NEAR Protocol co-founder Illia Polosukhin has sketched a consumer crypto roadmap that reads more like a product wishlist than a whitepaper, and the star of the list is a chain-native take on a 1990s toy: an AI Tamagotchi where every pet is a non-fungible token with its own wallet and its own inference budget.
An NFT pet that pays for its own brain
In an Oct. 3 post on X, Polosukhin outlined the concept in unusually concrete terms. Each digital character would be represented by an NFT, and that character would carry a NEAR balance dedicated to funding AI inference. The AI behavior would run through NEAR AI with verifiable inference from a prompt encoded onchain, meaning the personality and logic that drive the pet are anchored to a smart contract rather than a company server that can be switched off.
The gameplay loop he described starts familiar: feeding and dressing the character. From there it escalates to battle arenas and competitions, giving the tokens a use beyond static collecting. Because the pet holds its own balance, the owner tops it up rather than paying a subscription, and the creature’s intelligence keeps running as long as the wallet stays funded.
The idea lands at a moment when AI companion apps are among the fastest-growing consumer software categories, while NFT collections have struggled to hold attention after the trading mania cooled. Polosukhin’s framing effectively merges the two: the collectible becomes the container for a persistent AI agent, and blockchain rails handle ownership, payments and verifiable computation in one stack.
Payment plumbing already exists
The Tamagotchi pitch is less speculative than it sounds, because the payment mechanism it leans on is already live. On July 31, NEAR introduced staking-based AI payments that convert locked tokens into monthly computing credits. The system covered 43 AI models at launch, including models from OpenAI, Anthropic and Google, and users keep ownership of their staked tokens, recovering them after unstaking. The locked balance determines how much inference the account can draw each month.
That model maps directly onto the pet design: an NFT with a staked balance becomes a self-funding AI character without the owner managing per-request payments. It also sidesteps one of the recurring criticisms of consumer crypto products, namely that users are asked to become amateur payment processors. Here the economics are bundled into a single deposit-and-forget action.
Checkout widgets, escrow check-ins and chain-hopping wallets
The AI pets were only one item on Polosukhin’s list. For merchants and application developers, he proposed a universal checkout widget that lets customers pay through NEAR Intents using whatever wallet and blockchain they already hold assets on, with Shopify floated as a distribution target. For event organizers, he suggested a registration system that holds attendee deposits in escrow until the host checks them in, with attendees who show up splitting the contributed funds — an extension of the older Kickback project across chains and assets.
He also pushed wallet compatibility as a distribution lever. Through Aurora’s Intent Connect, users could interact with applications from other chains without arranging a bridge or learning NEAR’s mechanics. That suggestion has a working precedent: on Sept. 17, Aurora Labs announced one-signature Sui execution, letting supported assets from other networks move into Sui applications through Intents Connect, with NEAR Intents supplying liquidity, settlement and cross-chain connections.
The through-line in the post is distribution over infrastructure. Polosukhin wrote that new and returning developers had been building on NEAR over the preceding weeks and that building itself had become easier, so the real questions are what to build and how to stand out. His answer is to stop asking users to adapt to crypto and start making crypto invisible inside products people already understand: a shop button, an event ticket, a virtual pet.
Why the pet could matter beyond the gimmick
Skeptics will note that Tamagotchi nostalgia is a thin hook, and that AI companions with crypto rails have been attempted before with modest results. But the design has one property most NFT projects never achieved: a recurring reason to interact. A collectible that must be fed, funded and battled gives owners a daily touchpoint, and each touchpoint exercises the underlying rails — token balances, onchain prompts, verifiable inference.
It also gives NEAR a distinct narrative among layer-one chains. Rather than competing on throughput claims, the ecosystem is positioning itself as the place where AI agents are owned, paid and verified onchain. If the Tamagotchi concept ships and finds an audience, it becomes a demo that other developers can copy; if it fails, the payment and cross-chain infrastructure it exercised remains useful for less playful applications.
For now, the proposal is a post, not a product. No launch date, tokenomics or official team was attached to the pet concept. But as a signal of where a major protocol co-founder thinks consumer crypto attention should go, it is a telling one: the next wave of NFTs may not be profile pictures, but creatures with a wallet and a mind of their own.
Disclaimer: This article is for informational purposes only and does not constitute financial or investment advice. Digital assets are volatile and readers should do their own research.
the pet running its own inference budget is cute until a grifter deploys 10k of them as a sybil farm with better prompts than everyone else. arena pots need anti bot design on day one
sybil farming pets for arena rewards is just airdrop farming with extra grief. honestly kind of on brand for this space
a tamagotchi with its own wallet so it can pay rent on its own brain. two weeks max before someone teaches theirs to beg for NEAR
verifiable inference with the prompt onchain is the real news, the pet is just the demo. expect everyone to copy this pattern
hard disagree, the pattern only matters if inference stays cheap. the moment arena rewards beat NEAR staking yield the prompts get gamed into pure spam
this is the real economics question. the second a pet’s arena ev beats its inference cost, the network is 90 percent bots by design
cap arena entries per pet per day and the bot math dies. nobody will implement it because volume looks good in dashboards
you know someone is already training theirs to dm people for top ups. the grift layer writes itself
Illia pitching a 90s toy nostalgia play and somehow making the inference budget per NFT sound reasonable. might actually print in asia
An NFT that pays for its own inference is the first consumer crypto idea in years that doesnt ask users to become payment ops. Staking 43 models into compute credits already works, so the pet wallet part is the easy bit.
The prompt living onchain is the part people sleep on. If the personality is a contract, nobody can quietly patch your pet into a different creature after you paid for it.
Tamagotchi died when the battery ran out. These die when the NEAR balance hits zero, which is somehow more brutal. Kids in 2027 will learn gas fees through grief.
kids learning gas fees through grief is dark but real. first thing mine would ask is why the pet costs money to think
top up or it dies is the most honest subscription model ever shipped. my tamagotchi never warned me, this one at least shows you the balance running out
learning grief through gas is unhinged and also the best onboarding trick ever shipped. the balance bar is a countdown you can actually do something about
Battle arenas for self funded AI pets sounds like bot farms with extra steps. Whoever scripts the smartest prompt wins the pot and the meta gets solved in a week.
Prompt onchain is tamper-proof but not bug-proof. One vague function in the personality contract and the arena meta exploits it forever, upgrade forks the whole pet lineage. Devs will hate that tradeoff.
the arena part will absolutely become prompt golf within a week. still, a pet whose brain is a verifiable contract beats a server toy that dies the day the studio pivots
contract brain beats server toy until the arena pot gets big enough for prompt laundering. someone will sell champion personalities on the secondary within a month
selling champion personalities on secondary is the actual endgame here, the pet is just the license key
meta gets solved in a week and then the devs patch the prompt, which is exactly what the onchain contract prevents. the personality being frozen is the feature and the bug
bot farms is the boring take. the losing pets still hold value as collectibles since the personality contract is permanent, the arena is just the funnel
a tamagotchi that budgets its own inference is the first consumer gas meter that feels alive. kids will grok nonce management before multiplication tables
my tamagotchi starved back in 98, my nieces pet will die of gas fees instead. same grief, fancier ledger
the staking to credits mechanism from july is the actual product here. the pet is a demo skin for prepaid ai compute with extra steps
the july staking to credits update was the roadmap hint all along. the pet is a demo skin for prepaid compute and it might work
onchain prompt means the pets personality is auditable. imagine checking a smart contract to verify your tamagotchi is actually loyal lol