frol.near

Rust and NEAR Protocol, NEAR AI, IronClaw technical evangelist building @TrezuApp

[frol.near @Freol]

Sharding is like shelves in the library, where books starting with A-D are known to be on the first shelf, and so you don’t need to scan the whole library.

If you haven’t designed the shelves from the day one, your library will never be efficient.

This is exactly how NEAR is different from any other chain

[Vadim (AI, ⋈) @zacodil]

Nice to see people loving NEAR's dynamic resharding. But I want to flag what it actually is: not a 2026 feature, but the payoff of a bet NEAR placed on day one, to shard the chain.

Sharding splits the chain into parallel lanes so no single lane jams. But there's a catch. To check a lane, a validator (a machine that verifies the chain) normally has to store everything in it, so the number of lanes is capped by what one machine can hold, and adding a lane meant a slow, coordinated, network-wide upgrade. Real sharding needs validators that can verify a lane without storing it. That was the missing piece for years.

NEAR shipped it in 2024, and that's the actual breakthrough: stateless validation. A validator now checks a lane from a small proof package, a slice of data plus a cryptographic receipt that it's correct, and never stores the lane. The correctness of a split is proven inside that package. So lanes stop being capped by one machine, and a lane can even split while the traffic is still moving. Dynamic resharding is that capability put on autopilot.

This is why almost nobody else has it. Solana, Sui, Aptos, Monad never sharded, so they never needed it: one big machine does everything. That has a cost, though. A single machine has a ceiling, so the harder you push it the more brutal the hardware gets and the fewer people can run a node, and a big enough spike can stall the whole chain. Ethereum does need sharding's relief, but its original data structures make those proof packages huge, so getting there means migrating its entire live state to a new design, a multi-year job they keep redesigning, and once Ethereum pushed scaling onto L2s the urgency drained out. NEAR is the rare chain where sharding was the whole thesis, so solving this was do-or-die.

And the bet keeps compounding. NEAR adds capacity by adding lanes while staying one road, one shared state, no bridges. Ethereum scaled by splitting into 20+ separate roads (its L2s), and this March it had to launch a whole project just to glue them back so they can talk instantly again. NEAR never split the road.

So enjoy dynamic resharding. Whether the agent traffic it's built for actually shows up is a separate question. The engineering under it is a years-old bet finally cashing out.

[frol.near @Freol]

@near_intents $20B and counting!

[frol.near @Freol]

$20B

[NEAR Protocol @NEARProtocol]

All-time volume on NEAR Intents just crossed $20 billion.

This milestone is the result of a series of technical breakthroughs that removed some of the biggest constraints in cross-chain execution.

A deep dive on Intents' rapid momentum 🧵

[frol.near @Freol]

@mert Trezu

[Trezu @TrezuApp]

Privacy is not optional in finance.

A salary agreement is private. Vendor terms are confidential. Strategic timing belongs to the team making the move.

On a public blockchain, none of that exists.

Confidential Treasuries are changing that default.

[frol.near @Freol]

GOAT

[Trezu @TrezuApp]

Most crypto teams get hacked not because of bad tech. But because someone had access they should not have had.

Setting up roles and permissions correctly is the difference between a team that operates cleanly and one that is one compromised wallet away from losing everything.

Three roles. One clear separation.

Trezu gives you three roles: Requestor, Finance, and Governance.

Requestors submit payment proposals. They cannot approve them. Finance members approve and execute payments. They cannot change the rules. Governance controls the treasury configuration. Adding members, changing thresholds, updating rules.

If one person holds all three, you do not have a multisig. You have a single point of failure with extra steps.

Thresholds matter as much as roles.

A 3-of-3 setup sounds secure until one wallet is lost and the entire treasury is frozen permanently. A 1-of-1 means anyone on the Finance role can move funds alone.

The sweet spot for most teams is 2-of-3. Enough redundancy to survive a lost key. Enough friction to prevent unilateral action.

What this looks like at different stages.

Early stage: Finance at 1-of-2 on everyday wallets. Governance at 2-of-3 on hardware wallets only.

Growing startup: Finance at 2-of-3 held by Finance Lead, Ops Manager, and CEO. Governance at 3-of-5 with founders and board.

Active DAO: Finance at 3-of-5 for your ops working group. Governance at 4-of-7 to prevent hostile takeovers.

When to update your setup?

New hire in a financial role. Someone leaving. A new funding round. A promotion to leadership.

Any of these should trigger a review. Your treasury should grow with your team.

Getting this right from day one costs nothing. Getting it wrong can cost everything.

[frol.near @Freol]

@mnaeraxr It was hosted on our Ukrainian DOTS platform running UCUP-2026 (an offline Kharkiv universities cup). We didn’t really publish much online about it.

I will share the links in DM

[frol.near @Freol]

Since the problem starting doesn’t set the constraints, it forces the participants to think why int a, b; cout << (a*b); would give a wrong answer and guide the LLM through the prompt to take that into account. The actual solution needs to implement the fastest big numbers multiplication algorithm and it is not feasible for majority of the participants to know how to code it, but it was quite straightforward to instruct the LLM to use “the fastest multiplication algorithm for arbitrary precision numbers”

[frol.near @Freol]

We recently run a fun new format - vibe coding contest for competitive programmers. It was a challenge to author the right tasks that won’t be one-shotted by LLMs indeed, but it is not impossible.

Participants were given the problems in a regular competitive programming contest format with a bit underdefined constraints (just like most of the problems we deal with in the real world), but the solutions must be the prompt that would generate C++ solution and that code is then being tested. Obviously, the first submission attempts were the copy-paste of the problem statement, but that would only give 10 points out of 100.

The first task was to compute A*B, but int a, b; is not enough and participants would have to think why they get WA reports. Think what you would prompt before checking the solution in the next thread message

[frol.near @Freol]

@TrezuApp https://t.co/CD3mf0A16I

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

Running payroll? Paying invoices? Receiving payments?

Historically doing this on-chain meant all the key company information was visible to everyone in the world.

With @TrezuApp using confidential intents you can now keep this information private as it's suppose to be.

[frol.near @Freol]

Bitcoin tried to normalize public balances, but it never sounded sane to majority of people, so confidential crypto is rapidly taking off

[Trezu @TrezuApp]

Running payroll on a public multisig means every salary is public.

Every vendor payment. Every trade size.

Not every team signed up for that level of transparency. Trezu lets you choose.

Public treasury for open governance. Confidential treasury for everything else.

[frol.near @Freol]

The chart that speaks for itself

[Alex Shevchenko 🇺🇦 @AlexAuroraDev]

3 months ago we shipped Confidential Intents on NEAR.

This chart is the single most important metric for Confidential Intents. 🧵

[frol.near @Freol]

Crypto adoption is NEAR

[Alex Shevchenko 🇺🇦 @AlexAuroraDev]

3 months ago we shipped Confidential Intents on NEAR.

This chart is the single most important metric for Confidential Intents. 🧵

[frol.near @Freol]

@AlexAuroraDev 🎉🎉🎉

[frol.near @Freol]

@ilblackdragon @Cameron_Dennis_ @dragonisnear Best way to express it indeed!

Our core is the mission: bring the value to the humanity, to every single person on the earth

[frol.near @Freol]

@MarcinRedStone @StablREuro https://t.co/hqRCo2hz3Y

[frol.near @Freol]

1-of-3 multisig significantly increases the risks for the funds!

The main reason I hear people doing it is how slow the governance is otherwise. And that is mostly due to terrible UX users have to deal with unless they know about @TrezuApp

[frol.near @Freol]

1-of-3 multisig significantly increases the risks for the funds!

The main reason I hear people doing it is how slow the governance is otherwise. And that is mostly due to terrible UX users have to deal with unless they know about @TrezuApp

[Marcin Kazmierczak ♦️ @MarcinRedStone]

Stablecoins from @StablREuro (EURR and USDR) got exploited for ~$10M.

Both have depegged by over 20%, with the attack reportedly still ongoing.
Another key compromise attack, this time a 1-of-3 minting multisig.

Seems like no weekend this year without a hack...

[frol.near @Freol]

@arthur0x 100% and @TrezuApp built on top of this foundation bringing the confidential non-custodial multisig. Check it out

[frol.near @Freol]

@rhea_finance 🎉

[frol.near @Freol]

@ilblackdragon @NeerajKA 100%
I have onboarded 80+ people to https://t.co/d2HrxK0RwH today at Vienna Bitcoin Pizza Day and everyone was shocked how simple that was