Platform

LOCI Line-of-code intelligence

LOCI is a Line-of-Code Intelligence platform that revolutionizes system observability. At its core is the proprietary Large Code Language Model (LCLM), a vertical LLM specializing in comprehensive system workload analysis focusing on power, performance, and reliability. This advanced model middleware, trained on SoC real-time tracing data since 2017, leverages extensive experience in mission-critical safety projects and HPC in the automotive sector. Our tuned middleware LCLM boasts a vocabulary 1000 times smaller than LLM, with reinvented tokenizers and pipeline training using only 6 GPUs, making it exceptionally efficient. LOCI uncovers anomalies in static data, shifts observability left, and predicts emerging trends in dynamic data discovering hardware and software reliability issues.

Value Proposition

LOCI provide Ai powered analysis of binary files, bringing your static code to life.
Binary files typically require to shift quality left and conduct reliability check before initiating lengthy testing processes. LOCI helps to unveil unknown issues providing root cause analysis, even in the absence of suppliers’ source code, leading to reduced mean time to repair (MTTR) and employee’s development sprint.

Reduce mean time to repair (MTTR)
Increase employee efficiency and team development velocity up to XXXX%
Improve system reliability and quality
Reduce development cycle time (work on feature from start to finish)
Enhance code coverage and untested code path
Change failure rate (# of failed deployment/ # of deployments - check DENSO)

Key Features

System Coverage
Achieve full system coverage using symbol names, eliminating the need for suppliers’ source code. This enhances safety and understanding among SOC Cores islands running QM or ASILx software components while also addressing time and performance impacts.

Time and Performance Degradation Prediction
Predict time and performance degradation across different suppliers’ versions without needing the source code, allowing for root cause analysis. This approach increases system reliability and quality, ensuring functionality uptime and availability for users.

Test Strategy
Based on behavioral analysis, LOCI Indicate which symbols to focus on during unit and system tests. This increases the coverage of system test scenarios and enhances functional quality.

Power optimization (available Q2, 25)
AI-driven suggestions for reducing energy usage without compromising performance.
Track energy consumption across all Functions and tasks, indicating where to optimize energy