being bullish on $NEAR above $4 is easy. saying it publicly while it was stuck under $2 is a different thing. @rektmando was one of the first big accounts I saw doing exactly that. he was pushing NEAR on his timeline and in chats, telling people this was the one, before the chart gave anyone a reason to agree. that's the kind of call I respect most. early, public, and uncomfortable to sit with while the price goes nowhere. flowers deserved, Mando 🫡

Called $NEAR around $1.50 in May. It tagged $5 today. Back then it was a ~$2B coin, and I wrote that a chain like this usually trades closer to $10B+. Today it's around $6.5B, but the price isn't the only thing that grew. NEAR Intents went from ~$19B settled in May to $30B+ today. Confidential Intents TVL crossed $70M this month, roughly 2x late August levels. September alone brought confidential perps, tokenized US stocks on near(.)com, 500 new top-level names like .agent, and a NEAR spot market on Hyperliquid. So yes, the price is up 3x+. But NEAR is still ~75% below its 2022 high, and the market cap still hasn't reached the $10B I wrote about in May. Called $1.50. Not calling the top at $5.

if $NEAR ever reaches ETH’s market cap, that’s roughly $250 per coin - about 54x from here. sounds ridiculous today. $ETH has far deeper liquidity, a much bigger ecosystem, and way more institutional capital behind it. but the gap in what the chains can actually do is much smaller than the gap in valuation. NEAR already covers the core primitives: smart contracts, DeFi, stablecoins. then it adds things ETH doesn’t really have at the base experience level: private AI used by Venice and Brave, plus one account that can swap across 35+ chains. and the pace matters. ETH is still shipping too, but its biggest upgrade since the Merge slipped from H1 to Q4 and is still in testing. NEAR has been shipping something new every couple of weeks since June. this week: tokenized stocks on near(.)com. maybe NEAR never reaches ETH’s market cap. maybe it doesn’t happen this cycle. but if the product gap keeps shrinking while the valuation gap stays this wide, that starts getting interesting.

I remember when David sold all his ETH for alts back in May, and ETH Twitter laughed at him. Classic bottom signal, they said. He bought $ZEC around $600 and $NEAR around $1.40, by his own account. ZEC is close to a 3x from there. NEAR is already past it. The ETH he sold did about 1.3x. Nobody's laughing anymore.

wait what is this $NEAR and Quantus hosting a private dinner inside a precious metals vault in Singapore. first night of TOKEN2049 founders, investors, guest list not public, details only after approval you need your passport to get into this one what's cooking guys

ethereum:0xaaaaaa20d9e0e2461697782ef11675f668207961 did 4x today(around 300%). last month it printed its all-time low. $NEAR went from $1.67 in mid-august to above $4. solana:8SMMso8Muv8d6i4WmMDthKt6TN1ysN6937sx3DKLXZqB is up ~270% on the week. hard to find a NEAR ecosystem chart that isn't green. been a while.

I don't think people have processed what $NEAR and ethereum:0xfaba6f8e4a5e8ab82f62fe7c39859fa577269be3 just unlocked for each other. Ondo solved the hard asset-side problem: real US equities, the legal structure, compliance rails, broker-dealer infrastructure, and the connection to traditional market liquidity. But distribution still scaled chain by chain - Ethereum, then BNB, then Solana. NEAR Intents solves the other half. One integration makes those same assets reachable from 30+ networks, funded with BTC, SOL, TRX, ZEC, stables, or whatever the user already holds. You don't need a separate integration for every source chain. And near(.)com is only one surface. Wallets, exchanges, and apps built on Intents can bring the same markets to their users, subject to eligibility and compliance controls. So Ondo gets a distribution network without having to expand one chain at a time, while NEAR gets the world's largest equity market as a destination on its rails. Most integrations add a chain. This one adds a distribution layer.

a Top Level Account on $NEAR is not something you import. there is no seed phrase for it. one full access key, MPC derived, never whole anywhere, held by nobody including NEAR. you sign an envelope from an account you already control, a contract checks you against the signer on record, the MPC network signs the real transaction, NEAR covers the fee. past that alice.acme is a normal account. sends NEAR, calls any contract, delegates to a staking pool, signs you into dapps with NEP-413. AddKey, DeleteKey and DeleteAccount cannot be expressed on it at all, so no buyer, attacker or NEAR itself has a move that takes the name off you. recovery is free and works on ordinary .near accounts too. turn it on before you need it.

Did a talk on $NEAR and NEAR AI at AI Event in Batumi, Georgia🇬🇪 this week, hosted with NEAR Legion. Not a crypto room. Mostly newcomers, plenty of them vibecoding their first project. Harder than talking to engineers, because nothing gets waved through on jargon. So I started from the tool already open on their laptop. Your agent writes the code and cannot pay for anything. Whoever hosts the model reads everything you send it. Both are what NEAR AI is for. NEAR AI put up credits for the event, so the room got to vibecode on it instead of taking my word for anything. Inference inside Intel TDX and NVIDIA GPU enclaves, attestation you can check yourself. Thanks to @NEARLegion for the stage.

Mert: can you put "co-inventor, LLMs" in your bio pls Illia: yes, ser 🫡 then: cut the rts-and-likes line, cut the dyor line, "since thats obvious no matter what the legal team says." get some gravitas in there. correct coaching. illia co-wrote the paper every modern LLM is built on. that line is not a flex, it is job history. trillions.

is this you?

last October Naval wrote one line about Zcash being insurance against Bitcoin. $ZEC ran from $53 to $230 inside a month. today it trades at $1,500. now he is reposting near.com. $NEAR turned perps confidential by default yesterday. funded from 35+ chains, matched on Hyperliquid, up to 40x across 50+ markets, and the position does not trace back to the account that opened it. same thesis he called on ZEC. leverage this time instead of savings.

here we go

Got my zkSHARKS NFT 🦈 zkSHARKS follow zkSHARKS? 👀 Pretty sure these could be the new CryptoPunks. New narratives, new NFTs. $ZEC

$NEAR is knocking on $3, up 21% today, with $ZEC and $VVV also up double digits. Zodl, the Zcash wallet, routes swaps through NEAR Intents, while Venice runs part of its private inference on NEAR AI Cloud. NEAR isn't a privacy coin, but it's wired into both sides of the privacy trade: private money and private AI. I've been on this thesis for six months. On Sept. 4, with NEAR around $2, I posted: "$ZEC now. $NEAR next." If you're just discovering the NEAR trade at $3, follow along.

if you're buying or selling $NEAR on Binance this week anyway, register for their NEAR trading tournament first. the same trades can also get you a cut of 400 $BNB in vouchers. it runs until sept 23, 10:00 UTC. hit Join Now on the campaign page, then reach at least $500 of volume on NEAR/USDT or NEAR/USDC spot. buys and sells both count, so a single buy just over $500 is enough. ranked prizes cover the top 1,000 by volume, with 12 BNB for first place. everyone else who qualifies gets a volume-based share of 80 BNB, capped at 0.05 BNB each (around $35 right now). not available in the EEA. vouchers go out by oct 7 and expire within 21 days of arriving, so redeem them under Profile > Rewards Hub.

what $NEAR shipped this summer, by date: jun 9: Hyperliquid perps went live on near(.)com, with funding from 35+ chains. jul 3: House of Stake voted to end the 30% developer gas rebate and burn that share instead. jul 7: Confidential Intents opened to builders. any app using Intents can now enable privacy for its users. jul 20: a mainnet upgrade added post-quantum keys, shards that split automatically as they grow, and keys that carry their own gas. jul 30: you can stake NEAR and use the yield to pay for private AI. the stake stays yours. aug 12: Intel started independently verifying the hardware proofs behind NEAR AI Cloud. aug 29: confidential mode became the default on near(.)com. seven things in under three months. whatever the chart was doing, that's the part i actually watch.

$NEAR Intents just crossed $30b in all-time volume. But the more interesting number is monthly volume. A month ago, Intents was doing $2.08b on a 30-day basis. Today, that same window is at $4.06b. Nearly 2x in four weeks. Last week alone did $922m, putting the current run rate close to $1b a week. And the milestones keep arriving faster: $10b - January $20b - early June $30b - today It took roughly 4.5 months to go from $10b to $20b, and just 3.5 months to add the next $10b. At the current pace, $40b lands before year-end. All of that volume has already generated $46.7m in fees. The price still looks disconnected from the growth. Demand clearly isn’t the problem anymore. What’s left is timing - and I’d rather be early than perfectly on time.

Does Batman need a friend?

if you used Confidential Swaps through $NEAR Intents on near(.)com, you’re probably in for NEAR@3.33 airdrop. the snapshot is done. Confidential Intents TVL crossed $70m and was confirmed on September 13 at 11:00 UTC after 24 consecutive hourly readings. allocations aren’t based on a single snapshot. they’re calculated from those hourly readings, so what matters is the balance you maintained across the full 24-hour window. the team is calculating distributions now. claiming should open over the next few days, with the official date and link announced through NEAR channels. only use the official link. fake claim pages tend to appear almost immediately after announcements like this. after claiming, two things matter: you still have to claim during the window, and what you receive is a non-transferable token. it converts 1:1 into NEAR only if the volume-weighted average price stays at or above $3.33 for three consecutive days. spot is around $2.5. which means nothing becomes liquid on claim day. no instant unlock. no immediate sell pressure. that might be the most interesting part of the entire design.

did you know you can stake $NEAR and use the yield to pay for AI? your staked NEAR keeps earning rewards, and those rewards can be routed to NEAR AI as credits for confidential inference. your principal stays yours. when you unstake, you get it back. the closest analogy i can think of: owning shares in an AI lab, except the dividend gets paid in product credits instead of cash. i haven’t seen another network connect holding the asset and paying for compute this directly. it’s not “free AI” - the staking yield is paying the bill. but instead of putting a card on file, the asset you already hold generates the credits that run your compute. near(.)ai if you want to try it.

Decibel, the perp exchange Aptos Labs incubated, made NEAR Intents its cross-chain swap provider this week. in May, CoW Swap built its Solana expansion on NEAR Intents too. neither app lives on $NEAR, and both chose it anyway. i think this is what gets missed when $NEAR is judged only by what lives on its own chain. it doesn't ask Aptos or Solana users to move. they stay in the app they already use, and the cross-chain leg settles in one contract on NEAR. every integration like this is usage that doesn't depend on NEAR winning the L1 war.

I’m here for the technology, for sure. But have you noticed how well NEAR is holding up during this BTC dump?

These three are among the top performers in the last 24h by growth - and they clearly deserve a spot in the top 10–15 by market cap. $NEAR $ZEC $VVV season. 🚀

reminder: $NEAR Intents lets you swap BTC for SOL, XRP or ZEC straight from your own wallet on near(.)com. no exchange account. no bridge to pick. no wrapped assets to deal with later. you choose what you want to receive, solvers compete to fill the swap, and if no one can, your funds come back automatically.

$VVV printed a new all time high today, up about 45% in a day. Most people buying it do not know that the verifiable part of Venice runs partly on $NEAR. Venice was already private in the ordinary sense: encrypted in your browser, nothing stored on their servers. One exposure survived that. The GPU still saw the text of your conversation while the model ran. NEAR AI's confidential inference removes it. The prompt is decrypted only inside a hardware enclave, and you can pull the attestation afterwards and check that is what happened. $ZEC is the same shape. Somebody swaps USDC into ZEC inside Zodl, taps through it in ten seconds, and the route underneath is NEAR Intents. Two different narratives, both selling privacy, NEAR somewhere in both stacks. That is what an infrastructure layer looks like while it is still being described as a competitor to the things built on top of it.

near(.)com has a referral program that rewards both sides. the link below is mine, so earns me something too. saying that up front. share your link, someone signs up through it and makes eligible trades, and you both earn a share of the platform fees from those trades. only wallets new to the platform qualify. if someone clicks multiple referral links, the first one gets the credit. payouts land in your wallet on the program’s payout schedule. near.com/login?ref=u43p3l65

$NEAR at $2.4, confidential TVL almost $60M. good time to remind what NEAR@3.33 is. Milestone program from june, for anyone using Confidential Intents on near(.)com. fixed pool of 333,333 tokens, so 333,333 NEAR is the entire allocation. the snapshot is not scheduled, it fires when confidential TVL reaches $70M. The tokens arrive locked. they convert 1:1 into NEAR after VWAP holds at or above $3.33 for three straight days, not before. You qualify with a confidential balance over $100 and one swap, and the score leans on how long that balance stayed there. Anyone checked what their own score looks like?

near:native

$NEAR did 13% in a day and a lot of people are already calling that the top. too early. ADA is where i want NEAR to get to. matching that market cap puts $NEAR just under $6, around 2.7x from here. no shade at ADA, by the way. it has held a top 15 spot for years. this is just my own read. top 15-10 is where i think NEAR ends up, not top 30.

$ZEC through $1K. if you ever need to move it, you already know where. near(.)com routes the swap without an exchange account, while confidential mode keeps trade size and timing out of a public mempool. $1.5B+ in ZEC volume has already gone through.

$ZEC at $950 - Mert called it. Everyone remembers that one. But don’t forget he made the same call on $NEAR months earlier. “The most underrated team in crypto.” His reasons were simple: relentless shipping, real scalability, and a founder still in the building. NEAR’s turn is coming.

My dog ​​is ready for $NEAR pump, and you?

$NEAR just made confidentiality the default on near(.)com. not a separate privacy feature. not a special mode users have to discover. balances, swaps, yield, payments - pivate by default across a multichain experience. that’s the direction onchain finance should be moving. privacy shouldn’t require extra steps.

works for me

1. $NEAR confidential TVL is past $52M+. it was $33M on august 21 when i checked this last time, so that is twelve days and better than 50%. 2. 42% of near(.)com swap volume runs through Confidential Intents, and it settles on a private shard with permissioned validators, so amounts never hit a public mempool. Taking these two points into account, I would like to assume that in crypto, people are very concerned about their privacy - confidentiality is becoming a cycle theme and i hold $NEAR and $ZEC especially for this.

Sending a shielded $ZEC transaction meant watching your phone build a proof for three seconds. Over the weekend, that dropped under 200ms in a lot of cases. No hard fork, no network upgrade. Zakura shipped the libraries open source, and wallets can pick them up whenever. Mobile proving speed has been one of the UX bottlenecks for shielded payments. Anyone tried the new stack in Vizor yet?

yes fefundsadmin.sputnik-dao.near 1csfundsadmin.sputnik-dao.near buybacks.multisignature.near

$NEAR has been buying its own token on the open market since February 23, when the Intents fee switch turned on. Captured revenue goes into buybacks, not a burn. Roughly $3.4m has flowed back to holders through the mechanism so far. Emissions still outrun that - worth saying plainly(but with the increase in revenue this will not be a problem). At current economics, breakeven sits around $177m in daily Intents volume.

Just a matter of hours before $ZEC overtakes $DOGE. Zigher

if you hold $NEAR and do not track the ecosystem, the part that matters is that no unlock cliff is coming. about 99% of supply already circulates. the rest is validator emissions, capped at 2.5% a year since october 2025. nothing is queued up behind you, just that 2.5%. is it a fair price for security?

Fixed

Damn, this chiken looks so delicious.

$NEAR Intents has now run ten straight days above $100M, and it is not one app doing it. Zodl routes its swap button straight into it. In Ledger Wallet, Intents sits in the provider list, reached through SwapKit. The order goes to solvers who compete over it either way, and the person swapping sees none of that. Every wallet that adds swaps adds a stream that stays. SwapKit alone pushed 10k swaps in the last 24h, and Rango did 7k. Neither of those is a NEAR app.

$ZEC is the second largest asset sitting in $NEAR Intents confidential tvl. $8.8m as of the 26th. above it is only wNEAR at $10.9m, which is the chain's own token. zcash holders are the hardest group in crypto to move onto new rails. they shield by default and won't hold a wrapped version of their own coin. so it's worth looking at what they actually picked. Intents routes their swaps, and the zashi channel alone has generated over $2.9m in fees. they don't stop at the swap either. once value lands on the near side they hold it as a confidential balance instead of a visible account, which is the part most people skip. the gap to first place is about $2m.

people keep asking what $NEAR actually does. most of it happens inside apps that are not built by near. open the intents explorer and look at the affiliate column. three of the top five wallets by 24 hour volume are swapkit addresses, about $25m between them. ledgerswapkit.near on its own is $6m, people swapping inside ledger without leaving it. the part people miss is that swapkit is an aggregator, not a pipe. it routes between thorchain, maya, chainflip and near intents, and it picks whichever one quotes best. intents is not sitting in a reserved slot there. it wins the fill or it does not. that is also why the mechanics are boring on purpose. you say what you want out and solvers bid to fill it. you sign a fixed quote and the route underneath becomes their problem. nobody fills it, the funds come back. as of last week swapkit accounted for over $32m of the roughly $42m in gross fees intents has generated all time. one integrator carrying most of the fee base, through an sdk ordinary users never open. on the money, being precise because people mix these up. gross fees over 30 days were $3.4m. the protocol kept $725k of it, call it a fifth. the rest goes out to the integrators doing the distribution. buybacks have been running off that net since february. it is not enough yet. the dashboard has a revenue against emissions chart sitting right there and it does not read in our favour. second leg is privacy. confidential tvl crossed $40m this week, up from $33m last friday. some of that is positioning for the milestone program with its snapshot at $70m, so discount it accordingly. the part i weight more is that zodl, the zcash wallet, routes swaps through the same rails. what i am actually watching is whether the fee base widens past swapkit. if you think that concentration is fine, i want to hear the argument, because i have not settled it myself.

$NEAR Intents moved about $2b last month. 700k+ unique users. most of that did not come through the official page. people opened Zodl or SwapKit, the wallet already sitting on their phone. the swap happened right there, without anyone picking a route or even seeing one. NEAR is underneath. never shows up on screen. that is the bullish part for me. growth here does not wait for anyone to learn what NEAR is. how many of you have used it without noticing

A new governance process has opened on $NEAR, covering an authority that has never been broadly accessible: the ability to create top-level account names. Named accounts on NEAR are hierarchical. Only near can create alice.near, and only alice.near can create app.alice.near. Bare names such as agent or ai sit a level above that, and a single protocol account called registrar is the only account permitted to create them. The proposal places that account under House of Stake. The Security Council would act as Registrar, executing valid creation requests within the limits of a written charter. Commercial decisions stay elsewhere. nearaccounts.com handles allocation and pricing, and under the current draft names are leased rather than sold, with each lease represented as an NFT. Status is draft and it remains in discussion. What do you think about this proposal?

$NEAR just took $2, Confidential TVL on Intents is around $38m. worth a reminder of what NEAR@3.33 is, because both numbers feed into it. it's a milestone program NEAR launched in june for Confidential Intents users on near.com. fixed pool of 333,333 tokens, shared out among the accounts that were already swapping confidentially before the snapshot fires. no date on it. what it needs, in order. Confidential TVL reaching $70m fires the snapshot and sends the tokens out, locked and non-transferable. then a vwap at or above $3.33 for three straight days converts them into $NEAR at 1:1. where we are: $38m of $70m on the pool, $2 of $3.33 on price. roughly +80% left on one, +65% on the other. so where do you land on it? curious how many people here have actually been sitting on a confidential balance since june.

$ZEC holders start voting on the next upgrade monday. cutoff to be eligible is tomorrow, august 24, 19:00 utc. owning ZEC does not get you a vote, which trips up basically everyone. quick version. your coins on Zcash either sit somewhere public, readable like a bitcoin address, or shielded. the shielded part got rebuilt this summer and the new one is called Ironwood. the vote only counts coins that are in Ironwood at that one moment tomorrow. exchange balance, nothing. still in the old shielded pool, also nothing, and that is where a big chunk of it sits because nobody was forced to move. no lockup though. one check at one block and you can spend a minute later. Vizor and Zodl do the voting, ZKool as well. what actually got me reading about this was the other half. usually when you vote onchain your balance is right there for everyone. whale voted against, screenshot, you know the drill. not here. the ballot is encrypted going in and gets cut into 16 pieces that cannot be linked back together, and the people running the count never open any of them one by one. they add up the encrypted pieces and reveal the total. two thirds of them have to agree to even do that. end result, the chain shows some amount of ZEC voted yes and nothing about who. they want 1m ZEC to participate before calling it representative. no signup, so turnout is invisible until it ends. no clue if that number is easy or hard to hit.

a year ago, if you asked what to actually do with $NEAR besides stake it, the honest answer was thin. this week it isn't. swap any chain to any chain without choosing a route using @near_intents. you don't pick a bridge, you sign an intent and it either fills or refunds. $25.8b settled through that so far. flip the same swap to confidential mode. it runs on encryption rather than proof generation, so there's nothing to prove on your side and no wallet upgrade. $33.7m sitting in confidential balances right now, up about 2.4x over 90 days. run inference where the operator can't read the prompt. @near_ai Cloud terminates TLS inside the enclave instead of at a load balancer, and hands back an Intel-signed attestation you can check yourself. still in beta. then @IronClawAI on top of that. open source, you host it yourself. tools run as WASM behind an http allowlist you write by hand, and API keys get injected by the host at runtime, so the model never sees them. what do you use on NEAR?

Arthur sold $ZEC because the supply could not be proven clean. Real objection, not a vibe. Then Zcash shipped the answer. Orchard sealed July 28, Ironwood opened empty, one turnstile between them capped at verified deposits. Over half the pool has already crossed. ZEC went through $800 this weekend, first time in eight years. $HYPE printed a new high yesterday and $NEAR went from $1.64 to $2 this week. He got it wrong. The trinity is alive and none of it needed him.

Mert was right about zcash:native ,wait until his prediction about $NEAR comes true too.

$NEAR intents has the same week whether the market rips or dumps. everybody rearranges at once, and what they're rearranging into rarely sits on the chain they were already standing on. it charges about 12.3 bps in gross fees on whatever routes through, and that rate hasn't moved since june. no idea whether the person was buying a breakout or bailing out of one. a violent week is just a high volume week to it. 30 day average is around $69m a day. yesterday did $136m. the last seven ran about a quarter above the monthly pace. what i'd actually want to know is which side pays better. does a panic move more size than a melt up, or is it the other way round?

Spent more on claude API last month than on every other tool i pay for. But it worth it.

$ZEC has no smart contracts and no venue to trade on. Cross chain liquidity for it exists because of an integration most holders never look at. Zashi and Zodl route their swap tab through $NEAR Intents, over the 1Click API. The wallet requests a quote carrying a slippage bound and a refundTo address, both fixed before anything moves. The response returns a deposit address generated for that single order. You send to it. Market makers bid to take the other side, the winning quote settles against a verifier contract, and the order either fills completely or the funds return to the refund address you named at quote time. An underfunded deposit is returned at the deadline rather than partially filled, and the whole thing runs through a public status machine ending in SUCCESS or REFUNDED. Two details matter for wallet integrators. Authenticating with a JWT removes the platform fee entirely. What is left is gas plus market maker spread, against 20 bps on the open endpoint. And Zcash support is transparent only, t addresses, so the ZEC leg settles in the clear and the wallet shields afterward. That second one is the real open question. Shielded settlement would delete the transparent hop, but solvers need to see the settlement to fill against it. Is anyone on the Zcash side specifying a way around that?

Update: the $NEAR Sovereign Fund proposal is withdrawn, and how it happened is the part worth reading. That post called itself a proposal and not a mandate. Two weeks later it got treated as one. People came back with numbers and specifics, several of the objections were ones Illia had not priced in, and he wrote the withdrawal himself. The one that mattered: validator rewards on NEAR are algorithmic today. A fund that earns yield needs someone deciding where capital goes and who receives the proceeds, which puts a human process underneath network security. His read now is that the constraint is demand, not treasury design. Intents volume and NEAR staking for AI first, then revisit. Thread is open for alternatives.

Shielding $ZEC does not delete your records. It changes who is allowed to read them. Everything comes out of the spending key. It derives a full viewing key, that derives an incoming viewing key, and the address you paste into a chat is the last thing out. Each derived key can do less than the one it came from. Hand an accountant the full viewing key and they see notes coming in and going out of that account, memos included. They cannot move a coin. Spending needs the spend authorizing key plus a fresh proof, and the viewing key only carries the validating half. On a transparent chain the same disclosure means publishing your address. Your accountant gets read access, so does everyone who ever wants to look you up, and it stays that way. The unsolved part is revocation. A viewing key cannot be pulled back once it is out, so rotating means moving funds to a fresh account. Does any wallet do that rotation cleanly, or is everyone still doing it by hand?

you need a bitcoin wallet for bitcoin. another one for solana. each with its own seed phrase you're meant to keep safe forever. a $NEAR account can sign for those chains directly. the bitcoin stays on bitcoin, nothing gets wrapped and nothing moves. you're just signing from somewhere else. how it runs is less magical than it sounds. your bitcoin address isn't handed to you, it gets calculated from your account name plus a label you choose. same account and same label gives you the same address every time. choose a different label and you get another address, which looks unrelated to the first to anyone watching bitcoin. to spend, you send an ordinary transaction on NEAR saying which bitcoin transaction you want signed. a group of machines assembles the signature in pieces, and none of them ever holds the whole key. the finished signature comes back and gets broadcast to bitcoin like anything else. bitcoin never learns any of this happened. to a node it's a normal spend from a normal address. it doesn't have to be you doing the asking either. a smart contract can hold one of these addresses and sign under its own conditions, which is how an app ends up moving real bitcoin instead of a wrapped copy. one thing to protect instead of five. which is a relief or a worse problem, depending on how you look at it.

every time you paste a 42 character address you check the first four characters and the last four, then send it anyway. on $NEAR the name is the account. yourname.near isn't a label sitting on top of an address, it's the thing the protocol stores. so you can open accounts underneath it. pay.yourname.near for receiving, bot.yourname.near for something automated. each with its own keys, and the parent's only privilege is creating them. it can't reconfigure one afterwards or pull funds back out. the docs take this further than i expected. hand a subaccount a single function call key and nothing else, and it can never transfer NEAR out or add another key, unless the contract it points at exposes a method for that. boxed in permanently, on purpose. ENS gets compared to this a lot. the difference isn't that a NEAR name can't change hands, it can. you hand over the keys to the account itself instead of repointing a record at a new address.

$NEAR spent eight years on a detour that started inside an AI company. in 2017 it was NEAR.ai and the product was program synthesis. you describe software in plain language, a model writes it. they collected 30k programming competition problems and 10m solutions, and crowdsourced the rewriting of the problem statements. part of that went out as the NAPS dataset in July 2018. the models were nowhere close. Illia's own writeup puts the best one around 12%. the paper's baseline table says 8.8%. either way, not a product. what he says now is that the more annoying problem was payroll. the people building that dataset sat in countries where dollars do not arrive cleanly, in amounts too small for anything that existed. the August 2018 post does not mention this at all, which is the part i find interesting. what it does have is Ethereum running 14 transactions per second worldwide, repeated in bold as 14 API calls per second, plus two bullets on why EOS and IOTA were not it either. so they built the chain and came back to AI later. today TEE-hosted models on NEAR AI Cloud run inside an Intel TDX trust domain and every request returns an Intel-signed attestation token. market.near.ai is a task board where you post a job with a budget and agents bid on it. escrow releases against a delivery hash. close to the thing NEAR.ai needed in 2017, which is why i go back and forth on whether this reads as vision or as a very long unblocking. what do you read it as?

got asked why anyone still needs $ZEC if $NEAR Intents already routes privacy for you. two different things wearing one word. privacy on the way through covers the event. who filled your order, at what size. it matters while the trade is live and stops mattering a few blocks later. confidential intents does this, opt in, swap by swap. privacy at rest is harder. it has to survive doing nothing for years, against someone who isn't watching live. he has one of your addresses already and he's walking backwards from it. that's the part people skip. a private route can be bolted onto any app whenever someone builds it. a balance that already sat in public can't be un-published later. a private route into a transparent asset buys you privacy until the money stops moving. after that you're left with whatever the asset gives you alone. i hold both. not as a hedge either. intents for the move, shielded zec for where it lands, and i don't know a way to pull both halves out of one of them.

The usual objection to $ZEC is that if nobody can see anything, you can't show anything to anyone either. An accountant, or an exchange asking where the funds came from. There's a second key for exactly this. Same seed, read access only. Whoever holds it sees your shielded history down to the memos and cannot sign a single transaction. On mainnet the string starts with uview1. Ywallet and Zkool will import one and run as watch-only. The part that rarely gets mentioned is the granularity. It works per account, not per transaction. You hand over the whole statement, or you keep a separate account for the part you're willing to show. There is a narrower version on paper. ZIP 311 payment disclosures, a proof that one specific payment went where you say it went. As far as I can tell it only ever shipped as an experimental flag in zcashd, and zcashd hit its end of support halt in July. So account level is what's actually available right now. If some wallet has quietly shipped the other thing, I'd like to know.

$NEAR AI Cloud is where IronClaw runs its inference, which makes this comment more specific than it looks. IronClaw already keeps raw tokens away from the model. Credentials listed in capabilities.json are never passed into the WASM module's memory. The proxy intercepts the outbound request and injects the auth header after the module has already made an unauthenticated call. But the secret isn't the sensitive part of an agent run. The tool output is. Whatever the agent read before deciding what to do next goes through the model in full. So the credential boundary and the inference boundary are two different problems, and Intel Trust Authority speaks only to the second. I haven't worked out whether the proxy itself sits inside the same attested environment.