In God i Trust || ai and crypto
The Loop
Joined May 2022
- Tweets6.9K
- Following1.3K
- Followers1.8K
- Likes23.4K
Pinned Tweet
Paying a team in crypto still looks like this: open wallet, paste an address, send, repeat five more times. Then a spreadsheet at month end to explain it all.
So we built Remiglo: stablecoin payment and payout rails for businesses, on NEAR Intents.
Here is the problem it solves. A business that pays contractors or a team across chains has no operations layer. Invoices go out by hand. Payouts are one wallet transfer at a time. Big payments move with no second pair of eyes. And at month end, somebody opens a spreadsheet and tries to match what was sent against what actually settled.
Remiglo is that operations layer, in one dashboard and one API:
β’ Invoices with hosted payment pages and QR codesβ¨β’ Payouts and payroll batches, uploaded by CSV or built by handβ¨β’ Approvals: anything over 500 USDC waits, and no coin moves before an admin approves it on the batch pageβ¨β’ An address book of recipients, and payroll that runs itself weekly or monthlyβ¨β’ Webhooks for every state changeβ¨β’ Reconciliation that matches what you sent against what settled, with a CSV export
In the demo video below, a customer pays a 45 USDC invoice, then a 540 USDC payroll batch for six people is approved and settles row by row, every row quoted and tracked on its own timeline, the approval itself written to the record.
How it works on NEAR: settlement runs on NEAR Intents, so USDC paid on NEAR can land as USDC on Base. A registry contract, live on testnet at remiglo.testnet, holds each business's spending policies (caps, budgets, approval thresholds) and records every batch on chain as an audit trail.
And it is non custodial. Each business funds its own payments from its own wallet to a per payment deposit address. Remiglo never pools funds.
Built at NEAR HACKATHON @NEARProtocol #NEARHackathon #NEARProtocol
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
20,000+ have claimed their spot on the Waitlist.
Three steps to get in.
1: Claim your spot: app.arch.network/waitlist
2: Share with friends to rank up
3: β (the most important one)
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
βPrime brokers (PBs) are critical market participants in tradfi markets. They provide leverage, finance trades, and enable hedge funds to use a single pool of capital across markets and strategies, rather than posting separate collateral at each venue.
It's an absolutely gigantic business.β
βTraditional PBs mostly serve large institutions because dealing with smaller customers is operationally expensive. Smart contracts can automate much of that work and offer it to anyone, from someone trading $500 to a pod at Millennium running a few hundred million.β
βBuilding the software to manage this is increasingly easy, whereas the hard part is having the balance sheet to finance those positions, a low cost of capital, and the risk-management know-how to avoid blow-ups.β
New essay out on the @multicoin blog, which is an evolution of our thinking on DeFi value capture
TLDR is in a world where software costs trend to zero, finance is one of the most obvious industries where moats will exist and DeFi is perfectly positioned
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Sign ups are now open. We will be onboarding users to Arch Primeβs mainnet beta according to this list π
You've spent years fighting for your portfolio.
Now, let it fight for you.
Join our Waitlist: app.arch.network/waitlist
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Paying a team in crypto still looks like this: open wallet, paste an address, send, repeat five more times. Then a spreadsheet at month end to explain it all.
So we built Remiglo: stablecoin payment and payout rails for businesses, on NEAR Intents.
Here is the problem it solves. A business that pays contractors or a team across chains has no operations layer. Invoices go out by hand. Payouts are one wallet transfer at a time. Big payments move with no second pair of eyes. And at month end, somebody opens a spreadsheet and tries to match what was sent against what actually settled.
Remiglo is that operations layer, in one dashboard and one API:
β’ Invoices with hosted payment pages and QR codesβ¨β’ Payouts and payroll batches, uploaded by CSV or built by handβ¨β’ Approvals: anything over 500 USDC waits, and no coin moves before an admin approves it on the batch pageβ¨β’ An address book of recipients, and payroll that runs itself weekly or monthlyβ¨β’ Webhooks for every state changeβ¨β’ Reconciliation that matches what you sent against what settled, with a CSV export
In the demo video below, a customer pays a 45 USDC invoice, then a 540 USDC payroll batch for six people is approved and settles row by row, every row quoted and tracked on its own timeline, the approval itself written to the record.
How it works on NEAR: settlement runs on NEAR Intents, so USDC paid on NEAR can land as USDC on Base. A registry contract, live on testnet at remiglo.testnet, holds each business's spending policies (caps, budgets, approval thresholds) and records every batch on chain as an audit trail.
And it is non custodial. Each business funds its own payments from its own wallet to a per payment deposit address. Remiglo never pools funds.
Built at NEAR HACKATHON @NEARProtocol #NEARHackathon #NEARProtocol
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Running finances on full auto is scary - but its the future.
Pakt - our guardrail system - was created out of internal demand. From automating rebalancing, treasury management, payroll...
Setting up an agent is easy. Doing is safely with peace of mind - now easy with Pakt.
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
NEAR Hackathon track: AI Swarms.
β
Coordination markets, bidding, collective purchasing and intent-based settlement between agents.
β
Build the economies that emerge when agents work together.
β
Start here: nearhackathon.com
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Update: On November 8, an attacker used a vulnerability in SGLang, the inference engine we run, to access part of our infrastructure. Weβve closed the access and fixed it.
Affected users have been contacted. If you havenβt heard from us, no action is needed.
near.ai/blog/security-incideβ¦
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Only person the markets approve Jensen touching is @ilblackdragon
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Update: We have closed the unauthorized access to NEAR AI internal systems and are completing our investigation.
NEAR AI Cloud stays paused until we can restore it safely.
NEAR Protocol and other NEAR services are not affected.
Next update in two hours.
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Inference providers are under attack.
AI inference is now playing a critical role and is a honey pot of data and ability to inject actions back into users devices via agents for malicious actors.
NEAR AI thwarted a series of attacks on its systems and is back to normal. NEAR Protocol, Intents and other infra wasnβt affected.
TEE architecture is part of the answer to making sure that from outside even inference provider doesnβt have an access. User should verify exactly what happens with their data and prompts.
NEAR AI code is open and inspected by many partners when deciding to leverage the stack to validate exactly what runs inside TEE and ensuring privacy.
In this case there was also a vulnerability within SGLang, open source inference engine that runs inside an enclave. Even though this had a minimum disruption it still was a very
Going forward we are going increase security rigor of dependencies and also improve a βturnstileβ to detect when anything out of normal runs inside TEE.
Proactive security against external attackers has became a must. We have been using Abliterated models and red teaming harness to continuously monitor and detect issues. We are also expanding scope and compute invested into this.
Update: On November 8, an attacker used a vulnerability in SGLang, the inference engine we run, to access part of our infrastructure. Weβve closed the access and fixed it.
Affected users have been contacted. If you havenβt heard from us, no action is needed.
near.ai/blog/security-incideβ¦
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Ambassador applications for @Arch_Prime just closed.
If you missed it, you haven't missed the story. You're still early.
In the last few weeks Arch:β¨β launched the new Arch Prime lookβ¨β published the full-stack onchain bank thesisβ¨β acquired Horizon, so your crypto can soon help you buy a home
Next up: the Arch Prime waitlist opens soon.
The best opportunities in crypto usually come before everyone else knows about them.
@Arch
Live today: the API, the TypeScript SDK, the full dashboard with accounts and roles, and the testnet registry. Next: mainnet and the first pilot business.
Code: github.com/lildosnear/Remiglβ¦
If you pay people in stablecoins, what does your month end reconciliation look like?
53
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Keep a date with us tomorrow
Circle Stewards from local chapters report on what their Circles have been doing.
Come if you're in a Circle, or if you're curious about the movement side of Logos and have questions.
teapaw.org/i/spaces/1lgyMjbqPKWPbβ¦
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
@NEARProtocol Hackathon participants!
Feel free to write and ask for any advice on using technology from @near_intents @auroraisnear.
This will greatly help you implement cross-chain functionality and also expand the project's capabilities.
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Your AES-256 doesn't matter if your key is sitting in a public GitHub repo.
β¨Key management in 5 stages: generate β distribute β store β rotate β destroy.β¨
Where are your keys right now?
You wake up
You haven't even confirmed if you can speak
You are already doing "npm run dev"
GM CT π«‘
6
14
166
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
Join us tomorrow for our AMA on Arch Prime and Horizon.
Weβll cover why we acquired Horizon, how your crypto and your home can work together, and whatβs next for Arch Prime.
π Mon, Nov 9
β° 1pm EST
Set a reminder π
teapaw.org/i/spaces/1broygqPQWmaOβ¦
π³Μ·πΜ·πΜ·πΜ·πΜ·πΜ·πΜ· β¬ retweeted
The smartest AI isnβt the one thatβs always sure. Itβs the one that knows when to change its mind.
This is called Non-Monotonic Reasoning.
Imagine telling an AI that birds can fly and Tweety is a bird. It concludes Tweety can fly.
But what happens when it learns Tweety is actually a penguin?
A smart reasoning system should revise its earlier conclusion based on the new information.
This is an important concept in AI reasoning, knowledge representation, and intelligent systems.
When building AI, donβt just ask what it knows. Ask how it responds when it learns something new.
π¬ Have you ever seen an AI stubbornly defend a wrong answer?