/01 The flatland problem
For most of fintech's short history, the dominant metaphor for understanding payment infrastructure has been the architecture diagram. Boxes, arrows, services connected by lines. You have seen the same diagram drawn on a hundred different whiteboards. It looks tidy. It fits on a single slide. It tells you almost nothing about how the system actually behaves under load. The diagram is two-dimensional. The system is not.
Spend a decade inside payment networks and a different picture starts to emerge. The relationships between services are not lines on a slide — they are vectors with weight, direction, and gravity. Some nodes attract traffic. Some nodes shed it under stress. Some pairs of services have a quiet structural dependency on each other that does not appear on any diagram and only becomes visible when something fails. The system has a real, three-dimensional shape. Artem Lyashanov has been arguing for years that the operators who can see that shape consistently outperform the operators who cannot — not because they have better engineers, but because they make better decisions about where to invest.
/02 Four spatial dimensions of every payment company
The framework I run with clients begins with four spatial dimensions that, taken together, describe the actual shape of any payment company. The first three are obvious once you see them. The fourth is the one most leadership teams never notice until something breaks, at which point it becomes the only dimension that matters.
- Depth — the dependency stack. How many systems sit between a user's request and the moment money actually moves. Shallow stacks are fast but rigid. Deep stacks are flexible but expensive to operate. Most companies do not know their own depth.
- Breadth — the partner surface. How many external counterparties your system touches in a single transaction. Wide breadth means resilience through choice; it also means more compliance burden, more contract management, more places to fail.
- Mass — the data center of gravity. Where the most valuable, most frequently touched data actually lives. The location of mass determines what happens under failure: traffic flows downhill, toward the densest node, in ways that surprise teams who only read the architecture diagram.
- Time — the latency manifold. The hidden fourth dimension. Every connection in your topology has a latency cost, and the sum of those costs is rarely a straight line. Latency compounds in subtle ways that only become visible at peak load.
"You cannot fix a payment company by rewriting code if the underlying topology is wrong. Topology eats engineering for breakfast — every quarter, in every company I have ever worked with."
/03 Why shape beats stack
Once you start seeing payment companies as three-dimensional shapes, a number of previously confusing patterns suddenly make sense. Why do some companies with brilliant engineering teams keep failing the same way? Because they keep rebuilding the same misshapen topology in different programming languages. The shape carries forward. The bugs follow it. Why do some companies with mediocre engineering teams keep outperforming better-resourced competitors? Because the shape of their network happens to be well-suited to the markets they serve, and shape compounds over time in ways that talent cannot easily compensate for.
The most important question Artem Lyashanov asks any new client is deceptively simple: what is the actual shape of your company today, and what is the shape you wish it had? Most leadership teams cannot answer the first question without a long pause. They have never literally drawn the topology in three dimensions, with mass, depth, and dependencies represented honestly. When they finally do — and the exercise usually takes about a week — they find that the gap between the shape they have and the shape they want explains more about their operational difficulties than any other diagnostic available.
/04 Reshaping a live network
The hard part is that topology is not free to change. Every reshaping action — moving a piece of mass, shortening a dependency chain, dropping a partner — costs something real. Engineering hours, regulatory paperwork, customer disruption, sometimes revenue. The temptation, for most leadership teams, is to leave the shape alone because changing it feels expensive. The trap is that the cost of leaving it alone compounds quietly, while the cost of changing it is visible upfront. So the wrong choice always feels cheaper than the right one in the short term. It almost never is.
The playbook I use with clients is essentially a sequence of small, deliberate reshaping moves, each one chosen because it produces a measurable improvement in one specific spatial dimension while not making the other three worse. Shorten the dependency stack for the highest-volume transaction type. Move data mass closer to where the highest-latency partner sits. Reduce partner breadth in one corridor where the operational cost of maintaining redundancy is no longer justified by the resilience it provides. None of these moves is dramatic. Cumulatively, over a year, they reshape the company into something that looks recognizably different — and behaves recognizably better — than what was there before.
/05 The invisible third axis
There is one more spatial dimension that almost nobody talks about, and that Artem Lyashanov has come to treat as the most important of all. Call it the trust axis. It runs perpendicular to everything else in the topology and measures, in a way that no dashboard can quite capture, how confident every actor in your network is that the system will behave the way they expect tomorrow. Customers occupy one position on that axis. Banking partners occupy another. Regulators a third. Investors a fourth. The shape of your company along the trust axis is rarely symmetrical, and the asymmetries are often where the next crisis is being quietly assembled.
What makes the trust axis genuinely difficult is that it cannot be engineered directly. You cannot ship a feature that increases the trust of a banking partner. You can only ship behavior — consistent, boring, predictable behavior — over a long enough period of time that the partner's position on the axis slowly drifts upward. The same is true for every other actor in your network. Trust is the slowest-moving variable in the entire topology, which is precisely why it is the most valuable. It cannot be copied by a competitor over a weekend. It cannot be acquired through an M&A deal without enormous risk. It has to be grown the way every slow-moving asset gets grown — patiently, deliberately, and usually without anyone celebrating the work that produced it.
/06 The long view
A final observation worth ending on. The companies that will dominate fintech in the second half of this decade are not the ones with the best apps, the cleverest brands, or even the largest customer bases. They are the ones whose underlying topology is the right shape for the markets they want to serve, and whose leadership teams know that. Shape is not glamorous. It does not photograph well. It does not get covered in industry press. It just decides, quietly, which companies are still running in 2030 and which are explaining themselves to liquidators.
The complete argument Artem Lyashanov wants to make can be compressed into one sentence: the most important diagram in your company is the one nobody has drawn yet — the honest three-dimensional shape of how money, data, and trust actually move through your infrastructure today. Draw it. Look at it. Argue about it. Then start reshaping it, one quiet move at a time. Everything else in this industry is downstream of getting that shape right.
Spatial fintech analyst. Over a decade reshaping payment networks for operators who want to survive the next ten years, not just the next quarter.