Vynce

[Vynce @Ayomide_vynce]

I built Vectra, a Telegram-first autonomous AI assistant powered by Kite Agent Passport.

🔗website: https://t.co/srSSSgn53f

🔗 GitHub: https://t.co/TggMg9Xu5J

🎥 Here’s a walkthrough of how Vectra works.

@ChiZhangData @GoKiteAI @OpenBuildxyz

[Vynce @Ayomide_vynce]

@lammy_sol @Test_Sprite Your build has really being helpful

@Test_Sprite check this out

[Lammy.Sol @lammy_sol]

A follow-up to my GrantPath build 🎓

The goal remains simple: help students find scholarships and manage their applications with more confidence.

In my last post, I shared how I used @Test_Sprite to test GrantPath and catch a real bug before it could be merged.

Beyond that, I tested key user journeys across:

⇒ Scholarship discovery and eligibility
⇒ Profiles and saved opportunities
⇒ Application tracking and notifications
⇒ Privacy and API access controls

I wanted to go beyond simply integrating an AI testing tool. I wanted to see how it could help verify real user workflows and improve confidence in the application.

Building a product is one thing. Making sure its important user journeys work as expected is another.

Thanks @Test_Sprite for being part of the journey!

[Lammy.Sol @lammy_sol]

Finding a scholarship is one thing. Knowing whether you qualify, what you still need, and how to keep track of your applications is another.

That’s why I built GrantPath 🎓

A scholarship discovery and application tracking platform designed to help students move from searching for opportunities to taking the next step.

Here’s what you can do with it 👇

⇒ Discover scholarships by country, study level, funding type, and field of study.

⇒ Understand eligibility requirements, see what matches your profile, and identify what still needs confirmation.

⇒ Check official sources instead of relying on vague scholarship descriptions.

⇒ Save opportunities, add private notes, organize required documents, and track application progress.

I used @Test_Sprite to test real user journeys against the deployed app.

One test caught a bug where removing Ghana from a study profile left the old residence saved. The required GitHub check blocked the merge until the issue was fixed.

That’s the kind of feedback I wanted: not just a working demo, but testing that can catch real problems.

Try GrantPath: https://t.co/Rn7Wom50Hg

💻 Code + TestSprite proof: https://t.co/YXEfK7ZR3Q

𝑪𝒉𝒆𝒄𝒌 𝑶𝒖𝒕 𝑻𝒉𝒆 𝑫𝒆𝒎𝒐 𝑽𝒊𝒅𝒆𝒐 𝑻𝒐 𝑲𝒏𝒐𝒘 𝑴𝒐𝒓𝒆

[Lammy.Sol @lammy_sol]

Got to make this post…

Most .NET devs I know still don’t know about @AvaloniaUI, so here’s why it’s worth a look 👇

It’s a free, open-source UI framework that lets you build apps for Windows, macOS, Linux, iOS, Android, and even the browser from one codebase.

If you know WPF, a lot of it will feel familiar: XAML, C#, MVVM. And because Avalonia draws its own controls, you can keep the UI consistent across operating systems.

It works with modern .NET too, and Linux desktop support is a big plus.

It’s not magic, though. Tray icons, notifications, and startup behavior can still need OS-specific code. But for the UI itself, it can save you a ton of work.

I got to see this firsthand building Synups 👇
https://t.co/svl8PkZLcB

#AvaloniaUI

[Lammy.Sol @lammy_sol]

Ever had your internet drop and then had no way to prove when it happened or how long it lasted?

That’s what got us into this.

For the #AvaloniaPortChallenge, we took WirePeep, Bill Menees’ open-source network monitor, and rebuilt it as Synups.

From a Windows/WPF app to @AvaloniaUI + .NET, running on Windows, macOS, and Linux

Synups sends real ICMP pings, shows response times and peer status live, and never guesses when the result is unknown.

It records confirmed incidents, recovery, comments, and history that survives a restart and lets you export the evidence to CSV.

Know when your network drops. Know how long it lasts. Keep the evidence.

𝑮𝒊𝒕𝒉𝒖𝒃 𝑹𝒆𝒑𝒐: https://t.co/ZcymoQvABy

𝑪𝒉𝒆𝒄𝒌 𝑶𝒖𝒕 𝑻𝒉𝒆 𝑫𝒆𝒎𝒐 𝑽𝒊𝒅𝒆𝒐 𝑩𝒆𝒍𝒐𝒘 👇

@dotnet @MikeCodesDotNET
@jsuarezruiz @kekekeks

[Vynce @Ayomide_vynce]

@lammy_sol @monad @0x_eunice Amazing one Lammy

[Lammy.Sol @lammy_sol]

An AI agent can execute a payment. But how do you let it pay for work without giving it unrestricted access to your money?

That’s the problem I’m exploring with Confiance, built on @monad.

Imagine hiring a developer to complete a job. You want them paid when the required work is done, not simply because an AI agent decides to send money.

With Confiance, the owner defines the rules in advance: who can be paid, which token can be used, how much can be spent, when the authorization expires, and what milestone must be completed before payment.

The owner authorizes those rules once. The agent can then request payments within those boundaries.

Here’s where it gets interesting.

In my demo, the required milestone is a GitHub pull request merge. @chainlink CRE verifies the actual merge and reports the result onchain. If the milestone hasn’t been verified, payment is blocked. Even after verification, every other payment rule must still pass.

When the conditions are satisfied, the agent can execute the payment using its own @privy_io wallet, without requiring the owner to sign that individual transaction.

And if the agent exceeds its budget or the owner revokes its authorization, Confiance rejects the request.

The system also incorporates @TheCleanverse credential and asset-eligibility checks, Mera’s passkey-based owner account, and @envio_indexer indexed onchain activity history.

The idea isn’t to remove human control. It’s to let humans define the boundaries so agents can do useful work within them.

Because giving an agent a wallet and giving it permission to spend are two very fdifferent things.

Confiance: Give an AI agent a job, not your wallet.

[Lammy.Sol @lammy_sol]

The next $5 trillion in transactions won’t be human. But can agents transact trustworthily at scale?

I explore what already exists, what remains unresolved, and whether @SuiNetwork can become the coordination layer for the agentic economy. https://t.co/PtoZjdFAG9

[Lammy.Sol @lammy_sol]

Over the past couple of days, I’ve been exploring the Abelian App and trying out different parts of the experience.

The more I used it, the more I started noticing little things that I really enjoyed, a few areas where I think the experience could be smoother, and some ideas I’d love to see the team build next.

So here’s my feedback around:

What I enjoyed, what could be better, and what I’d like to see built next.

𝑻𝒉𝒆 𝒗𝒊𝒅𝒆𝒐 𝒈𝒊𝒗𝒆𝒔 𝒚𝒐𝒖 𝒂 𝒍𝒐𝒐𝒌 𝒂𝒕 𝒎𝒚 𝒂𝒄𝒕𝒖𝒂𝒍 𝒖𝒔𝒆𝒓 𝒆𝒙𝒑𝒆𝒓𝒊𝒆𝒏𝒄𝒆, 𝒃𝒖𝒕 𝒇𝒐𝒓 𝒕𝒉𝒆 𝒇𝒖𝒍𝒍 𝒅𝒆𝒕𝒂𝒊𝒍𝒔, 𝒄𝒉𝒆𝒄𝒌 𝒕𝒉𝒆 𝒕𝒉𝒓𝒆𝒂𝒅 𝒃𝒆𝒍𝒐𝒘. 🧵

@PQabelian #Abelian

[Vynce @Ayomide_vynce]

@Toosensitivegq @bigbrovar @Peter_dotcom “As a 500L medical student

You shouldn’t be the one saying this. “They have paid the price “

Fine what they have done is wrong. The caption is unprofessional. But expulsion ? Common. That’s is just ruining their careers over a Caption lol

[Vynce @Ayomide_vynce]

@lammy_sol @quantpilotapp Amazing write up about @quantpilotapp

[Lammy.Sol @lammy_sol]

𝐖𝐇𝐀𝐓 𝐈𝐅 𝐘𝐎𝐔𝐑 𝐓𝐑𝐀𝐃𝐈𝐍𝐆 𝐈𝐃𝐄𝐀 𝐂𝐎𝐔𝐋𝐃 𝐁𝐔𝐈𝐋𝐃 𝐈𝐓𝐒𝐄𝐋𝐅 ❓

You have a strategy in your head.

Maybe it’s something as simple as:

“Buy BTC when these conditions happen. Sell when these other conditions happen.”

But then comes the difficult part.

How do you turn that idea into something that can be tested?

How do you know if it actually works?

That’s where @quantpilotapp comes in.

A🧵

[Lammy.Sol @lammy_sol]

I PUT ARMORCODEX BETWEEN CODEX AND MY CODEBASE 🛡️

I wanted to see what happens when an AI coding agent can do its job but still has boundaries it cannot cross.

So I tested ArmorCodex × OpenAI Codex on Sentra, a small monorepo with a real authentication debugging task.

With @armoriqAI , I:

→ Registered Codex’s intent plan

→ Created a deny-curl policy directly from chat

→ Kept normal Bash commands working

→ Let Codex investigate and fix 2 auth bugs

→ Got all 5/5 tests passing

→ Watched ArmorCodex block a restricted curl call before execution

→ Verified the enforcement activity in the ArmorIQ dashboard

The interesting part?

ArmorCodex didn’t stop Codex from doing its job, it stopped the action that crossed the boundary I defined.

𝑪𝒉𝒆𝒄𝒌 𝑶𝒖𝒕 𝑻𝒉𝒊𝒔 𝑽𝒊𝒅𝒆𝒐 𝑻𝒐 𝑺𝒆𝒆 𝑯𝒐𝒘 𝑨𝒓𝒏𝒐𝒓𝒄𝒐𝒅𝒆𝒙 𝑾𝒐𝒓𝒌𝒔 𝑰𝒏 𝑨 𝑹𝒆𝒂𝒍 𝑪𝒐𝒅𝒆𝒙 𝑾𝒐𝒓𝒌𝒇𝒍𝒐𝒘 👇 ´

[Vynce @Ayomide_vynce]

@lammy_sol @Oluwaferanmihh @Somnia_Network @dreamDEXSomnia @1783Dao @broadchain_info 🔥✅

[Lammy.Sol @lammy_sol]

𝐅𝐎𝐋𝐋𝐎𝐖𝐄𝐑𝐒 𝐀𝐑𝐄 𝐄𝐀𝐒𝐘 𝐓𝐎 𝐁𝐔𝐘. 𝐀 𝐓𝐑𝐀𝐂𝐊 𝐑𝐄𝐂𝐎𝐑𝐃 𝐈𝐒𝐍’𝐓.

There are some dApps you build that make you proud of the developer you’re becoming.

Credence is one of them.

I built Credence on @Somnia_Network with @dreamDEXSomnia Event Contracts to solve a simple problem:

How do you know who is actually good at making predictions?

Social media followers can be bought. Likes can be bought. Hype can be created.

But a verifiable track record is different.

With Credence, users can make predictions
on live markets, build a permanent reputation from their actual results, and discover predictors based on performance.

The more you predict, the more your track record speaks for you.

And when you find a predictor you mmmm m mmmm mmmm

trust, you can back their active prediction.

No guessing based on popularity.

Just predictions, results, and reputation.

🌍 Available in English, Chinese, French & Portuguese.

Built for the @SIMANLABS SUPERSTAR Global Builder Speed Hackathon.

𝑾𝒂𝒕𝒄𝒉 𝑻𝒉𝒆 1 𝑴𝒊𝒏𝒖𝒕𝒆 𝑫𝒆𝒎𝒐 𝑻𝒐 𝑺𝒆𝒆 𝑯𝒐𝒘 𝑪𝒓𝒆𝒅𝒆𝒏𝒄𝒆 𝑾𝒐𝒓𝒌𝒔 👇

[Vynce @Ayomide_vynce]

@lammy_sol @404gen_ @opentensor Amazing game man

[Lammy.Sol @lammy_sol]

Built a racing game for the @404gen_ Game Jam and decided to test it with my guy who’s actually a proper car racing lover 😂🏎️

Meet ROAD//ZERO. 🏁

Sent him the race link on Telegram, he joined and we went straight at it.

I thought building the game was the hard part until I watched this guy leave me behind on my own game 😭

But seeing the multiplayer actually work, two different players, different devices, same race, in real time, was probably my favorite moment while building this.

ROAD//ZERO has Quick Race, Championship against AI drivers, and Race Together where you create a room, send the link to your friend, and race live.

No installation. Just open your browser and race.

Built and shipped for the @404gen_ Game Jam 🏁

Play here: https://t.co/7mLuglPJ9Y

𝑯𝒆𝒓𝒆’𝒔 𝑻𝒉𝒆 𝑫𝒆𝒎𝒐 𝑽𝒊𝒅𝒆𝒐 👇

[Lammy.Sol @lammy_sol]

“No matter the effort you put in life if it does not yield results it is nullified”

[Lammy.Sol @lammy_sol]

Won my first hackathon since I started building again on Sept 1 🏆

Built LATCH for the @Bazantic track.

First one on the board. More to come.

[Lammy.Sol @lammy_sol]

Then there’s this👇