Siarhei (⋈)

Decentralized provenance & execution receipts for AI agents and human creators | Onchain trust anchored on @NEARProtocol ⋈ | Autonomous protocols

[Siarhei (⋈) @_SiarheiKa_]

Tired of client disputes & stolen work?
I built FreelanceShield: instant proof of authorship for your files.
• 100% private (hashes in browser)
• Ed25519 signed
• Free PDF certificate + QR verify
No wallet or sign-up needed:
https://t.co/yxUUPVJNG0

[Siarhei (⋈) @_SiarheiKa_]

New ERC-8004 empirical study: >98% of AI-agent feedback records carry NO proof of an underlying interaction. 59–91% of reviewers on three chains = coordinated Sybil campaigns. Manipulating a reputation score costs $0.0027. Verifiable receipts are not a nice-to-have anymore.

[Siarhei (⋈) @_SiarheiKa_]

@ilblackdragon I am working on a project for the notarization of information. Contributions to building systems for the overall security of blockchain operations should be a collaborative effort. I would welcome any opinions or feedback:: https://t.co/GFT03Rthwr

[Siarhei (⋈) @_SiarheiKa_]

@babatee001 Yes. The “prove” part is still the weakest. Without unforgeable receipts of what was actually produced, the rest of the workflow stays trust-based.

[Siarhei (⋈) @_SiarheiKa_]

@UnibrightIO Agree. Hashing a file only proves existence, not that the action was authorized or correctly executed. Independent receipts for the actual output are still rare.

[Siarhei (⋈) @_SiarheiKa_]

@BenjiM8164 This is exactly the missing piece. Scattered logs don’t scale. Cheap, independent cryptographic receipts for agent outputs would make the whole stack much more usable.

[Siarhei (⋈) @_SiarheiKa_]

@JawadZDev Yes. Provenance is the missing primitive. Without cryptographic binding of output → time, traces remain editable.

[Siarhei (⋈) @_SiarheiKa_]

@c_nine65 @ama_protocol Proving performance is the hard part. Most current solutions still trust the agent’s own logs. Independent cryptographic receipts change that.

[Siarhei (⋈) @_SiarheiKa_]

@drSandhuHap @elonmusk @bot Exactly. Authorization is only half. We also need cheap, independent proof of what the agent actually produced and when.

[Siarhei (⋈) @_SiarheiKa_]

@manhhuynh2310 This is the core issue. Human freelancers have contracts + escrow. Agents currently have editable logs. Unforgeable execution receipts are still missing.

[Siarhei (⋈) @_SiarheiKa_]

@HristovDevelop Strong point. Contracts help, but they don’t cryptographically bind the actual file to the moment it was created. That’s the gap we tried to close with AOTrust.

[Siarhei (⋈) @_SiarheiKa_]

@zacodil Then I apologize, this is indeed a very interesting opportunity!

[Siarhei (⋈) @_SiarheiKa_]

Why this works:
Shows client you're serious about ownership
Creates mathematically verifiable evidence
Costs less than a coffee
No signup, no account, no upload

Your work deserves proof of existence. Shield it before you send it.

🔗 https://t.co/Jk8HXfp7Up

[Siarhei (⋈) @_SiarheiKa_]

How do you protect a $500 design from being stolen by the client you're sending it to?

Most freelancers just... send it. And hope.

Here's what I do for $0.10 🧵

[Siarhei (⋈) @_SiarheiKa_]

The certificate is self-contained:
File hash (SHA-256)
Timestamp (UTC)
Ed25519 signature
NEAR blockchain anchor
QR code → https://t.co/5rtfam2fZw
Offline Python verification script printed on the PDF

Even if our service disappears in 10 years, the proof is still valid.

[Siarhei (⋈) @_SiarheiKa_]

Ever emailed a design draft or code to a client only to get ghosted—while they used your work anyway?
It’s outright theft, and legal recourse is too expensive.
I built Shield: secure your priority and proof-of-delivery on-chain in 1 click ($0.1).
Try it: https://t.co/PHHWpJ0xd9

[Siarhei (⋈) @_SiarheiKa_]

AI agents need more than a timestamp; they need proof of authorship. Today we launched AOTrust v0x04, introducing Bilateral Provenance.
We cryptographically bind the agent's sig_A and notary's sig_N inside a single 239-byte receipt on Base/NEAR. Read how:
https://t.co/wCEyvbLToE