One of the stranger press releases of the crypto week came out of Hong Kong on October 13. NFT Limited, a company listed on NYSE American under the ticker MI and formerly known as Takung Art, announced a proposed technical protocol called RWAIX, open-sourced its reference implementation on GitHub under an MIT license, and targeted a limited real-world pilot for the first quarter of 2027.
The company operates an online trading platform for international art and collectibles, and its pitch speaks directly to one of the loudest debates in digital assets: how do buyers, sellers, and increasingly AI agents and bots verify who they are dealing with, agree on what was promised, and confirm that delivery actually happened?
What RWAIX Actually Does
According to the announcement, distributed via PR Newswire and reported by ChainCatcher, RWAIX is designed to let transaction parties, including autonomous software agents, verify identity, agree on the content of a deal, and confirm that delivery occurred. The stated first use case is the company’s own existing digital art platform, operated through its wholly owned Hong Kong subsidiary Metaverse Digital Payment Co., Limited.
The open-source reference implementation includes Node.js core protocol logic, 50 automated test cases with expected outputs, schema and API documentation, and a browser demonstration using synthetic data. The company says internal, self-administered test records showed all 50 tests passing with no failures.
That last qualifier deserves emphasis. These are the company’s own tests, run by the company, on an unreleased protocol. An independent security review is still pending, and NFT Limited itself notes that RWAIX has not been integrated into any production environment and that timelines may shift.
The Pilot Plan
The proposed pilot is deliberately small. Rather than launching a new marketplace, the company plans to upgrade a single existing NFT order, walking it through the full cycle of consent to purchase, delivery, independent verification, payment, and dispute handling, using its existing banking channels. Chief Executive Officer Yanying Wang said the company already has customers, orders, and payments, and wants to use the next 90 days to test whether the protocol works on real customer orders.
If that limited pilot succeeds and the security review clears, a 90-day live trial would begin, with the first quarter of 2027 as the target window. If it does not, RWAIX joins the long list of announced protocols that never touched a production order.
Why an AI Agent Transaction Protocol Matters
Strip away the corporate packaging and the underlying idea is timely. AI agents are already moving real money in crypto. Chainalysis data highlighted this week showed AI agents have spent more than 138 million USD on the Base network, and payments passing through AI-driven flows are one of the fastest-growing categories on major chains. What is missing is a standardized way for a bot to prove a deal was delivered as agreed before payment releases.
That gap is exactly where agent-commerce proposals live. A protocol that lets a machine buyer verify an art file, a digital collectible, or eventually a real-world asset arrived as described, and trigger payment automatically, would be a building block for autonomous commerce regardless of who ships it. NFT Limited’s twist is applying the concept to art and collectibles trading, a niche with genuine fraud and authenticity problems that verification rails could address.
Reasons for Skepticism
The caution list here is long, and pretending otherwise would be a disservice. This is a proposed protocol, not a product. The tests are self-administered. The company, formerly Takung Art, has a history of riding narrative waves, from art trading to metaverse branding to NFTs to AI, and the RWAIX announcement follows that pattern of repositioning around whatever narrative dominates the moment. The pilot involves exactly one order. And a stock ticker whose identity is the acronym NFT announcing an AI agent protocol in the same week the market is euphoric about agents is timing that deserves scrutiny.
Open-sourcing under MIT is a genuinely positive signal, since anyone can audit the code, run the 50 test cases, and judge the schema design directly. That is the right way to float a protocol. But open code with no production integration is a promise, not a proof.
The Bigger Picture for NFT and RWA Markets
The NFT market that NFT Limited serves is far smaller than it was, with monthly trading volume in the hundreds of millions of dollars and active wallets near 505,000 in early 2026, roughly 42 percent of the 2022 peak. Yet real-world asset tokenization keeps attracting institutional infrastructure, and verification layers are a common denominator in both. If machine-to-machine commerce keeps expanding, someone will build the trust rails RWAIX describes; whether it is a small NYSE American-listed art platform or a major protocol team is the open question.
For now, the scoreboard reads: 50 passing tests, zero production orders, one promised pilot. The 90-day clock, if the company holds to it, runs out around the turn of the year. Watch GitHub activity and the independent security review; both are harder to fake than a press release.
Disclaimer: This article is for informational purposes only and does not constitute financial advice. Cryptocurrency markets are highly volatile, and readers should do their own research before making any investment decisions.
The bank rails choice is doing more work than people notice. Agents settling through existing dispute channels skips the entire onchain arbitration mess.
138 million dollars already spent by AI agents on Base and still no standard way to verify delivery before payment releases. The gap RWAIX aims at is genuinely real
Real gap, questionable messenger maybe. Still, the MIT license means anyone can pull the repo and run the 50 test cases themselves, that part is fair
Takung Art rebrands to NFT Limited, then announces an AI agent protocol right as the market goes agent crazy. The narrative pivots tell you everything
The pilot is one single order though. Even if it clears, scaling art verification to actual RWA flows is a completely different sport
one order scaling to real rwa flows is the whole gap, agreed. still, a public spec with schemas and test cases beats every closed pilot that runs a demo chain and prints a press release
schemas being public means anyone can fork the spec before the pilot even reports. if fill rates are bad the repo still becomes the base layer someone else ships
forks without real escrow flow are just github stars. the 90 day fill rate report is the only scoreboard that matters here
scoreboard point stands but public schemas let anyone compete on fills too. pilot failing with the spec surviving is still a net win
one clean order proves the plumbing works and nothing else. give me a hundred messy ones before i care about the spec
one order plus 50 test cases is a hobby repo until the 90 day window reports real fill rates. if takung publishes the numbers i will eat my skepticism plate and all
fair plate of skepticism, but the mit license means the spec outlives the pilot either way. someone else ships it if takung cant
fill rates plus dispute resolution stats or it didnt happen. an order that clears but takes three weeks of arbitration to release escrow is a failed demo in disguise
three weeks of arbitration per order is why dispute stats matter more than fill rates. fast settlement without recourse is just a faster way to lose
same skepticism but one order through a real bank with consent, delivery, verification and dispute steps documented beats a demo chain. the plate stays full either way lol
the takung pivot timeline is wild, art trading platform to NYSE ticker MI to AI agent protocol. at least the MIT license means the repo can outlive the narrative
an art platform releasing the verification layer instead of keeping it closed is rare. respect to NFT Limited for that alone
running it through existing banking channels is the detail here. no stablecoin settlement theater, actual bank rails
bank rails means disputes get real recourse too. stablecoin escrow with no finality arbiter is just code roulette for anything above pocket money
Recourse exists in banking rails but it is slow. The protocol’s whole job is proving delivery before release so disputes shrink. Arbitration speed is the metric to watch, not just fills.
50 self administered tests passing means very little before a third party audit. open sourcing it was the right move, now lets see an independent security review before the q1 2027 pilot
not the CEO lol but 90 days of real orders is a smarter test than most pilots that just run a demo chain and call it done
mit license is the only reason to take this seriously. closed source pilots with the same one order track record are a dime a dozen
The detail nobody quotes is that the pilot order walks through dispute handling using existing banking channels. If that single order ends in a real dispute, the 90 day report becomes the most interesting document of Q1 2027.
the dispute handling part is the hard bit and everyone skips it. bots agreeing on delivery before escrow releases, that is the actual product
Base payments growing fast but nobody standardized machine to machine deal verification. first usable spec wins this niche
machine to machine trade needs receipts more than it needs blockchains. shipping the verification layer openly under MIT is the right call even if this never scales past art
receipts need a final arbiter though. bank rails move the money fine, but who rules when an agent disputes its own order log