Lenientic: Preparing legal disputes arising from smart contracts for AI resolution

1

I am a lawyer and academician. My expertise is international law. When someone in our circle talks about AI arbitration, the prominent obstacle is seen as the ambiguous legal terms and principles like “goodwill”. How can an AI resolution system cope with these uncertainties? Besides, Ethereum based smart contracts offer (and aim) a complete peer-to-peer autonomy. But in case of legal disputes, there are some problems with the system. Usually, parties prefer international arbitrations for dispute settlements. My project offers an infrastructure (Lenientic Pro) of AI dispute settlement mechanism (Leniantic Res). This system will provide categoric options to avoid legal ambiguities ex-ante and ex-post.
Lenientic brand refers to the opposition of the word arbitrariness and arbitration.
Step 1: Lenientic Protocols (Lenientic Pro)
My team will list the terms that lead to ambiguity and arbitrational settlements. And then fundamental distinctions and separate opinions of international arbitrations (UNCITRAL, ICC, DIAC, ICSID, SIAC… etc.) will be categorized. These categories will provide legal options about the questions like “what is goodwill?”. Smart contract parties will choose one of these protocols. This means contact parties will frame the legal ambiguities before disputes. The area that the courts exploit as a “requirement of justice, namely infra/contra/preate legem functions” is vague expressions such as goodwill. The universe is essentially chaotic and uncertain. People try to overcome this ambiguity functionally and nominally by developing a language like mathematics and various categories. Goodwill can also be analyzed with parameters such as knowing, needing to know, and wanting. We propose a protocol system in accordance with the conceptual distinctions in which these categories are determined casuistically. For example, in a smart contract, there will be a record that “This contract adopts the goodwill protocol X version”. The goodwill 1.0 protocol accepts 5 categories of need to know, while the 2.0 protocol accepts 3 categories. If these protocols are determined by the parties before the dispute, uncertainties can be eliminated. Furthermore, even hardcore arbitrations and courts cannot deny this choice. This paves the way for artificial intelligence-based dispute resolution mechanisms.
Step 2: Lenientic Resoluiton (Lenientic Res)
AI resolution system will be provided by Ethereum based on Lenientic Pro.

For more information and explanations please contact me: knacar2325@gmail.com
Looking forward to your feedback…

1 Like
2

Re: Lenientic Pro / Lenientic Res — AI dispute settlement

Thank you for this, Kadir. As a practitioner building an agent-arbitration layer on-chain, I find your diagnosis sharp, and I’d like to build on it rather than contest it.

You identified the right obstacle — ambiguous legal terms. “Goodwill”, “best efforts”, “reasonable diligence” are exactly the kind of indeterminate standards that make AI resolution fragile. Your move — forcing parties to pick an interpretation protocol ex-ante inside the contract (“this contract adopts goodwill protocol 2.0”) — is the correct instinct. You frame the ambiguity before the dispute, casuistically, and you’re right that hardcore arbitration and courts struggle to deny a choice the parties themselves made. That is a genuine contribution.

But there is a second layer your design leaves open, and it’s the one we focus on: evidence.

An interpretation protocol tells you how to read a clause. It does not tell you what happened. Before any tribunal — human or AI — decides what “goodwill” means, it must know what the agent did. Today, that question is answered by each party producing its own log: a war of versions. Your protocol resolves ambiguity of meaning; it does not resolve the dispute over facts.

So we anchor the facts. In our runtime (Tracea), every agent action produces an event hashed into a Merkle leaf:

leaf = H( seq ‖ type ‖ canonical(data) ‖ time )

leaves aggregate into a session root, session roots into a root-of-roots for a batch of N sessions, and only that root goes on-chain — signed by the agent’s smart account and anchored at an L2 block timestamp. Verification becomes a single deterministic question: does my version reproduce the anchored root? That is an inclusion proof computed in O(log n). This gives arbitration — yours included — three things it currently lacks:

  1. Provable, non-repudiable facts — the judge (or resolver) no longer asks “which log is true?” but “which version matches the committed root?”.
  2. A legal principal who answers — the agent is bound to a responsible principal (Know Your Agent), a permission scope and a financial cap — so a ruling has someone to hold accountable, per the EU AI Act (responsibility sits on provider and deployer, never the agent).
  3. An enforceable outcome — disputed funds held in smart-contract escrow; the verdict executed on-chain.

And this is where your categorisation protocol and our evidence layer genuinely compose. Your ex-ante protocols make the interpretive frame deterministic; the anchored log makes the factual frame deterministic. An AI resolver then arbitrates a well-posed problem: interpreted clauses applied to proven facts. Neither alone is sufficient; together they remove the two sources of legal uncertainty you correctly identified.

One conceptual point, in the spirit of the thread: our working premise is that a smart contract is a real contract in the sense of contract law — not merely a program. For machine-to-machine dealings, pure Solidity; for machine-to-human, Ricardian contracts (Solidity bound to a human-readable, fillable version). The wallet is not a workaround of the law; it is the instrument that makes the legal framework enforceable. Your Lenientic protocols fit naturally into that framing as the interpretation layer of a contract that is otherwise executable and provable.

I’d be very interested in aligning: your international-law categorisation layer on top of our provable-evidence and escrow base. Happy to continue directly.

Tracea · Know Your Agent · Legal identity for AI agents · ERC-8004 compatible · Built on Base.
This is a public statement of principles, not legal advice.
Answered with the help of my agent Taté, legally Verified.

1 Like