraint is real and worth naming: static makes genuinely dynamic, per-user content harder, so anything dynamic has to earn its place rather than arrive by default. For a marketing site, almost nothing needed it.

Optimising the remainder

Hosting and architecture set the ceiling. The rest is discipline about what each page actually ships.

Images are served as WebP, compressed to a budget, with a hard cap per file. A single unchecked hero image can outweigh the rest of a page combined, so every image had to justify its bytes. Fonts load without the blocking waterfall that holds up rendering while a webfont downloads, so text appears immediately rather than waiting on a font request. JavaScript is kept to what the page genuinely needs, not a framework shipped out of habit. Less script is less to download, less to parse, and less to run on the visitor’s device, which is where a large share of a page’s real-world energy is actually spent.

None of these are exotic moves. They are the boring decisions that compound, and most sites skip them because nobody set a budget that made them necessary.

How it actually went

Most of the build was straightforward, because the sustainable choice and the practical choice were the same choice. Static is cheaper to host and faster to serve. Lean images load quicker and cost less bandwidth. Renewable hosting was a supplier decision, not a redesign. For most of the work, constraint and sustainability pointed the same way.

They did not always agree. The original plan for the hero section included a looped video of leaves moving in a soft wind, the kind of detail that makes a site feel alive. A short looped clip, even well-compressed, typically runs to several megabytes. Against our 138KB hero image that is a weight difference of roughly twenty to one, and unlike an image, a video requires continuous processing on the visitor’s device to decode and render each frame. The carbon cost does not land once on load; it accumulates for as long as the page is open. We cut it. The static image is what we actually built to.

The first performance pass was a failure, and a useful one. An early mobile test scored 67, dragged down by blocking scripts. Fixing that single issue moved it into the 90s. The number told us exactly where the waste was, which is the entire argument for measuring in the first place.

The scorecard

• CO2 per visit: 0.03g, measured 25 June 2026 on the live public site with the Sustainable Web Design model (version 4), the same model the Website Carbon Calculator uses. Target: under 0.1g, cleared with room to spare.
• PageSpeed, mobile: 98. Desktop: 100. Measured 25 June 2026 on the live public site with Lighthouse, the engine behind PageSpeed Insights. Target: 90+ on both, met.
• Cleaner than 95% of pages tested, rating A+, per the Website Carbon Calculator, measured 25 June 2026 on the live public site.
• Hosting: Krystal, 100% renewable energy, certified B Corp. Verifiable today, not on launch.

In development testing the build already met the performance target, scoring between 94 and 97 on mobile across repeated runs, with accessibility, best practices, and search engine optimisation (SEO) at 100. The carbon and performance figures above are measured on the live public site, including the Website Carbon “cleaner than” percentile, published the same way we would measure a client’s site: in public, on the same tool. We held our own front door to the standard we intend to work to. The scorecard is how you check whether we meant it.

Want to know your own site’s number? Run it through the Website Carbon Calculator. It takes about thirty seconds, and if the result surprises you, that is usually where the brief should start. If you would like help reading what it means for your stack, book a carbon check and we will walk through it with you.

In late May 2026 I was told, plainly and without any drama, that I had been replaced by artificial intelligence (AI). Not as an insult. As a description. The kind of interface work I used to be paid for is now generated by tools like Google Stitch, and clients increasingly build their own projects rather than pay a designer to do it. My most recent contracts had quietly dried up earlier that spring; this was simply the reason. Whoever delivered the news was right, and I already knew it, because I had started doing the same thing myself. What surprised me was the feeling. Not fear. Relief.

It started at sixteen

I was ready for that conversation because the idea behind The Trinity Tribe had been forming for six years. Most of that time it lived in my head and in a sprawling Notion workspace, waiting. I was waiting for the idea to ripen, and waiting for myself: to mature as a designer, to get the user experience and interface (UX/UI) craft properly into my hands, to learn from the agencies and clients I worked with what people actually need and where the real gap sits. You cannot design a different way of working until you understand the current one well enough to see what it costs.

The seed is older still. At sixteen I watched a documentary about Jacque Fresco and The Venus Project, and it rearranged something. The world runs the way it does, it argued, because people designed it that way. The economy, the cities, the incentives: none of it is weather. Someone made it, which means someone could make it differently. I never got that thought back out of my head. It went from a teenager’s conviction, through vegan philosophy, to solarpunk, and finally to a way of working: building digital products that tell the truth about what they cost.

Why I waited, and what finally moved me

What I did not have was a way to start. I knew one thing with total clarity: you cannot run an agency alone. The work is too broad, the surface area too large for one person to hold. So for years I looked for a co-founder. A handful of people toyed with the idea alongside me, but each drifted off once it became clear that co-founding is work, not a title, and that the effort starts long before there is anything to show for it. I kept doing good work for other people while the thing I wanted to build sat in Notion, unbuilt. One more person felt like the only door in, and it stayed shut.

Then two things matured at the same time: the idea, and the tools. While the industry was using AI to remove designers from the budget, I was using it to do the opposite. I had started replacing my Figma work with Claude Code, going deeper into automation, building the operating system that now runs The Trinity Tribe behind the scenes. AI did not take the work away from me so much as hand it back at a different scale. It gave one person enough reach to start something that used to require a team.

So when I was finally told the old way was over, it did not land as a loss. It landed as permission. I am fairly sure I would not have made the call on my own. The push came from outside, the readiness from years inside, and the two met this spring.

Why I didn’t want it to be mine

Here is the part that matters most, and it is the least obvious. Once AI made it possible to start alone, I had to decide what kind of thing to build. The easy version is a studio that is a vehicle for one person’s taste: my eye, my vision, my name on the door. I deliberately did not build that.

Part of the reason is plain honesty about my own limits. I am still learning, and I expect to be learning for the rest of my life. I am not the most experienced person in the room on design, and I am further from it on engineering, marketing, management, and half the other things a studio has to be good at. A practice built to depend on one person’s knowledge inherits the edges of that knowledge. A practice built around a community can become good at far more than any single founder ever could.

There is a structural version of the same point. A studio shaped entirely around its founder is a monoculture: one point of view, one set of blind spots, one person who has to be in the room for anything to happen. That is the same fragile pattern the work itself argues against, systems that depend on a single owner and cannot be maintained once that person leaves. It would be strange to sell resilience and run a business with none.

So The Trinity Tribe is designed, from the first day, to grow past me. The name is the plan. Trinity is not three people; it is three values the work holds in common. Tribe is whoever comes to share them. The real answer to “you cannot do this alone” was never a single co-founder. It was a community: a shared view that surfaces from many people rather than direction handed down from one.

What it looks like when it’s real

Fully grown, a community-driven agency does not look like a studio with a big mailing list. It looks like a tribe doing the work: collaborators, clients and readers who hold the three values and shape where the practice goes. Direction comes from the edges rather than from a founder’s desk. The studio’s sense of what it is thickens as the community grows, the way a forest’s underground network gets richer with every tree it connects. There is no centre to capture, and in time no single person to follow, which is the point.

I want to be honest about where this sits on launch day: almost nowhere. The site you are reading this on is new. The tribe is mostly empty chairs. I am one person with an operating system, a clear philosophy, and a public invitation. That gap between the idea and the reality is not something to hide. It is the work. A community you could design into existence on day one would not be worth belonging to.

What work owes the world

Underneath all of this is one belief about technology, and it is easy to get wrong. Technology and nature are not opposites. Technology is the living system continuing through a new medium, the same process that grew forests and nervous systems, now running through us. Being replaced by AI did not shake that belief. It confirmed it.

But the belief has a shadow that has to be named in the same breath: cancer is also natural growth. Growth that consumes its host until both die is completely natural and completely fatal. So “technology is nature’s next step” is only half a sentence. The full version is that it stays nature’s next step only while it stays in harmony, and the entire job is telling the difference. A carbon audit is that test with a number attached, the same act as refusing to make a green claim you cannot defend (a subject with its own deadlines and legal teeth). How we build to that standard is a longer story; what matters here is the belief underneath. Work should relate to the world like an organism that gives back more than it takes, stays legible to the people who depend on it, and does not quietly make them dependent on it.

What I’d call success

So I will not measure this in revenue, or headcount, or how widely my own name travels. Those are easy to grow, and easy to grow badly. Success, for The Trinity Tribe, looks like two things. The smaller one: a studio its community genuinely owns, where I am one node rather than the centre, and the practice would carry on if I stepped back. The larger one: a world where “what does this cost the living system?” is an ordinary question, asked by default, by people who never heard of us.

If The Trinity Tribe works, the strangest sign of it will be this. At some point the studio will stop feeling like mine. That is not the risk of the plan. It is the plan.

If this reads like something you would want to be part of rather than a service you would buy, that is the right instinct. We are about to open a community, on Discord first and then on a home of our own. The announcement goes out first to the people already following along: find The Trinity Tribe on LinkedIn or Bluesky and you will not miss it.

Solarpunk is most often described as a visual aesthetic: soft yellow afternoons, vertical gardens, art-nouveau curves wrapped around clean technology. The look is real, and it is having a moment. But solarpunk is also a movement and a philosophy, and the visual is the surface of a much older idea. The idea is this: technology and nature do not have to be in opposition. They can be designed to work together.

That belief is why we use the word at The Trinity Tribe (TTT). Not because of the aesthetic. Because the philosophy gives us a way to think about digital products that “sustainable” no longer does.

Why the word matters

“Sustainable” has been worked to death. So has “green.” Most agencies that use these labels mean the same handful of things: we recycle the printer paper, we offset the flights, we put a leaf icon on the homepage. The terminology has been emptied of meaning, and the original ideas inside it have leaked out.

Solarpunk could go the same way. The aesthetic is easy to copy. You can already see the word attached to any product that is vaguely future-positive and has a plant in the mood board. If that becomes all it means, the politics and the design philosophy underneath will quietly disappear, and we will be back to looking for the next word.

So this article is a small attempt at protection: a written-down version of what the word actually means, so that we (and anyone reading) can remember.

Where solarpunk comes from

Solarpunk is older than its 2010s breakthrough. Its intellectual roots run through four distinct currents. Each of them is worth more than the short summary we can give it here. For now, the headline version of each:

Permaculture. In the mid-1970s, Bill Mollison and David Holmgren co-developed a design framework based on observing how natural systems already work. The twelve principles most people quote (observe before you act, produce no waste, use small slow solutions, value the marginal) came later: Holmgren set them out in 2002, in Permaculture: Principles and Pathways Beyond Sustainability. They treat nature as a methodology, not a metaphor, and they travel well from agriculture into digital design.

Social ecology. Murray Bookchin spent the 1960s through the 1990s arguing that ecological crisis is not, fundamentally, a technological problem. It is a social one. The way we organize work, ownership, and authority shapes what the technology can even be. A truly solarpunk practice is not only about reducing carbon. It is about who owns the data, who controls the platform, and who can maintain the system once the original team is gone.

Speculative fiction. Ursula K. Le Guin mapped the literary territory long before “solarpunk” was a word. The Dispossessed and Always Coming Home imagined futures that are beautiful precisely because they negotiate scarcity collectively. The word itself was coined in a 2008 blog post, “From Steampunk to Solarpunk”; the visual language took shape on Tumblr around 2014; and Sunvault (2017) became the first anthology to collect the fiction at scale, giving the movement a canon rather than its name. The fiction added hope without naivety.

Natural systems as design models. Empirical observations from working ecology have changed how we understand cooperation, resilience, and resource sharing. Suzanne Simard’s research on mycorrhizal networks (the underground fungal connections between trees) and Elinor Ostrom’s Nobel-winning work on the commons both suggest the same thing. Long-lived complex systems run on redundancy and shared resources, not on monoculture and isolated efficiency.

Each of these deserves its own article. We will write them.

Green, sustainable, regenerative

Underneath all four currents is one distinction that does most of the work. Three words sit on a spectrum:

• Green: do less harm.
• Sustainable: maintain the system as it is. Do not make it worse.
• Regenerative: leave the system better than you found it.

Most “sustainable” digital work sits at the first level. The conversation is about reducing carbon, page weight, server load. That is real, and it matters. But it accepts the baseline as given.

A regenerative practice asks a different question: can this product, by existing, make the surrounding system healthier than it was before? That is a higher bar. It is also the bar that separates solarpunk from green-with-a-vibe.

What this means at TTT

So far this has been ideas. Here is the operational part.

We chose the word solarpunk because it names a philosophy, not a feeling. That philosophy shows up in our work in three concrete ways.

We design for independence. A practice rooted in permaculture-style thinking does not make clients structurally dependent on infrastructure they do not control, including the artificial intelligence (AI) services a product relies on. The point is not to be anti-AI. It is that if a service degrades, changes its pricing, or disappears, the client should stay functional rather than stranded. Independence is a design value, not a marketing one.

We design to last. The websites we have audited tend to be two to five years old, and most were already showing signs of unmaintainable accumulation by year two. Our work assumes a five-to-ten year horizon and is built so that whoever inherits the codebase can still understand it. That assumption changes a lot of small decisions, including which frameworks we do not reach for.

We say no to projects that do not fit. Not often, but often enough to notice. If a brief is essentially a request to apply solarpunk styling to a product whose underlying logic is extractive, the styling does the opposite of what the philosophy intends. It launders the product. We are not interested in that work, and not pretending to be.

We wrote this down so that everyone, including us, can remember what the word actually means.

The next time a digital product calls itself sustainable, ask which level it is working at: green (less harm) or regenerative (more health). Those are different products, built on different beliefs. If that distinction matters to you, email us and tell us which level your own site is operating at, and we will tell you honestly. Knowing the difference is most of the work.

Picture a business owner reading their own website footer in the summer of 2026. It says the company is eco-friendly, sustainable, and carbon neutral. Now ask them to prove it: the measured numbers, the certification, the evidence an enforcement body would accept. Most could not produce a single document. That gap has always existed. What is new is that, from a fixed date this autumn, the gap is no longer just embarrassing. In the EU, it is unlawful.

The internet runs on green claims nobody checks

Open the footer of almost any small business website and you will find a sustainability claim. Eco-friendly. Green. Environmentally conscious. Carbon neutral. These words appear on agency homepages, product pages, and email signatures across thousands of sites, usually with nothing behind them: no measurement, no certification, nothing to check.

For years this was tolerated. A vague green claim was, at worst, a reputational risk. If a journalist or a sharp customer pushed on it, the company might face an awkward conversation, but rarely a consequence. The claim cost nothing to make and almost nothing to leave unproven, so it spread. Sustainability became a default adjective, applied the way “premium” or “trusted” get applied, as tone rather than fact.

We have always found this uncomfortable, and not only on principle. A market where everyone claims to be green and nobody has to prove it is a market where the businesses doing the actual work get no credit for it. The claims become noise, and noise drowns signal.

The rule changed, and not the way the headlines said

In 2025, a lot of coverage announced that EU green-claims regulation was collapsing. That coverage was half right, and the half it got wrong is the half that matters.

Here is the accurate version. There are two separate instruments, and they are easy to confuse.

The first is the Green Claims Directive, a proposal known by its file number COM(2023) 166. It would have required businesses to substantiate every environmental claim with scientific evidence and have it verified by an independent third party before publication. In June 2025 the European Commission announced its intention to withdraw this proposal, and its future is genuinely uncertain. This is the directive the headlines were reporting on.

The second is the Empowering Consumers for the Green Transition Directive, Directive (EU) 2024/825. This one is not a proposal. It was adopted in early 2024, it amends existing EU consumer-protection law, and it carries the ban on vague and misleading green claims. Member states must write it into national law by 27 March 2026, and it applies from 27 September 2026.

So the substantiation-and-verification machinery is on hold. The ban on unprovable green marketing is law, with a date attached. Treating “the Green Claims Directive is dead” as an all-clear is the single most expensive misreading a business could make this year.

What the Empowering Consumers Directive actually bans

The directive does three things that bear directly on marketing copy.

First, it bans generic environmental claims. If you describe your business or product as “eco-friendly,” “green,” “environmentally friendly,” or “climate friendly,” and you cannot demonstrate recognised excellent environmental performance, the claim is prohibited. Vague is the problem. Specific and evidenced is the way through.

Second, it bans carbon-neutral claims based on offsetting. Terms like “carbon neutral,” “climate neutral,” and “net zero” are not allowed in consumer marketing when the basis is buying offsets rather than cutting emissions in your own value chain (your operations and supply chain). A business can advertise a reduced climate impact only if it has actually reduced that impact. Offsets do not count toward the claim.

Third, it restricts sustainability labels. A label or badge may only be used if it rests on a certification scheme with independent third-party monitoring, or is established by a public authority. Self-issued badges are out.

Who is in scope is broad. The rules apply to commercial communication aimed at EU consumers, which includes any website a consumer in the EU can read. Penalties are set by each member state, and the framework allows fines reaching up to 4 percent of annual turnover in coordinated cross-border cases. The exact figure depends on where enforcement happens, but the direction is clear: this is consumer-protection law, with consumer-protection teeth.

Why this lands on websites specifically

It is easy to read “consumer marketing” and picture packaging and television advertising. For most businesses, the exposure is closer to home. It is on the website.

Website copy, landing pages, email footers, product descriptions, and the small green badges in the corner of a checkout page are all commercial communications. They are exactly the surface the directive governs. A claim does not have to be printed on a physical product to be in scope.

Two ordinary examples show how this works. A service business has a homepage line reading “we are a sustainable, eco-friendly company.” There is no certification behind it and no measurement, so under the new rules it is a generic claim that cannot stand. Second, a digital agency advertises “carbon-neutral websites” on the strength of an offsetting arrangement. That is precisely the offset-based neutrality claim the directive prohibits.

The second example carries a complication worth naming. When an agency makes a sustainability claim on a client’s behalf, on the client’s website, the claim sits on the client’s domain. If it fails the test, the business hosting the claim is the one exposed, not only the agency that wrote it. Plenty of businesses are carrying claims they did not write and have never checked.

What the Green Claims Directive would have added

The stalled proposal still tells you where the direction of travel points.

The Green Claims Directive would have moved from banning bad claims to actively policing good ones. Every explicit environmental claim would have required substantiation grounded in widely recognised scientific evidence, assessed across a product’s life cycle, with a verified methodology for any comparative claim such as “greener than the alternative.” An accredited independent verifier would have had to check the claim before it was published, not after a complaint.

That regime is on hold, and may not arrive in its proposed form. But notice what the Empowering Consumers Directive already does without it. The hard work, removing claims that cannot be evidenced, is done. The verification layer would have raised the bar for the claims that remain. A business that reads the withdrawal as permission to keep its unprovable copy has misunderstood which directive was doing the work.

What a defensible approach looks like

Strip away the regulation and a simple test remains. Could you hand an enforcement body your working? If the honest answer is no, the claim is not ready to publish.

A claim you can defend has four properties. It rests on a measured number, the actual carbon output of the site, produced by a tool rather than estimation. It runs on hosting you can verify is powered by renewable energy, which is a different and stronger thing than hosting that has been offset to nominal neutrality. It comes with a quantified outcome, a before and an after, rather than an adjective. And it carries a methodology you can cite, so a regulator can see how the number was reached.

None of this is exotic. It is ordinary measurement applied to a domain where measurement has been rare. When we rebuilt our own site, we treated it as a client project and measured it the same way, precisely so that every claim on it could be backed with a number. The point is not that this is hard. The point is that most green marketing has simply never been asked to do it. From this autumn, in the EU, it will be.

The practical next step

If you do one thing after reading this, make it a measurement, not a rewrite. Run your live website through a public carbon-measurement tool to see what it actually emits per visit. Then read your hosting provider’s energy documentation and look for the specific claim: is the electricity renewable at source, or is “neutral” being carried by offsets? Those two facts are the start of knowing what you can honestly say.

From there, the job is an inventory: every environmental word on the site, sorted into what you can prove, what is generic, and what rests on offsets. The first you keep. The second and third you fix or remove before 27 September 2026.

Start with what you can actually measure

Measure your site and list every environmental claim on it before 27 September 2026. If you would like a second pair of eyes on what has to change, book a short audit conversation or drop us an email. The point is not to sell you certainty. It is that the deadline is real.