@banamlas@SECGov@The_DTCC You said indeed. We built the indeed. One KYC, follows the investor, scales offshore onboarding without the separate whitelist tax.
I can dm more info 🫡
Tokenization does not mean that tradfi is guaranteed to come buy your bags.
It means tradfi is using chains and wallets as distribution to sell you theirs.
This can be good. Public access on globally distributed networks, transparency, self-custody, etc.
But it's not a magical permanent bid unless the assets are fantastic and can justify portfolio allo.
@regintel2000 I see a lot of people saying we have to throw AI into the compliance back office.
…but there is a more fundamental problem that needs to be addressed
AI will hit a wall in the back office: Trust
Air France 447 didn't fall because the pilots lacked software. It fell because three iced-over pitot tubes fed the cockpit bad data. The instruments lost contact with reality, and the automation trusted the lies.
The answer was not more automation. It was better sensing: redundant, heated signals the system could trust before it acted.
Without verifiable inputs, finance is just processing fiction at machine speed.
Compliance, KYC, underwriting, market surveillance, credit risk. All still running on PDFs, screenshots, exports, and hand-assembled attestations.
So we layer agents on top:
read the PDF, click the portal, fill the form, route the exception.
It still doesn't know whether the underlying claim is true.
AI makes the cockpit faster. It does not fix the pitot tube.
@cryptoreine I take the point but I think we need to completely replace reg NMS.
It’s going to present many challenges to getting Defi tokenized securities.
…and it wasn’t even a good solution for investors in tradfi. Arguably it’s the root cause of payment for order flow.
Random but in 2016 I was convinced Trump was going to pull up to the RNC to accept the nomination
And Al Gore was going to unzip his Trump costume and proclaim “this is the amount of dedication you need to be president of the United States”
Giving every Blockbuster employee a PC in 1998 didn't produce Netflix.
AI inside the average enterprise is the same bet. New engine, old chassis. Bolt it onto the org chart you already have and you get a faster version of the thing that was already dying.
The gains are real. But they won't accrue to incumbents, because capturing them means rebuilding the company from scratch. And no incumbent eagerly rebuilds the machine that pays everyone's salary.
That's the opening.
The giants can see the door. They just can't walk through it.
You can.
Build the company that assumes the tech, instead of the one that bolts it on.
Blockchains do things trad rails can't.
But none of that matters until a network actually forms.
A neutral, composable state machine with nobody on it is a ghost town with great plumbing.
Credible neutrality was never the hard part. The hard part is giving people a reason to show up before it's worth anything.
@maurelian_@jbrukh@AFDudley0 Proof of work only works if it’s cheap/fast to verify an expensive computation right?
I think it’s harder than just solving the bench
4% to borrow against your S&P 500. 20% on a credit card. Same person. That gap is what decentralization is worth.
Visa handed ownership to a club of banks. The toll didn't move, because they were the ones charging it.
You can give up control late. But the toll, once it's flowing, never comes off.
Leave the gatekeeper standing and the market fragments around it. Every time.
Morpho is the example that proves Dmitriy right.
It didn't isolate risk. It cherry-picked the profitable matches and left Aave holding the residual, on reserves Morpho never capitalized.
Aave eventually noticed. That's why Morpho users quietly disappeared from Merit.
"Utility outweighs security" sounds nice until you realize the utility was funded by Aave depositors.
You built the proofs, Eli. You never stopped to ask what they actually prove.
ZK proves a computation ran correctly. It can't tell you the inputs were real. Feed it a lie and it proves the lie, flawlessly.
Finance's problem was never the math. It's where the number came from. Pull the covenant directly from the bank's server instead of a quarterly mark the manager grades himself, and the opacity premium private credit feeds on disappears. You fund cheaper than Apollo. You take every deal they wanted.
You built the cockpit and never heated the pitot tube. That was the half that flies the plane.
AI will hit a wall in the back office: Trust
Air France 447 didn't fall because the pilots lacked software. It fell because three iced-over pitot tubes fed the cockpit bad data. The instruments lost contact with reality, and the automation trusted the lies.
The answer was not more automation. It was better sensing: redundant, heated signals the system could trust before it acted.
Without verifiable inputs, finance is just processing fiction at machine speed.
Compliance, KYC, underwriting, market surveillance, credit risk. All still running on PDFs, screenshots, exports, and hand-assembled attestations.
So we layer agents on top:
read the PDF, click the portal, fill the form, route the exception.
It still doesn't know whether the underlying claim is true.
AI makes the cockpit faster. It does not fix the pitot tube.
You found the tell everyone will scroll past. The root wallet.
You never really owned a stock certificate. You owned a line in a ledger that a registrar could edit. The entry was ownership. The paper was just a receipt.
The root wallet is that edit power rebuilt as code. So the trusted middleman blockchains were built to kill didn't die. It got a key and moved inside the asset.
Every tokenization pilot from here is a fight over who holds that key.
Intermediary-centric vs Investor-centric
We either revolutionize the capital markets or give the existing machine better margins.