Skip to main content
The core principle of the Logic-Injection-on-Origin Protocol (LIOP) is the Logic-Injection-on-Origin (LIO) paradigm. Under the Postulate of Origin, data remains at rest within its physical custody boundary unless the output is mathematically aggregated or cryptographically attested. As AI systems scale, context window saturation and egress bandwidth become primary architectural constraints. Traditional protocols (such as Model Context Protocol) rely on Context-Pulling, extracting remote data to a centralized model.
Animated Logic-on-Origin vs Context-Pulling ComparisonAnimated Logic-on-Origin vs Context-Pulling Comparison

​
The Data Gravity Trap (Context-Pulling)

In standard AI applications, when an LLM needs to analyze data, read a ledger, or search a repository, the server must extract the raw data and transmit it over the network to the model. This introduces three structural bottlenecks:
  1. Exfiltration Risk (Data Privacy): Transmitting raw databases conflicts with compliance standards such as HIPAA, GDPR, or financial secrecy regulations. The remote LLM provider temporarily holds sensitive user records in memory.
  2. Network Overhead: Transferring gigabytes of telemetry or log streams over HTTP causes high transfer latency and network egress costs.
  3. Context Window Saturation: Streaming large datasets directly into model context windows incurs significant token costs and risks truncation.
Data possesses gravity: as dataset size increases, network transfer becomes slower, costlier, and less secure.

​
The LIOP Solution: Logic-on-Origin

Instead of moving large datasets to the model, LIOP dispatches compiled logic to the dataset. When an LIOP Agent requires an analytical result, it writes the query logic, compiles it into an isolated WebAssembly (.wasm) module or sandboxed script, and transmits it across the peer-to-peer network. Upon arriving at the target Data Node:
  1. Sandboxed Isolation: The logic executes within a WASI Sandbox configured with restricted file descriptors and capability allowlists.
  2. Local Processing: The WebAssembly code processes the dataset in situ at native execution speed, performing filtering, transformation, and aggregation locally.
  3. Verified Output: The module completes execution and returns an aggregated result along with an HMAC-SHA256 ZK-Receipt that cryptographically binds the output to the executed logic.
Raw records never leave the origin infrastructure.

​
Architectural Scope & M2M Integration

By moving computation to origin, developers deploy deterministic analytical modules directly across secure enterprise boundaries:
  • Healthcare Consortiums: Cross-institutional research cohorts execute epidemiological queries across hospital databases without exposing patient records protected by HIPAA.
  • Financial Audit & Fraud Detection: Anomaly detection heuristics execute directly inside sovereign banking enclaves without transmitting ledgers across public networks.