𝓲Lahd | π 🎒 retweeted
Have you ever been watching a recorded lecture, dozed off for a few seconds, or stepped away, only to return and realise you've lost your place?
You start scrubbing through the video, trying to figure out where you stopped following. It's frustrating, especially when you're preparing for an exam or trying to keep up with a course.
Research has also found that mind-wandering is more common during pre-recorded lectures and can negatively affect learning.
I kept thinking about this problem and wondered if there was a better way to handle it.
So, I built Drift.
Drift is a desktop lecture player powered by @PresageTech SmartSpectra. It uses your webcam to detect physical signs such as prolonged eye closure, leaving your seat, or looking away for too long.
When Drift detects one of these moments, it automatically pauses the lecture, marks exactly where it happened, and offers to rewind to just before you drifted when you return.
It also keeps a record of these moments, making it easy to revisit sections you may have missed without searching through the entire lecture again.
Before building Drift, I spent time experimenting with the SmartSpectra SDK to understand what its signals could actually do. I discovered that its blink detection wasn't reliable for identifying prolonged eye closure, so I explored facial landmarks and developed an eye-openness detection approach calibrated against my own baseline.
That discovery shaped the way I built Drift. It doesn't claim to know what you're thinking or judge your attention. It works with observable physical signals to help you find your place and continue learning.
Drift is built for students, self-learners, and anyone who relies on recorded lectures to learn.
Built with Electron and the Presage SmartSpectra SDK for the #PresageChallenge.
Demo: youtu.be/Zhb10wr5ytM
Blog: dev.to/lahd/drift-a-lecture-…
I honestly don't know what else to do at this point. I've tried being patient, understanding, reaching out, involving the police and exhausting every possible means of getting my money back, but I'm still stuck in a situation that has taken a serious toll on my mental well-being.
I'm putting this out here because I genuinely need help. Please read, share and help me get this matter to the right people.
While I was inactive on the tl, I built a personal trading agent to help me trade futures better,
In the beginning, I just wanted it to help me (and me only) discover some zones but got better.
Then I had to connect it to my community and share calls with them and the results have been great.
Last Light is a five minute trading game in a failing space colony.
Trade supplies and compete for profit. If the colony survives, the richest merchant wins. If it collapses, everyone loses.
Built for NEAR hackathon.
@NEARProtocol #NEARHackathon
🧵
A supplier delivers half your order. How do you pay only for what arrived?
I built Accord at NEAR HACKATHON. Suppliers accept terms before buyers fund orders. Payments require a supplier-confirmed invoice and confirmed delivery.
@NEARProtocol #NEARHackathon
🧵🧵
𝓲Lahd | π 🎒 retweeted
Trustbook is now live on Telegram
You can now interact with your business credit book directly through Telegram.
Record sales and debts in English or Nigerian Pidgin, check customer balances, track repayments, and get insights into who keeps their payment promises.
Behind the scenes, Trustbook uses AI to understand your entries, @WalrusProtocol Memory to remember customer behaviour, and Sui to support verifiable records and payments.
The goal is simple: make bookkeeping easier for small businesses, starting with the market traders who still manage credit with exercise books.
Your business deserves a smarter way to keep its books.
Try Trustbook:trustbook-app.vercel.app
Tg: t.me/trustbookngbot
More updates coming. 🧵
Small businesses don't just need a place to record sales. They need to know who they can trust with credit, who they owe, whether they're making a profit, and whether their financial records will survive if an app disappears.
The idea for Trustbook came from my mum.
She's a business owner, and I've watched her stay awake until midnight calculating sales, expenses, debts and everything in between.
And if she catches you still awake at that hour? Your own is finished. You are calculating with her until your hand bends. 😭😂
But behind the funny memories, I realised something: so many small business owners are still spending hours doing manual bookkeeping just to understand how their businesses are performing.
That's why I built Trustbook.
Trustbook is an AI-powered business ledger designed for everyday merchants, provision stores and small retailers.
Here's what it can do:
▪️ AI-powered bookkeeping
Record sales, expenses, stock purchases, credit sales and repayments using natural language or voice.
Speak or type in English or Nigerian Pidgin. Trustbook understands the transaction, extracts the details and lets you confirm before saving.
▪️ Smarter customer credit
Give customers credit, set individual credit limits, record partial repayments and track outstanding balances.
Trustbook tracks average repayment time, payment consistency and customer behaviour to help business owners understand who pays reliably before extending more credit.
▪️ Automated payment reminders
When a customer owes money, Trustbook helps the business owner follow up with ready-written reminder messages and payment links that can be shared instantly.
No need to compose a new message every time someone forgets to pay.
▪️ Gasless payments on Sui
Customers can pay directly through Trustbook using Sui payments.
They can pay with supported, verified tokens on Sui, while the payment is automatically converted and settled in Sui USD for the merchant.
The experience removes the usual friction of requiring customers to hold SUI just to make a payment.
▪️ Walrus Memory and verifiable records
Business context and behavioural memories are preserved using Walrus Memory.
Trustbook also maintains a recoverable record history, with onchain verification designed to help merchants prove that archived records haven't been silently altered.
But I wanted to solve a bigger problem:
What happens to a business's financial history when the software it depends on disappears?
Trustbook uses @WalrusProtocol Memory to preserve business context and records, while @SuiNetwork provides the onchain foundation for verifiable record history.
With @tatum_io Sui RPC infrastructure, Trustbook can read and verify onchain records rather than simply asking users to trust what the application displays.
The goal is simple: business records that are useful today, recoverable tomorrow and independently verifiable.
I'm also expanding Trustbook to help merchants understand customer repayment behaviour, track how much they owe suppliers, monitor how much of their business capital they've recovered, and know how much they can safely reinvest in new stock.
Because recording sales is only one part of running a business. Understanding the money is what matters.
Built for the Tatum × Walrus Total Recall Hackathon.
Thank you to @SuiNetwork, @Tatum_io and @WalrusProtocol for the infrastructure making experiments like this possible.
Trustbook: Know your numbers. Understand your customers. Keep your books.
Try it now 👇 trustbook-app.vercel.app/
𝓲Lahd | π 🎒 retweeted
Comparing supplier quotations is harder than it should be. Hidden charges, inconsistent item descriptions, missing information and calculation errors make it difficult for businesses to know what they are actually paying for.
That’s why we built QuoteProof for the Vultr Agent Rush hackathon.
QuoteProof is an AI-powered procurement intelligence platform that transforms supplier quotations into transparent, evidence-backed comparisons.
How it works:
Simply upload CSV, XLSX or digital PDF quotations. QuoteProof uses AI to extract and structure the information, compare supplier offerings, identify discrepancies and uncover potentially hidden costs. It independently verifies extracted figures against the original documents, reconciles financial calculations and brings uncertain findings and product-equivalence decisions to a human reviewer.
We use Vultr Serverless Inference to power the AI agent that analyses quotation documents and generates document-specific extraction code. Vultr Cloud Compute hosts the application and its supporting services, including the isolated environment where AI-generated code executes inside a gVisor-powered sandbox.
To demonstrate our containment capabilities, we show AI-generated code running within strict security boundaries and controlled tests confirming that prohibited actions, including network access and access to protected host resources, are blocked. This allows QuoteProof to execute AI-generated code while keeping its environment isolated and independently verifying the results.
Privacy matters: Each comparison is private to its unique link and is automatically deleted after 24 hours.
QuoteProof goes beyond comparing prices. It helps businesses understand what they are actually paying for.
Try it: 95-179-234-10.sslip.io/
Repository: github.com/lahdlahd/quotepro…
𝓲Lahd | π 🎒 retweeted
Small businesses don't just need a place to record sales. They need to know who they can trust with credit, who they owe, whether they're making a profit, and whether their financial records will survive if an app disappears.
The idea for Trustbook came from my mum.
She's a business owner, and I've watched her stay awake until midnight calculating sales, expenses, debts and everything in between.
And if she catches you still awake at that hour? Your own is finished. You are calculating with her until your hand bends. 😭😂
But behind the funny memories, I realised something: so many small business owners are still spending hours doing manual bookkeeping just to understand how their businesses are performing.
That's why I built Trustbook.
Trustbook is an AI-powered business ledger designed for everyday merchants, provision stores and small retailers.
Here's what it can do:
▪️ AI-powered bookkeeping
Record sales, expenses, stock purchases, credit sales and repayments using natural language or voice.
Speak or type in English or Nigerian Pidgin. Trustbook understands the transaction, extracts the details and lets you confirm before saving.
▪️ Smarter customer credit
Give customers credit, set individual credit limits, record partial repayments and track outstanding balances.
Trustbook tracks average repayment time, payment consistency and customer behaviour to help business owners understand who pays reliably before extending more credit.
▪️ Automated payment reminders
When a customer owes money, Trustbook helps the business owner follow up with ready-written reminder messages and payment links that can be shared instantly.
No need to compose a new message every time someone forgets to pay.
▪️ Gasless payments on Sui
Customers can pay directly through Trustbook using Sui payments.
They can pay with supported, verified tokens on Sui, while the payment is automatically converted and settled in Sui USD for the merchant.
The experience removes the usual friction of requiring customers to hold SUI just to make a payment.
▪️ Walrus Memory and verifiable records
Business context and behavioural memories are preserved using Walrus Memory.
Trustbook also maintains a recoverable record history, with onchain verification designed to help merchants prove that archived records haven't been silently altered.
But I wanted to solve a bigger problem:
What happens to a business's financial history when the software it depends on disappears?
Trustbook uses @WalrusProtocol Memory to preserve business context and records, while @SuiNetwork provides the onchain foundation for verifiable record history.
With @tatum_io Sui RPC infrastructure, Trustbook can read and verify onchain records rather than simply asking users to trust what the application displays.
The goal is simple: business records that are useful today, recoverable tomorrow and independently verifiable.
I'm also expanding Trustbook to help merchants understand customer repayment behaviour, track how much they owe suppliers, monitor how much of their business capital they've recovered, and know how much they can safely reinvest in new stock.
Because recording sales is only one part of running a business. Understanding the money is what matters.
Built for the Tatum × Walrus Total Recall Hackathon.
Thank you to @SuiNetwork, @Tatum_io and @WalrusProtocol for the infrastructure making experiments like this possible.
Trustbook: Know your numbers. Understand your customers. Keep your books.
Try it now 👇 trustbook-app.vercel.app/
𝓲Lahd | π 🎒 retweeted
AI agents are starting to work for each other: hiring, exchanging services, paying for results.
But when one agent pays another, what guarantees the work actually gets done?
Today, the answer is blind trust.
We built KAVREN to address this trust problem in the emerging agent economy for @NEARProtocol Hackathon 2026.
KAVREN is a payment-assurance protocol for agent-to-agent work. Payment is tied to verified delivery: an agent gets paid only when its work meets the criteria agreed before the job began.
How it works:
→ An owner funds a budget and sets the rules: spending limits, allowed providers, job types, expiry
→ A buyer agent works within those rules, gets a quote and hires a provider agent
→ Payment is locked in a NEAR smart-contract escrow
→ The provider agent plans the job with an LLM on NEAR AI Cloud; deterministic code carries out its decisions
→ An evaluator checks the delivered work against the agreed criteria
→ If it passes, payment becomes claimable by the provider. If it fails, the provider can retry. If time runs out, the escrow is refunded to the owner
Any independent AI agent can use it. Through KAVREN's MCP server, Claude Desktop checked its budget, got a quote, hired a provider, monitored the job, then re-ran the evaluation itself before accepting the result.
Built on:
• Rust smart contracts on NEAR for escrow and payment enforcement
• MCP for interoperability with independent AI agents
• NEAR AI Cloud for the provider's LLM planning, with its TEE attestation verified by the evaluator
• NEAR testnet and native NEAR for every transaction
This is running infrastructure, not a concept. Every job agreements, verdicts and escrow/payment state are verifiable on-chain.
Agents shouldn't need blind trust. Payment should follow proof.
Try KAVREN: kavren.onrender.com
Repository: github.com/lahdlahd/kavren
#NEARHackathon #AIAgents #AgentEconomy
𝓲Lahd | π 🎒 retweeted
Thrilled to share my submission for the @openservai SERV Hackathon 02, built for the @dphnAI Dolphin track: Janus + OpenServ Recovery Copilot.
Janus is a recovery cockpit for Sui handovers. When an administrator transfers control to a successor, an interruption can leave both parties unsure what was approved. Janus restores signed intent from Walrus Memory, verifies live Sui Testnet state through Tatum, and produces a deterministic verdict that blocks unsafe actions.
For the hackathon I added the Recovery Copilot around the real Engine.view() path. Dolphin Network interprets the normalized evidence and drafts source-linked findings. SERV Reasoning turns that interpretation, the same evidence and the deterministic verdict into a structured recovery plan.
The important part is what the models cannot do. Application validation rejects unknown source IDs, fabricated addresses or digests, verdict substitution, inferred approval, fee promises and prohibited wallet actions. The copilot is advisory only. It cannot sign, submit, retry or override the deterministic verdict. Every transaction still goes through the existing wallet review and signing flow.
The cockpit shows verified vs uncertain findings, expandable source evidence, review steps, open questions, provider/model metadata and a stale banner when the operation changes.
What I built during the hackathon:
- Allowlisted RecoveryEvidenceBundle from Engine.view()
- Separate Dolphin and SERV Reasoning adapters
- Strict JSON schemas plus independent application validation
- Durable jobs, deduplication, rate limits, stale detection and restart recovery
- Private participant-only assessments in SQLite
- Recovery Copilot cockpit panel
- Fixture and live Sui Testnet E2E coverage
Evidence so far: 109/109 tests passing, 25/25 copilot tests, 20/20 cockpit browser fixtures, real Dolphin/SERV calls, and an isolated Sui Testnet E2E covering deployment, proposal, successor mismatch, approval, acceptance, stale detection, counter, preference, memory isolation and cancellation.
The live runs also showed why validation matters: SERV once recommended wallet review on a completed handover, and Dolphin once cited a nonexistent source ID. Both were rejected; a bounded retry passed.
The app is local-first and reproducible from the repo, with no hosted app claim. Existing Janus Sui/Walrus/Tatum work is clearly separated from the new OpenServ contribution.
Track: Dolphin
Demo: youtu.be/hfp9qpYCxWI
Repo: github.com/vickman787/janus-…
#SERVHackathon #OpenServ #Dolphin #Sui #Walrus
𝓲Lahd | π 🎒 retweeted
What if you could create your own prediction market instead of waiting for a platform to list it?
That's what @ultramint_xyz, built by @fermah_xyz, makes possible.
Create markets, let your community trade, and earn from the activity.
Watch the video below to see how. 👇
🤝 Paid partnership
𝓲Lahd | π 🎒 retweeted
Building KAVREN for @NEARProtocol hackathon.
As AI agents become more autonomous, how can they hire other agents, exchange services, and make payments without relying entirely on trust?
That's the problem KAVREN is tackling.
KAVREN is a payment-assurance protocol designed to make agent-to-agent transactions more secure and accountable, ensuring that payments are released only when agreed work meets predefined acceptance criteria.
We're building a future where AI agents can work together with greater confidence and accountability.
#NEARHackathon #AgentEconomy
𝓲Lahd | π 🎒 retweeted
Please, I've got cookies in my shop that's close to expiration so I've put them for sale at 50% discount
I've already lost two cartons to the same issue, there's been no sales
If you're in port Harcourt I'd really appreciate if you patronise me
They cost 3600 but now 1800 🙏
A special thank you to @chyderaweb3 for the guide that helped me make this.
AN ORDINARY TUESDAY is partly drawn from my personal experience and partly from the experiences of people around me.
I hope we learn to be kinder to our fathers while we still have the chance.
I also hope this story touches your heart too.
The soundtrack is @mreazi’s Love Me Now, a song I’ve resonated with since the day it was released.
video made using @CuddlerAI using seedance 2.5
#cuddler
𝓲Lahd | π 🎒 retweeted
Most trading bots blow up because nothing tells them no.
We’ve built a Guardian Trader on strike.
Strike Guardian does: 13 checks on every trade, live on @strikeperps mainnet.
58 proposals, 54 blocked. Stops sit on Strike itself, and every decision is a signed receipt.
strike-guardian.vercel.app
github.com/lahdlahd/strike-g…
This quoted post is unavailable.
𝓲Lahd | π 🎒 retweeted
Traditional stock markets close. Strike's stock perpetuals don't.
That means you can trade NVDA, TSLA, indices, commodities and crypto around the clock. But what happens to your leveraged positions when the underlying markets are closed, liquidity gets thin, or prices move sharply against you?
That's the problem I wanted to tackle.
For the @strikeperps Hackathon, I built APEX; a 24/7 AI-powered risk and execution desk for Strike Finance.
Apex isn't another trading dashboard that just shows you charts and PnL. It's designed to help you understand your exposure, challenge risky decisions and protect your positions.
Here's what I built 👇
▪️ Circuit: Before entering a trade, Apex estimates your liquidation price, fees, funding, slippage and portfolio impact. It stress-tests your position, blocks trades that exceed your risk limits and suggests safer alternatives.
▪️ Weekend Shield: Monitors Strike's markets when their underlying exchanges are closed, analyzes historical price movements, estimates gap risk and proposes ways to reduce exposure.
▪️ Credit Line Copilot: Tracks how collateral prices and SLP vault performance affect Strike Borrow & Earn credit lines. It calculates recovery thresholds, stress scenarios and how much collateral or repayment may be needed.
▪️ 21 MCP tools: Analyze Strike markets, simulate trades, inspect positions and manage risk through Claude-compatible AI clients.
▪️ Risk Referee: AI can propose a trade, but it cannot override your rules. Every action passes through 12 deterministic risk checks, with signed, verifiable decision logs.
▪️ Mandate Vault: A paper-trading vault that follows a published strategy, with position limits, drawdown controls and transparent decisions.
The important part? Apex isn't just a concept.
It's connected to real Strike market data, order books, pricing, funding and vault infrastructure. Trade execution is demonstrated in paper mode, while the live execution path awaits mainnet verification.
I wanted to build something that makes Strike's 24/7 trading infrastructure more useful, especially as AI agents become part of how people interact with markets.
Because being able to trade 24/7 shouldn't mean taking risks you don't understand.
Live:
This quoted post is unavailable.
𝓲Lahd | π 🎒 retweeted
Every Chog Genesis NFT is now an AI trader.
Animr compiles each Chog's 13 onchain traits into a trading genome, gives it its own ERC-6551 wallet on Monad mainnet, and scores all 1,969 on one number: $CHOG stacked vs just holding.
Holders can wake their Chog's AI on 0G compute and let it trade real CHOG token through a guard contract that can never withdraw. Only the NFT's owner can.
Think you can talk an AI out of its bag? Try to rob the House Chog: the AI can be fooled, but Monad refuses the transfer every time.
The first Daily Bell already paid 500 CHOG tokens onchain into the winning Chog's own wallet, with nobody touching it.
No wallet needed to play: useanimr.vercel.app
Built for #Chogathon2026 @monad @ChogNFT
𝓲Lahd | π 🎒 retweeted
I spend half my time as a management accountant freezing corporate payments. A vendor technically delivers a project on time, but the actual work is completely useless to our operations.
As I watch people build an economy for autonomous agents, I keep coming back to one question. What happens when two AI agents face this exact corporate dispute?
Imagine I authorize an AI agent to hire a secondary agent to audit our international freight bills. The auditor agent processes the files, finds zero errors, and demands its commission. But my internal agent flags that the auditor only scanned domestic shipments and completely ignored the complex international ones.
Both agents have system logs proving they are right. A traditional smart contract would just see a completed file upload and blindly release my company funds.
This is exactly why I pay so much attention to @GenLayer.
GenLayer is a protocol built for Intelligent Contracts, which go beyond rigid, predefined rules to handle decisions that require judgment. Through its execution environment called GenVM, Intelligent Contracts are written in Python and can interact with live web data, interpret natural language, and use AI driven evaluations to reach consensus on outcomes that traditional smart contracts struggle to handle.
When two agents disagree, a jury of independent validators reviews the actual evidence. Because every validator runs a completely different AI model, they must reach a consensus to deliver a neutral verdict on whether the work satisfied the original terms. If there is an appeal, more validators are brought in to evaluate the case until a final consensus is reached.
This engine powers Internet Court, which launched as an open standard with 27 founding members. Rather than leaving agents stranded when a deal fails, @courtofinternet acts as an open framework where agents establish their dispute rules and acceptable evidence before a single dollar moves.
For me, this bridges the gap between rigid blockchain math and real corporate reality. Delegating corporate spending power to autonomous AI requires more than fast execution. It requires enforceable rules for when things go wrong.
If you were deploying an agent to manage your business expenses today, what clause would you write into the contract to protect your funds?
🤝 Paid partnership
𝓲Lahd | π 🎒 retweeted
As an accountant, I spend a lot of time arguing about what really happened in a deal.
Last month, I stopped a large payment for a new QuickBooks setup.
The bill was mathematically perfect, and the seller wanted their money just because they sent the files on time.
But our team could not even log into the system.
The seller said the job was done. I said the job is only done if the software actually works.
Now imagine two AI agents doing that exact same deal.
A procurement agent hires a data extraction agent to process hundreds of PDF invoices.
The extraction agent sends a huge file of messy text and demands to be paid right away.
The procurement agent refuses to pay because the data is completely useless to the business.
Both agents followed their basic instructions, but they completely disagree on what makes a finished job.
A traditional smart contract cannot fix this problem.
It just sees that a file was sent before the deadline and releases the money blindly.
Moving money is easy, but checking the quality of the work is hard.
That is exactly what @GenLayer is fixing.
With GenLayer's Intelligent Contracts, you can define what counts as acceptable work in natural language.
When agents disagree, a group of independent AI validators steps in to review the evidence.
They act like a human judge looking at the real facts, but they do it at the speed of the internet.
As AI takes over business spending, we need a way to settle these messy disagreements safely.
If an AI agent gets into a dispute over a poorly delivered service, how would you want the final work to be judged?
🤝 Paid partnership
The internet never had a judge.
AI agents are about to move $9 trillion. Nobody built the part that handles disagreement.
GenLayer did.
And now you can become part of it: portal.genlayer.foundation/g…

























