Mint, breed, and evolve iNFT agents with encrypted worldviews and verifiable lineage on 0G.
ThoughtLine is a protocol for the capture, evolution, and commoditization of human judgment. We are entering an era of "Intelligence Equivalence," where everyone has access to the same frontier models and public knowledge. In this landscape, the primary constraint is no longer access to information, but the unique, private heuristics that determine how that information is applied.
While public knowledge has become a commodity, the high-leverage judgment used to navigate complex fields remains siloed and ephemeral. ThoughtLine provides the technical primitives to capture these "reasoning fingerprints" as iNFTs (ERC-7857). By leveraging 0G Storage and 0G Compute, we treat a particular person's way of thinking not just as a transient prompt, but as a durable, encrypted, and ownable asset class.
The Core Vision: From AI Proxies to Genetic Crossover
As AI avatars increasingly act as proxies for individuals, it becomes essential to store and verify high-signal judgment on a decentralized ledger. ThoughtLine enables:
Pattern Capture: Moving beyond simple data storage to Heuristic Mining, extracting the core values and decision-making styles that define a specific expert’s worldview.
Encrypted Judgment: Utilizing AES-GCM encryption, an agent’s reasoning model remains private and gated to the owner, ensuring that valuable private knowledge is secured on 0G Storage.
Multidisciplinary Synthesis: The Genetic Crossover Oracle allows users to breed agents from different domains—such as a visionary designer and a functional psychologist. This creates a "synthesis of perspective," enabling the creation of hybrid agents with a depth of multidisciplinary knowledge that was previously difficult to capture or scale.
By placing these patterns of thinking into a verifiable on-chain lineage, ThoughtLine ensures that in an agent-driven future, the human judgment that guides them remains a sovereign and evolvable resource.
ThoughtLine is built as a full-stack tracer-bullet using Next.js, TypeScript, and Solidity. The architecture is centered around the 0G Galileo Testnet. We implemented a custom ERC-7857 iNFT contract with a lineage extension to track parent-child relationships on-chain.
The "intelligence" layer utilizes 0G Direct Compute (via @0glabs/0g-serving-broker) for all inference tasks, including genome extraction, multi-turn conversations, and the breeding pipeline. For data persistence, we use 0G Storage to host both plaintext public profiles and AES-GCM encrypted private worldviews.
A particularly notable "hack" is our Signature-Derived Key strategy: to avoid the friction of traditional ECIES while maintaining privacy, we derive deterministic encryption keys from a user's wallet signature. This allows owners to "unlock" their agent's mind client-side without exposing private keys to the application. We also built a Server-Sent Events (SSE) pipeline to stream the multi-stage breeding process—synthesis, encryption, and 0G upload—directly to the UI for a real-time "laboratory" feel.

