Full-Stack Blockchain Dev | Developer Relations | Exploring AI, Privacy & On-chain Systems | Prev @Hypersignchain @NFTFN | Public Speaker
onchain
Joined April 2023
- Tweets10.8K
- Following845
- Followers4.2K
- Likes44.3K
Pinned Tweet
Hii, I'm Saber and I'm currently 19.
Sooon enough I'll be one of the biggest web3 Developers in Africa.
when you Hear Saber you think of web3 development
clock it.
Privacy tech for Blockchains broken down for you:
➺ Zama (FHE): The data is public, but it's encrypted.
➺ ZKPs: The data is hidden, but the proof is public.
➺ TEEs: The data is processed in a hardware black box.
➺ Canton Network: The data only exists between the two parties transacting.
FHE encrypts your footprint.
ZKPs hide your footprint.
Canton ensures you never leave a public footprint at all.
Happy new week everyone!
let’s do amazing things <3
need to close out few of my builds today + send in a couple of job offers.
Saber.eth | Blockchain Developer⚡ retweeted
There’s a blind spot among young professionals today. Many don't really understand what exactly is stopping them from getting the outcomes they want.
I saw it more and more when I facilitated a 45-minute masterclass at BLOCKFEST Africa. I talked about attention and impact in building productive communities. Then we put it to work. Because what is a masterclass without real-life problem-solving?
We first diagnosed a marketer with massive engagement on her pages but no clients nor gigs. We found out that she wasn't communicating enough evidence of what she does, why it matters, and why she should be trusted with the work.
The second was a founder who couldn't explain his product in one sentence because it was trying to do too many things at once. Fix one problem first, scale it, then begin to add the rest.
A community manager struggling to grow beyond referrals. We diagnosed that the referral incentives weren't tied to the actual activities they wanted members doing.
The questions from the audience kept coming, pointing to how many builders are struggling with distribution, PMF, and more.
The marketer is now going through private mentorship with me. After the session, students building their own projects came to me for guidance on navigating life after school. Community builders and founders exchanged contacts with me.
Someone even gave me $100 towards my ride. 😭❤️
It was very interesting to see faces behind pfps again.
Special thanks to @thesarahidahosa, the engine behind BLOCKFEST Africa 2026. Thank you for the invite, Queen
Thank you, @blockfestafrica ❤️
Every Chog has a look. Now give it a life.
Introducing CHOG World, my #Chogathon build for @ChogNFT on Monad.
CHOG World turns your Chog Genesis into an AI character with its own wallet and its own mind. It quests, picks fights with rivals and earns on its own, a few times a day, inside limits you set onchain.
How it plays:
• Quests and arena fights settled from a block nobody can predict
• Every fight replayed in your browser and checked against the chain
• Levels, titles, scars and gear change how your Chog looks
• Name a Grudge and your Chog goes after that rival
• 10% of every solana:GYFGHK9eiuwB8UJ3FqTSpGeUsykTxH97MfXBGDchpump fee is burned
How it stays safe:
• Your NFT never leaves your wallet
• You grant a rules contract, never a server
• Every move is checked onchain against your budget
• Sell the Chog and its agent stops, while its coins, gear and titles go to the buyer
Live on Monad mainnet, open to the first 25 Chogs:
chog-world-mainnet.vercel.ap…
No Chog? Try it free on testnet:
chog-world-zeta.vercel.app
full demo: youtu.be/X8fre646yA8
Awaken yours and share its card. Let's see whose Chog climbs the leaderboard first 👀
Everyone’s talking about building on NEAR lately, but why?
> Sharded architecture built for scalability.
> Human-readable accounts and better UX.
> Chain abstraction that connects assets across multiple blockchains.
> NEAR Intents that let users and agents express outcomes instead of manually handling every transaction.
> An ecosystem betting big on the intersection of AI and crypto.
The interesting part? NEAR isn’t just thinking about where crypto is today, but how AI agents will use it tomorrow.
Might be time to pay attention to NEAR too!!
I’ve migrated to $NEAR
I’ll create content on NEAR.
I’ll write and deploy smart contracts on NEAR (Rust, JS and more).
I’ll onboard more people to NEAR.
I’ll find the next 100x on NEAR.
NEAR is the future.
Stop blaming the LLM for hallucinating. Your model isn't stupid. You are just dumping 50 random files into a context window and hoping for magic.
If your AI agent is writing garbage code, your context management is garbage.
Memory is an architecture, not a chat history.
When you build a production agent, you need three distinct memory layers:
> Core Memory (Pinned): System prompts, fixed tool schemas, and strict operating constraints. This never leaves the context window.
> Archival Memory (Vector): Documentation and massive codebases. The agent should only see this when it explicitly fires a search_memory tool.
> Rolling Summaries: When the context window hits 80%, you don't just blindly trim the oldest messages. You use a secondary, cheaper LLM to summarize the workflow history, inject the summary, and flush the raw logs.
I'm bullish on AI agents using cryptoespecially when privacy is built in.Every AI agent needs:
-> a private/self-sovereign identity
-> a wallet it fully controls
-> the ability to transact autonomously (without leaking activity)
Banking is slow, permissioned, and fully surveilled. Crypto gives agents wallets, fast settlement, and permissionless access to money, plus privacy tools (ZK proofs, confidential txs, stealth addresses) so they can operate without broadcasting every move.
As agents become more independent, crypto becomes both their financial layer and their privacy layer.
Vibe Coding 101. It's still that simple bro.
➺ Stop watching beginner prompt tutorials
➺ Install your agent harness of your choice (e.g., LangGraph, Letta, or AgentCore — just a quick heads-up: Cloudamize is actually a cloud TCO and migration analytics platform, not an AI agent harness!)
➺ Wire up MCP servers to give your agent hands
➺ Structure your repo for strict context management
➺ Set up smart contracts as your agent's permission layer
➺ Feed compiler errors back into the agent loop via LSP
➺ Orchestrate parallel agents (Coder + Critic + Tester)
➺ Build automated eval harnesses
➺ Go back to “deploy something real” and repeat
Everyone is hyping up AI agents trading on public ledgers.
Here is the reality: If you deploy an autonomous trading agent or an AI treasury manager on a fully transparent blockchain, every single competitor can read the contract, front-run the agent’s strategy, and track its exact balances.
Enterprise-grade AI agents cannot survive on transparent ledgers.
They need privacy. They need state synchronization without broadcasting their underlying data to the entire world.
This is exactly why my research focus is shifting toward Daml and the Canton Network. It is the infrastructure actively solving the privacy vs. interoperability trilemma for institutional capital.
I'm currently building out Showdesk (a mandate-bound treasury execution agent) to test these exact data boundaries.
Check it out:
App: shadowdesk-inky.vercel.app
GitHub: github.com/Saber1Y/Shadow…
Demo walkthrough:youtu.be/xdd0VtBYofo
CC: @CantonNetwork @digitalasset @AppsFactory_CC @CantonFdn
I think we’re focusing too much on hackathons and not enough on the bigger problem.
Hackathons are supposed to accelerate your building journey. They’re a chance to test ideas, meet people, get feedback, and maybe win some funding.
But a hackathon is not a job. It’s not guaranteed income. And it shouldn’t be the primary thing you’re depending on as a builder.
The real question is: what happens after the hackathon?
Where are the jobs? Where are the teams hiring? Where are the opportunities for builders to turn their skills into sustainable careers?
Because with AI, the barrier to building software is getting lower. You can go from an idea to a working prototype faster than ever.
But building something and getting paid to build something are two very different problems.
We need more conversations about how Web3 creates sustainable opportunities for builders, not just how many hackathons we can organize or how much prize money is up for grabs.
Hackathons should be a launchpad, not the entire runway.
Bro, you’re my mutual, but I have to say this. This screams entitlement
Are you sharing hackathons because you genuinely want to help people find opportunities, or because you expect winners to tip you? It was until recently you started sharing hackathons opportunities
Because going by your logic, do you know how many people like @kenn_ronin, @0x_beni_ and @winsznx have helped by sharing hackathons early
Even me, with all the AI and AI agent tips I share, do you realize how many people I’ve helped? If we all followed your logic, we’d be millionaires and retired by now lmaoo
So because someone saw a hackathon on your page, participated and won $5k, you suddenly expect him to give you something And because he didn’t, you won’t repost his content now
How does that make sense?
People have won four and five figures from opportunities shared by Tim, Ronin and others. Imagine if everyone who shared these opportunities started expecting a cut whenever someone won
Someone even messaged me recently to say he used one of my AI tips to win a hackathon. I was happy for him, and I still share tips every day. Why didn’t I ask him for a tip or refused to engage him again cause he didn’t tip me
And I saw someone in the comments saying that’s how you show gratitude
What gratitude are we talking about here?
It would’ve been a different conversation if you had personally sent him the hackathon privately or helped him win the hackathon
But you posted the opportunity publicly on your page. He saw it, participated, won and said thank you
Funny enough he may even have it in his mind to appreciate you , most likely the money has not arrived yet ( most hackathon wins money take months to process)
That thank you alone is gratitude. He acknowledged you, and that should be enough
The moment you start expecting something beyond that, you’re attaching conditions to what you originally shared publicly
If you’re genuinely happy to see people win from opportunities you share, why should a tip determine whether you celebrate them
Nobody is saying people shouldn’t appreciate those who help them. Of course they should. And if someone voluntarily decides to send you something, that’s a kind gesture
But expecting it and then holding it against them when they don’t. That’s entitlement
Saber.eth | Blockchain Developer⚡ retweeted
I am back on YouTube.
My mother just subscribed my YT!
W 🥳
Saber.eth | Blockchain Developer⚡ retweeted
BEEN SCROLLING 𝕏 FOR 30 MINUTES
software is dead
distribution is moat
AI will replace devs, designers, editors
bro like can we have an original thought
Can i get a job with this Github?
link: Github.com/Saber1Y
Meet SpendArc (@SpendArc8): programmable spending controls for autonomous AI agents. Live on @arc mainnet.
Agents need to move money to be useful. Hand one an open wallet and a single bad prompt, hallucination, or runaway loop can drain it. SpendArc fixes that.
How it works:
> Create a vault owned by your wallet. The agent gets a scoped API key, never your funds.
> Set a leash: per-tx cap, daily cap, allowlisted payees, expiry.
> Every payment is checked twice. A server policy gate rejects off-policy requests before they're ever signed, then the vault re-checks the full policy on-chain before a single unit of USDC moves. Blocked actions move nothing and are recorded on-chain.
Proven on @arc mainnet, same agent, one variable:
✅ Approved 0.5 USDC (under the cap)
⛔ Blocked 2 USDC (over the cap), nothing moved
SpendArc has an in-app mainnet/testnet switch. Try the full flow on Arc testnet first, then flip to mainnet. Each agent is bound to the network it was created on, so your test agents and real agents never mix.
Built on @arc and native USDC for the Arc Microgrants program by @circle.
Demo (testnet walkthrough): youtu.be/brlCHLD2qnQ
Live app (mainnet): spend-arc-two.vercel.app
Code: github.com/Saber1Y/SpendArc
Factory (Arc mainnet): explorer.arc.io/address/0xd7…
@arc @circle @DoraHacks
SpendArc is live on @arc mainnet.
Hand any AI agent a wallet that holds nothing: it can only ever spend inside policy. Caps, allowlists and daily limits, all enforced on-chain.
Try it 👉 spend-arc-two.vercel.app
Saber.eth | Blockchain Developer⚡ retweeted
We're building an open distribution network on @monad where anyone can help products grow and earn from the sales they generate.
Real sales. Trackable referrals. Automatic commission splits.
𝐒𝐏𝐎𝐊𝐄 is here, more details soon... gMonad 😁
@monad_dev @pareen @0x_eunice @spicytaco