The technology
Why LOCI thinks ahead.
LOCI predicts software behavior from real workloads and platform traces — not source code alone.
Inputs
The moat
8 years
of labelled execution data
Real hardware traces, labelled against the binaries that produced them. That is what the model learned from, and why it can predict a change before it runs.
Execution model · LCLM
Powered by AI Physics
Outputs
Engineering judgment
Why source code isn’t enough
Source code explains intent. Execution reveals reality.
Source code cannot reveal runtime behavior by itself.
What the engineer intended
Source code
What actually happens on hardware
Unknown runtime behavior
- Cache misses
- Memory pressure
- Power spikes
True vs predicted
Held-out · real silicon
Up to R² = 0.96
Fit on unseen binaries
MAPE ≈ 8%
Held-out
Predicted vs measured on new, unseen binaries · matched to the real eval · up to R² = 0.96 · MAPE ≈ 8%.
The engine behind LOCI
Execution intelligence, by the numbers.
LCLM — a small code-language model with a ModernBERT backbone and FlashAttention kernels, domain-trained on real execution traces. Powered by AI Physics.
6 yrs
Of real-workload traces
Billions
Of execution traces
up to R² 0.96
On new binaries and code it has never seen
Real silicon
Verified on hardware
120+
Granted patents · US · EU · JP
ISO 27001
Certified · SOC 2 in progress
The signals
What LOCI predicts.
Timing
Per-function response time before it runs.
Latency
p95 / p99 bounds, predicted from the binary.
Power
Per-function energy cost, ahead of time.
Memory
Worst-case stack depth and pressure.
CPU
Throughput and scheduling behavior.
GPU
Occupancy, register spills, kernel cost.
Cache
Modeled miss-rate on the hot blocks.
System effects
Wake cadence and power-state cost.
Where LOCI fits
The execution-intelligence layer in your existing loop.
Engineer
Asks for a change
Claude
Writes the code
LOCI
Predicts execution
CI
Guards every PR
Merge
Within contract
Production
No surprises
LOCI augments the workflow you already have — it doesn’t replace it.
- No runtime
- No instrumentation
- No simulator
- No profiler
- Works offline
Know how your software behaves before it runs.
LOCI is the execution-intelligence layer for AI coding agents.