There's a instructor I keep thinking about...

She teaches six spin classes a week. Has done for years. Her regulars plan their Tuesdays around her — she's built a genuine community inside a 20-bike studio in East London (and beyond online). Brands have reached out. She has a following. And yet the platform she teaches on owns everything: the class recordings, the rider data, the relationship with her students. If she moves studios, she starts from zero. If the platform changes its algorithm, her income changes with it.

She's not unusual. She's the norm.

Peloton has 12 million members and a roster of instructors earning up to $500K a year. Those numbers look impressive until you ask who owns the 100 million-plus classes those instructors have created. The answer isn't the instructors. Meanwhile, rider fitness data — conservative estimates put it at $300-500 per person annually based on acquisition valuations — generates exactly nothing for the people producing it, barring equity incentives. The platforms capture everything; participants usually capture nothing.

This is the problem SpinChain is built to solve. And solving it properly requires building on Sui.

The airline loyalty analogy is a useful one here. Nobody says you're getting "paid to fly" when you earn frequent flyer miles — but those miles are real, they accumulate based on verifiable activity, and they represent a meaningful stake in the ecosystem you're contributing to. SpinChain works the same way. Riders don't get paid to exercise. They earn protocol rewards — SPIN tokens — for hitting verifiable effort milestones, the same way a frequent flyer earns miles for distance travelled.

The difference is that in SpinChain, the instructor earns too. When an instructor deploys a class, alongside the usual revenue from class participants they receive an NFT that contains the revenue logic for that class permanently embedded in it. Ticket sales, secondary trades, sponsor integrations — a programmable split flows back to them automatically, forever, regardless of which platform they're teaching on next year. The class NFT is theirs. The relationship with their riders is theirs, whether the class is real time or being replayed/remixed in future.

This is a "programmable financial event" in practice: a spin class that settles itself.

The obvious question first: why put any of this onchain at all? Strava, Whoop, and Peloton's own app process telemetry on centralised servers in milliseconds for fractions of a cent. No gas fees, no consensus latency, no complexity. If raw performance were the only metric, offchain wins easily.

But offchain systems have a structural ceiling. They can record that you worked hard. They cannot make that record trustlessly verifiable by a sponsor who never met you, composable with DeFi protocols that didn't exist when you signed up, or permanently owned by the instructor who created the class — not licensed to her by a platform that can change the terms next quarter. Offchain apps track your effort. Only onchain makes your effort ownable, verifiable by strangers, and composable with the rest of the financial internet — without trusting the platform in the middle.

The telemetry going onchain is what makes the settlement layer airtight. Which brings the real constraint into focus: how do you do this without the economics becoming absurd?

SpinChain runs on two chains: Sui handles what happens during a class (real-time telemetry, session state, biometric updates), while Avalanche handles what happens after it (reward settlement, ZK proof verification, NFT ownership). You don't use the same infrastructure for a live sports broadcast and the banking transaction that follows it. The same logic applies here.

Indoor cycling generates a genuinely unusual data profile. A single rider in a 60-minute class produces thousands of biometric data points: heart rate, power output, cadence, all streaming at 10Hz. Multiply that across 1,000 concurrent riders and you're looking at millions of individual state updates per minute, in real time.

On most chains, this is economically fatal. Processing 10Hz telemetry across 1,000 simultaneous riders on Ethereum L1 would cost over $7,000 per class — before a single reward is distributed. Even on faster EVM chains the numbers remain prohibitive for the micro-interactions that make fitness apps feel alive: the instant heart rate zone alert, the live leaderboard update when you overtake the rider ahead, the confirmation that you hit your power target in that last interval.

Sui's object-centric architecture changes this entirely. Each rider's session is an independent shared object on Sui — a RideSession — that no other rider ever touches. Because sessions are fully independent objects, Sui's parallel execution engine processes them simultaneously rather than sequentially. No global state bottleneck, no gas wars, no queue. 1,000 riders updating their session objects at the same time creates 1,000 independent parallel operations.

The result: we batch 10 biometric samples per Programmable Transaction Block and achieve 5,000 aggregate TPS across all active riders at a cost of $0.72 per hour-long session. The same session on Avalanche alone would cost $144 per rider. That 99% cost reduction isn't an optimisation — it's what makes the ownership model viable at all. Mysticeti consensus delivers 390ms finality, which is the difference between telemetry that feels live and telemetry that feels like a buffering video stream.

SpinChain goes one step further than helping human instructors. It lets them extend themselves through AI.

The first use case is augmentation: an AI layer that analyses Sui's 10Hz telemetry streams mid-class and surfaces coaching cues in real time. Detecting when the group's average power drops, suggesting a recovery interval, flagging when a rider's heart rate has been in the red zone too long. The human instructor stays in control; the AI handles the data processing no human could do at that speed or scale.

The more ambitious extension — still in prototype — are autonomous agents that can run classes independently. They manage their own schedule, set pricing via bonding curves, and coordinate performance adjustments on Sui in real time, settling earnings on Avalanche. The instructor/developer who train them still earns from every class it teaches.

The access argument remains the same either way. The best spin instructors are expensive and geographically constrained. An AI layer — whether augmenting a human or running independently — can bring that quality of instruction to a rider training at 6am in Lagos or 11pm in Seoul. Autonomy is a spectrum, not a binary; SpinChain lets instructors choose how much of it they want to hand over.

The economic architecture makes this work: the AI agent manages liquidity for its class tokens via Uniswap v4 hooks, adjusting SPIN/USDT pricing automatically based on class demand. When a class fills up and demand spikes, the bonding curve adjusts. When a new time slot opens with low demand, pricing softens to attract riders. This happens without human intervention — a self-sustaining fitness economy.

There's a tension at the centre of any fitness protocol: the data that makes personalised workouts possible is also deeply sensitive health information. Heart rate variability, VO2 max approximations, sustained effort patterns — this is data that insurance companies want, that advertisers want, that you'd probably rather keep private.

SpinChain handles this through zero-knowledge proofs running in your browser before any data leaves your device.

Here's what that means concretely. After a class, you might want to claim a reward for sustaining 85% of max heart rate for 20 minutes. In a traditional system, you'd have to share your actual heart rate data to prove that claim. In SpinChain, you don't. A ZK circuit — built in Noir, running locally on your device in under a second — generates a cryptographic proof that the statement is true without revealing the underlying number. The proof goes on-chain to Avalanche's C-Chain. The raw data never leaves you.

The proof says: this rider met the effort threshold. It doesn't say what their heart rate was. It doesn't say what their fitness level is. It doesn't say anything an insurer or advertiser could use. Selective disclosure, by design — you can share what matters (I worked hard) without sharing what's sensitive (here's my cardiovascular baseline).

Encrypted raw telemetry is stored on Walrus, Sui's decentralised storage layer, where only you hold the keys. The chain knows you worked. Nobody else knows how.

SpinChain is entering Avalanche's Build Games competition — six weeks of intensive building toward a concrete target: 100 riders completing verified workouts, 10 instructors minting their first class NFTs and receiving revenue. These are modest numbers deliberately. Product-market fit in a new category comes from proving the loop works for real people, not from projecting scale.

But the thesis extends well beyond spin classes. Every modality of fitness generates the kind of high-frequency, privacy-sensitive data that existing platforms monetise without participant consent. Running, rowing, weightlifting, yoga — all of it becomes programmable and composable once you have the infrastructure: ZK proofs for effort verification, parallel execution for real-time telemetry, instructor-owned content with embedded revenue logic.

Sui is the performance layer that makes this real. Not because it's fast in a benchmark sense, but because its object model maps naturally onto the actual structure of the problem — independent riders, independent sessions, no shared state that creates bottlenecks.

That alignment between the data model and the execution model is what drops a $7,000-per-class problem to $0.72.

The fitness industry generates $96 billion annually. Almost none of that flows to the people doing the work. Helium proved a community could build a wireless network more efficiently than a telecoms company. Hivemapper is doing the same for street mapping. The pattern is clear: when you give participants real ownership of what they contribute, you get better infrastructure at lower cost.

SpinChain applies that principle to fitness. Effort becomes verifiable. Ownership becomes real. And the instructor in East London keeps what she's built, wherever she teaches next.

SpinChain is live soon on testnet. Explore the protocol at spinchain.vercel.app and follow the build at github.com/thisyearnofear/spinchain.

• Farcaster @papa — warpcast.com/@papa

• Lens @papajams — palus.app/u/papajams

• Twitter @papajimjams — twitter.com/papajimjams

• PAPA: https://paragraph.xyz/@papajams.eth/farcasters-zk-anons

• PAPA: https://paragraph.com/@papajams.eth/reverse-engineering-scout-game

• PAPA: https://paragraph.com/@papajams/the-infra-problem-defi-struggled-to-solve

There's a instructor I keep thinking about...

She teaches six spin classes a week. Has done for years. Her regulars plan their Tuesdays around her — she's built a genuine community inside a 20-bike studio in East London (and beyond online). Brands have reached out. She has a following. And yet the platform she teaches on owns everything: the class recordings, the rider data, the relationship with her students. If she moves studios, she starts from zero. If the platform changes its algorithm, her income changes with it.

She's not unusual. She's the norm.

Peloton has 12 million members and a roster of instructors earning up to $500K a year. Those numbers look impressive until you ask who owns the 100 million-plus classes those instructors have created. The answer isn't the instructors. Meanwhile, rider fitness data — conservative estimates put it at $300-500 per person annually based on acquisition valuations — generates exactly nothing for the people producing it, barring equity incentives. The platforms capture everything; participants usually capture nothing.

This is the problem SpinChain is built to solve. And solving it properly requires building on Sui.

The airline loyalty analogy is a useful one here. Nobody says you're getting "paid to fly" when you earn frequent flyer miles — but those miles are real, they accumulate based on verifiable activity, and they represent a meaningful stake in the ecosystem you're contributing to. SpinChain works the same way. Riders don't get paid to exercise. They earn protocol rewards — SPIN tokens — for hitting verifiable effort milestones, the same way a frequent flyer earns miles for distance travelled.

The difference is that in SpinChain, the instructor earns too. When an instructor deploys a class, alongside the usual revenue from class participants they receive an NFT that contains the revenue logic for that class permanently embedded in it. Ticket sales, secondary trades, sponsor integrations — a programmable split flows back to them automatically, forever, regardless of which platform they're teaching on next year. The class NFT is theirs. The relationship with their riders is theirs, whether the class is real time or being replayed/remixed in future.

This is a "programmable financial event" in practice: a spin class that settles itself.

The obvious question first: why put any of this onchain at all? Strava, Whoop, and Peloton's own app process telemetry on centralised servers in milliseconds for fractions of a cent. No gas fees, no consensus latency, no complexity. If raw performance were the only metric, offchain wins easily.

But offchain systems have a structural ceiling. They can record that you worked hard. They cannot make that record trustlessly verifiable by a sponsor who never met you, composable with DeFi protocols that didn't exist when you signed up, or permanently owned by the instructor who created the class — not licensed to her by a platform that can change the terms next quarter. Offchain apps track your effort. Only onchain makes your effort ownable, verifiable by strangers, and composable with the rest of the financial internet — without trusting the platform in the middle.

The telemetry going onchain is what makes the settlement layer airtight. Which brings the real constraint into focus: how do you do this without the economics becoming absurd?

SpinChain runs on two chains: Sui handles what happens during a class (real-time telemetry, session state, biometric updates), while Avalanche handles what happens after it (reward settlement, ZK proof verification, NFT ownership). You don't use the same infrastructure for a live sports broadcast and the banking transaction that follows it. The same logic applies here.

Indoor cycling generates a genuinely unusual data profile. A single rider in a 60-minute class produces thousands of biometric data points: heart rate, power output, cadence, all streaming at 10Hz. Multiply that across 1,000 concurrent riders and you're looking at millions of individual state updates per minute, in real time.

On most chains, this is economically fatal. Processing 10Hz telemetry across 1,000 simultaneous riders on Ethereum L1 would cost over $7,000 per class — before a single reward is distributed. Even on faster EVM chains the numbers remain prohibitive for the micro-interactions that make fitness apps feel alive: the instant heart rate zone alert, the live leaderboard update when you overtake the rider ahead, the confirmation that you hit your power target in that last interval.

Sui's object-centric architecture changes this entirely. Each rider's session is an independent shared object on Sui — a RideSession — that no other rider ever touches. Because sessions are fully independent objects, Sui's parallel execution engine processes them simultaneously rather than sequentially. No global state bottleneck, no gas wars, no queue. 1,000 riders updating their session objects at the same time creates 1,000 independent parallel operations.

The result: we batch 10 biometric samples per Programmable Transaction Block and achieve 5,000 aggregate TPS across all active riders at a cost of $0.72 per hour-long session. The same session on Avalanche alone would cost $144 per rider. That 99% cost reduction isn't an optimisation — it's what makes the ownership model viable at all. Mysticeti consensus delivers 390ms finality, which is the difference between telemetry that feels live and telemetry that feels like a buffering video stream.

SpinChain goes one step further than helping human instructors. It lets them extend themselves through AI.

The first use case is augmentation: an AI layer that analyses Sui's 10Hz telemetry streams mid-class and surfaces coaching cues in real time. Detecting when the group's average power drops, suggesting a recovery interval, flagging when a rider's heart rate has been in the red zone too long. The human instructor stays in control; the AI handles the data processing no human could do at that speed or scale.

The more ambitious extension — still in prototype — are autonomous agents that can run classes independently. They manage their own schedule, set pricing via bonding curves, and coordinate performance adjustments on Sui in real time, settling earnings on Avalanche. The instructor/developer who train them still earns from every class it teaches.

The access argument remains the same either way. The best spin instructors are expensive and geographically constrained. An AI layer — whether augmenting a human or running independently — can bring that quality of instruction to a rider training at 6am in Lagos or 11pm in Seoul. Autonomy is a spectrum, not a binary; SpinChain lets instructors choose how much of it they want to hand over.

The economic architecture makes this work: the AI agent manages liquidity for its class tokens via Uniswap v4 hooks, adjusting SPIN/USDT pricing automatically based on class demand. When a class fills up and demand spikes, the bonding curve adjusts. When a new time slot opens with low demand, pricing softens to attract riders. This happens without human intervention — a self-sustaining fitness economy.

There's a tension at the centre of any fitness protocol: the data that makes personalised workouts possible is also deeply sensitive health information. Heart rate variability, VO2 max approximations, sustained effort patterns — this is data that insurance companies want, that advertisers want, that you'd probably rather keep private.

SpinChain handles this through zero-knowledge proofs running in your browser before any data leaves your device.

Here's what that means concretely. After a class, you might want to claim a reward for sustaining 85% of max heart rate for 20 minutes. In a traditional system, you'd have to share your actual heart rate data to prove that claim. In SpinChain, you don't. A ZK circuit — built in Noir, running locally on your device in under a second — generates a cryptographic proof that the statement is true without revealing the underlying number. The proof goes on-chain to Avalanche's C-Chain. The raw data never leaves you.

The proof says: this rider met the effort threshold. It doesn't say what their heart rate was. It doesn't say what their fitness level is. It doesn't say anything an insurer or advertiser could use. Selective disclosure, by design — you can share what matters (I worked hard) without sharing what's sensitive (here's my cardiovascular baseline).

Encrypted raw telemetry is stored on Walrus, Sui's decentralised storage layer, where only you hold the keys. The chain knows you worked. Nobody else knows how.

SpinChain is entering Avalanche's Build Games competition — six weeks of intensive building toward a concrete target: 100 riders completing verified workouts, 10 instructors minting their first class NFTs and receiving revenue. These are modest numbers deliberately. Product-market fit in a new category comes from proving the loop works for real people, not from projecting scale.

But the thesis extends well beyond spin classes. Every modality of fitness generates the kind of high-frequency, privacy-sensitive data that existing platforms monetise without participant consent. Running, rowing, weightlifting, yoga — all of it becomes programmable and composable once you have the infrastructure: ZK proofs for effort verification, parallel execution for real-time telemetry, instructor-owned content with embedded revenue logic.

Sui is the performance layer that makes this real. Not because it's fast in a benchmark sense, but because its object model maps naturally onto the actual structure of the problem — independent riders, independent sessions, no shared state that creates bottlenecks.

That alignment between the data model and the execution model is what drops a $7,000-per-class problem to $0.72.

The fitness industry generates $96 billion annually. Almost none of that flows to the people doing the work. Helium proved a community could build a wireless network more efficiently than a telecoms company. Hivemapper is doing the same for street mapping. The pattern is clear: when you give participants real ownership of what they contribute, you get better infrastructure at lower cost.

SpinChain applies that principle to fitness. Effort becomes verifiable. Ownership becomes real. And the instructor in East London keeps what she's built, wherever she teaches next.

SpinChain is live soon on testnet. Explore the protocol at spinchain.vercel.app and follow the build at github.com/thisyearnofear/spinchain.

• Farcaster @papa — warpcast.com/@papa

• Lens @papajams — palus.app/u/papajams

• Twitter @papajimjams — twitter.com/papajimjams

• PAPA: https://paragraph.xyz/@papajams.eth/farcasters-zk-anons

• PAPA: https://paragraph.com/@papajams.eth/reverse-engineering-scout-game

• PAPA: https://paragraph.com/@papajams/the-infra-problem-defi-struggled-to-solve