Trade Your Double Diamonds for Steel
AI does not compress the design process. It alloys discovery and delivery into a material neither could become alone
In 1976, Whitfield Diffie and Martin Hellman were trying to solve a strange modern problem: how do two people establish a shared secret in public, in full view of everyone, without ever having met before and without ever handing each other the secret itself?
For most of human history, secrecy depended on transport. You trusted a courier, a sealed letter, a guarded room, a diplomatic pouch, a chain of custody. Secret things stayed secret because you moved them carefully. But in a networked world, that model breaks. You need a way to create trust without carriage. A way to produce a secret without delivering one.
Two people sit across from each other at a table with jars of paint. They each start with the same public color: yellow. Alice pours her secret blue into the yellow and gets green. Bob pours his secret red into the yellow and gets orange. They slide their mixed jars across the table. Now Alice adds her secret blue to Bob’s orange. Bob adds his secret red to Alice’s green
They both arrive at the same muddy brown. The exact same shade. Neither one ever revealed their secret color, and yet they are holding identical results. The secret colors are still secret. The shared color is right there on the table.
If you have not seen this demonstration, watch it. It takes ninety seconds and it will change how you think about one-way functions forever. The demonstration is one of the most famous in cryptography. It describes the Diffie-Hellman key exchange, and the reason millions of people have watched it on YouTube is that it makes something mathematically elegant feel physically obvious. Of course the colors converge. You can see it happen. The math works because the paint works, and the paint works because mixing is a one-way operation.
This is my real debit card, put through shasum -a 256, there is no way to reverse it:
9219e0affa4b86cdfaed7927c91ce7600aadc4f38079c5a4aa0667e496700bf0But the part that matters is not the secrecy. It is the irreversibility. Once mixed, the colors cannot be unmixed. You cannot look at the brown and extract the blue. You cannot reverse-engineer the red from the final shade. The system moves forward into a new color space, and there is no going back.
Hold that thought.
Two Buckets of Paint
The Double Diamond has been the dominant model for the design process for nearly two decades. And for most of that time, it has been a genuinely useful one. Two diamonds, two phases: discover and define (the first diamond), then develop and deliver (the second). Blue bucket, red bucket. Think, then build.
The logic was sound. Execution was slow and expensive. Writing code took months. Building prototypes took weeks. Testing required recruiting, scheduling, facilitation, synthesis, and then a second round of all four because the first round surfaced more questions than answers. You could not afford to build in order to think. Building was the scarce resource, and you do not waste scarce resources on exploration. So you separated thinking from building, the same way you separate prep from cooking when ingredients are scarce. You think first, carefully, because every trip to the stove costs real time and real money.
And the industry optimized around that separation. Entire methodologies grew up to manage the handoff between the two diamonds: the translation layer where research becomes requirements, where insights become specifications, where understanding becomes something an engineer can act on. Design systems, sprint rituals, story mapping, journey-to-backlog pipelines, specification documents that exist primarily to survive the crossing from one diamond to the other. All of it designed to move work cleanly from the blue bucket to the red one without losing too much fidelity in transit.
The Double Diamond was not wrong. It was a correct description of a constrained system. And the industry got remarkably good at describing, optimizing, and certifying the mechanics of that constraint. An entire consulting economy emerged to service it, selling tighter diamonds, cleaner handoffs, more efficient phase transitions, better ways to manage the boundary between thinking and building. The boundary itself was never questioned. It was treated as a law of nature, not a constraint of cost.
That made sense when execution was the bottleneck. Execution is no longer the bottleneck.
The Colors Mix
When you can prototype in minutes instead of weeks, something fundamental shifts. Building stops being the expensive second phase and starts being part of thinking itself.
Prototyping is discovery. You build something rough, and the act of building reveals what you did not understand.
Testing is research. You ship a small change, and the data tells you what question to ask next.
Generating variations is exploration. You produce ten options in the time it used to take to produce one, and the comparison itself becomes the insight.
You do not need to theorize about which layout performs better when you can build both and watch real users choose.
The old sequence was sequential by necessity:
Before: Think in blue. Translate. Build in red. The handoff is the bottleneck.
Now: Think-build. Build-think. The colors bleed into each other.
The new sequence is not a sequence at all. What results is not a small efficiency gain. It is a structural collapse. The barrier between the two diamonds dissolves, and what emerges is not a faster version of the old process. It is a fundamentally different medium. Discovery and delivery mix continuously. Not blue, then red. Blue and red, together, producing a purple space where both happen at once.
The Diffie-Hellman property holds here, too. Once mixed, you cannot cleanly separate them again. You cannot point to a moment in the work and say “this is where discovery ended and delivery began.” The colors have converged. The system operates in a new color space, and there is no returning to the old separation (even if you wanted to).
But paint is still the wrong metaphor.
Paint Is the Wrong Material
Accept the paint for a moment. It is intuitive. Everyone has mixed colors. Everyone understands that you cannot unmix them (at least not without industrial chemistry and a lot of patience). The Diffie-Hellman analogy works beautifully as an entry point.
But paint is soft. Temporary. Imprecise. Paint mixing is a description of the process: what happens when you combine two things. It does not describe what you get.
Mixed paint is still paint. It has the same properties as its components, just a different color. Purple paint does not gain capabilities that blue and red lacked. It does not become stronger, harder, or more resilient under stress. It just looks different. The combination is additive, not transformative. If you are being honest about the chemistry, paint can be separated with enough effort. Centrifuges exist. Chromatography exists. The fusion is visual, not structural. The material did not change. Only the appearance did.
What is actually happening with discovery and delivery in the AI era is more permanent than that. The result is not a different color of the same material. The result has structural properties that neither element had alone. The combination does not just look different. It behaves differently. It bears weight. It holds an edge. It does things neither component could do.
We need a different material entirely. Not a color metaphor. A material metaphor. One where the combination produces something structurally new, something with emergent properties, something irreversible at the atomic level.
Iron is useful but soft. Carbon is essential but brittle. Neither alone does what you need.
Iron and Carbon
Pure iron is soft. That surprises people, but it is true. Iron in its elemental form is ductile and malleable. You can shape it, bend it, work it with relative ease. But it does not hold an edge. It deforms under sustained pressure. It yields when you need it to stand.
Carbon, on the other hand, is hard but brittle. Think graphite: pure carbon, and it flakes apart in your hand. Think diamond: also pure carbon, and it shatters under impact despite being the hardest natural material on Earth. Carbon resists deformation absolutely, but it does not flex. It breaks.
Now add a small amount of carbon to iron under extreme heat (we are talking about 0.2 to 2.1 percent by weight, no more). Something remarkable happens at the atomic level. The carbon atoms do not just sit next to the iron atoms like pigments floating side by side in a jar of paint. They migrate into the interstitial sites: the tiny gaps between atoms in the iron crystal lattice. Once there, they lock the lattice into a harder, stronger, more resilient configuration.
The iron atoms can no longer slide past each other easily (which is what made pure iron soft in the first place). The carbon prevents the slippage. It is not filling the gaps so much as restructuring what the gaps can do. Different carbon percentages produce different grades: low-carbon steel is mild and flexible, high-carbon steel is hard and holds an edge. The precise ratio matters. But what matters more is that every ratio produces a material that did not exist before the heat.
The result is steel. Not iron-plus-carbon. Not a mixture. An alloy. A material with properties neither element possessed:
Stronger than iron
Less brittle than carbon
More resilient under stress
Structurally useful in ways neither component is alone
Iron is useful but soft. Carbon is essential but brittle. Together, under heat, they become something neither could be alone. This is what is most important to conceptualize: you cannot un-alloy steel.
The carbon, the iron, they are not gone, but you cannot heat steel back into its constituent elements. The transformation is permanent. The lattice is a new thing. The atoms have reorganized, and the material that emerged from the forge is not a combination of its inputs. It is a creation.
It is no longer synthesis of parts into one.
It is genesis.
The New Material of Work
Seen from a long historical distance, process models can look permanent right up until they disappear.
Case in point: The Pony Express lasted about eighteen months. That is all. A technology iconic enough to symbolize an era did not even survive long enough to become normal. It existed in the gap between one infrastructure and another.
Of course, just as the average pony couldn’t imagine the telegraph, design as we know it may turn out to be closer to that than we think.
In the historic lens, the “design” we do is barely a blip. The gap between widespread ironworking and early steelmaking was measured in centuries. The gap between iron tools and modern industrial steel was closer to millennia.
Almost every person who actually developed the discipline of UX/product/digital design is still alive. And like those pony riders and the telegraph, those blacksmiths with carbon, a new factor has mutated our trade:
Iron was discovery. Research, understanding, insight, empathy, synthesis. Useful, absolutely. Essential. But soft. Discovery alone does not ship products. It bends under the pressure of deadlines, stakeholder demands, quarterly targets, and the sheer gravitational pull of organizational inertia. A brilliant research report deforms the moment it hits a sprint planning meeting. Pure insight, with no mechanism for execution, yields under the weight of reality.
Iron bends.
Carbon was delivery. Building, coding, shipping, testing, deploying. Essential but brittle. Pure execution without understanding breaks under stress. It shatters the moment requirements change, the moment users do something unexpected, the moment the market shifts beneath you. A feature built without research is carbon: hard, sharp, and fragile. It cannot absorb new information without cracking.
Carbon cracks.
Heat is AI. The energy that makes the fusion possible. Without sufficient heat, iron and carbon just sit next to each other. That was the old model: two diamonds, side by side, connected by a translation layer but never truly fusing. Two elements in proximity, never in solution. With enough heat, the carbon diffuses into the iron lattice. With AI as the forge environment, building diffuses into thinking. Discovery diffuses into delivery. The atoms reorganize.
Heat transforms.
Steel is the new material of work.
The space where thinking creates artifacts immediately, where artifacts generate insight immediately, where exploration and execution reinforce each other in a continuous lattice. The work has properties neither discovery nor delivery had alone. It is stronger. More resilient. Structurally useful in ways that “research phase” and “build phase” never were.
Steel changes how we build, which changes what we build, which changes us. The pony express didn’t evolve into faster horses. The telegraph was invented and the knowledge of the farrier did not apply.
Collapsing the Model
In a recent conversation with my good friend Jim Kalbach (author of Mapping Experiences), he recently framed the shift as a reversal of proportions: before AI, perhaps 20 percent discovery and 80 percent delivery; after AI, the balance appears to invert. I think even that understates the change. The smaller diamond does not merely shrink while the larger one expands. It moves inside it. They become concentric. Then they alloy. Jim showed me this illustration:
There is a connection here to the Jevons Paradox, which I wrote about in The Last Typesetter. Jevons observed that making a steam engine more fuel-efficient does not reduce fuel consumption. It increases it, because efficiency makes the engine useful in more contexts, which drives more total usage. The same principle applies: making execution more efficient does not reduce the need for design thinking. It increases it.
But Jevons describes the demand shift. The alloy describes the structural shift. Discovery and delivery do not just change proportions. They fuse. They reorganize at the lattice level.
They become steel.
You become steel.
Separate Furnaces
The most common response I see to “AI changes the design process” is the wrong one. But, it is also the most rational and understandable wrong response, which makes it harder to name.
A design team hears the message and does the obvious, sensible, organizational thing: they add AI to each phase separately. AI-powered research sprints in the first diamond. AI-powered development sprints in the second diamond. The two phases remain distinct, but now each one runs faster. The handoff between them is the same. The translation layer is the same. The org chart is the same. The process is the same process, just accelerated.
The team that runs “AI-powered research sprints,” then “AI-powered dev sprints,” then “AI-powered Figma Make sessions,” then “AI-powered agentic coding sprints” is heating iron and carbon in separate furnaces. You get hot iron. You get hot carbon. You do not get steel.
The alloy only forms when the elements share the same heat source. When discovery and delivery happen in the same environment, at the same time, with the same tools, in the same hands. The forge is not two separate workstations with a conveyor belt between them. It is one continuous space where the heat transforms both materials simultaneously, and the transformation happens because they are together. Carbon must be inside the iron lattice. Building must be inside the thinking. That is how the alloy forms.
So, the system protects itself because adding AI power to an existing process makes sense, but more than that, it protects existing understandings of identity. Profession is one of the deepest identity containers most people have. So when the process changes, it does not just threaten workflow. It threatens self-understanding. We were trained to prize diamonds. Now the work demands steel.
The entire cottage industry that grew up around managing the boundary between diamonds is now optimizing a seam that is fusing shut. Tighter sprints. Cleaner handoffs. More efficient phase transitions. All of it amounts to polishing the interface between iron and carbon while the forge melts them together underneath.
You cannot alloy steel in two separate furnaces.
The Forge Is Always On
We started with buckets of paint. The yellow, the blue, the red. The colors sliding across the table. The irreversible convergence on the same shade of brown.
But we have upgraded the material. We started with paint and ended with steel. The metaphor got harder, more permanent, more structurally useful as the argument progressed, and that progression was not decorative. Paint describes the mixing. Steel describes what the mixing produces. And the environment that produces it has a name.
The forge.
The forge is not a phase. It is not something you enter for build sprints and leave for research sprints. It is the environment where the work happens now. The heat is always on. The alloy is always forming. You do not step in and out of it. You work within it, continuously, and the material you produce bears the properties of both elements at once.
The old process assumed separation:
The alloyed process behaves differently:
Or more simply:
Over and over. Not diamonds. Not phases. A continuous forge where the act of building is the act of understanding, and the act of understanding is the act of building. The lattice reinforces itself with every cycle.
I am no metallurgist. But I know what it feels like to work in the alloy, because I have been doing it. Zero Vector is not a methodology for managing phases. It is not a way to make the diamonds spin faster. It is a methodology for operating in the forge, for working in steel, for staying in the space where discovery and delivery are no longer two materials but one.
I will not soften this: operating in the alloy is disorienting. You lose the comfort of knowing what phase you are in. Some days it feels like flying. Some days it feels like drowning.
We are not yet at the age of industrial steel. We are still close to the forge, still learning what this material can bear, still mistaking early tools for final forms. The first steel tools were mostly improved iron tools. It took much longer to understand what steel would make possible.
That is where we are now.
The colors cannot be unmixed. The lattice cannot be undone. The forge is always on, and the material it produces is stronger than anything either element could become alone.
Which means the real transformation is not just in the work. It is in the one who works.
think ↔ make
Alice and Bob are still here. The jars of paint are still here. But look at what happened while we were not watching. They did not just mix colors. They mixed what they knew with what they could build, and each arrived at something neither ingredient could have produced alone. When they slid their jars across the table, they reached the same place without ever exchanging the secret of how they got there.
That is the alloy. Not the act of mixing, but the fact that what you hold at the end is not what you put in at the beginning.
What remains is steel. And the hands that forged it are not the same hands that started.
But AI isn’t the steel; you are.