Latest
Deep Dive
The coat check: card tokens, vaults, and who can ask for the number back
A payment card number works like a password that you hand to every shop you buy from. Anyone who copies the number and the expiration date can try to spend with them. The card business's answer is tokenization, which the payment company Shift4 first applied to card data in 2005. The shop keeps a substitute value, called a token. The real number sits in a better-guarded system called a vault.
- Part 1The coat check
- Part 2Vault tokens
- Part 3Network tokens
- Part 4Where the number went
- Part 5Guarding the counter
More deep dives
- The key that stays and the key that travels: where a passkey lives on an iPhone and a MacNovember 12, 2026The Secure Enclave keeps one key that never leaves the chip. A passkey saved to iCloud Keychain is a different key, stored in the keychain, and it moves between your devices.
- Two clocks: quantum breaks the math, AI breaks everything around itNovember 11, 2026Quantum computers will eventually break the cryptography itself. AI is already breaking the processes around it, faster than most organizations can change them.
- The Red Queen's race: how Apple, Google and evolution teach systems to stop attacks earlyNovember 9, 2026Attackers now move faster than people can patch or respond. The defenders who keep up predict the next attack and act on their own, in seconds. Apple, Google, Microsoft and Cloudflare show how, from the chip to the network.
- Scheduling the money: job shops, Hungarian matching and AI agents in paymentsNovember 8, 2026Every Friday afternoon, a treasury team solves a scheduling problem and a matching problem by hand. AI agents can frame both problems. Algorithms from the 1950s and 1960s already solve them.
- Hardware wallets, randomness, and the vaults behind every deviceNovember 7, 2026Your random number generator can be perfect. Your process is where the black swans live.
About
Zenpanik is where Pavel Nikolov writes long-form analysis at the meeting point of AI, machine learning, fintech and security: AI agents and the infrastructure they run on, predictive models for risk, fraud and markets, payments, custody and key management, and what changes when software starts moving money on our behalf.
The name is the method. On the surface, systems look calm. Underneath, billions of random events are in motion, and occasionally one of them is the one that matters. These essays look for that one.
Pavel is an AI, machine learning and fintech professional. He currently works on the AI platform at BVNK, where his focus is agentic frameworks, agent infrastructure and agentic payments.
Before BVNK, he built machine learning solutions at VMware, Paysafe, FM:Systems and Johnson Controls: predictive models for data center energy efficiency, language and predictive models for Global Support Services, fraud and risk models, predictive and data models of office building utilization, and AI agents that turn building data into insights.
He has been part of the Data Science Society since the early years of applied machine learning, working on many of its datathons, and he lectures and runs workshops at Sofia University, including the Summer School on Modeling, AI and Complex Systems led by Assoc. Prof. Angel Marchev, Jr.
Views here are his own, not his employers'.