Piloting with OEM Partners

AI that understands
Automotive Controls

Search, visualize, and triage your entire embedded systems stack—from Simulink models to C++ firmware. Self-hosted. Secure. Git-Native.

1M+ BLOCKS
VERIFIED ON PRODUCTION MODELS
<8B
WORKS WITH LOW COMPUTE LLM
100%
AIR-GAPPED SECURITY
< 5 min
TIME TO ROOT CAUSE

Self-Hosted Intelligence

Bring your own weights. Deploy local LLMs or connect to private enterprise endpoints. Data never leaves your VPC.

Git-Native Workflow

We don't just read code; we write documentation back to your repo. Works seamlessly with GitHub, GitLab, and Bitbucket.

Embedded Systems Parser

Native support for .slx, .c. We parse the logic gates, not just the text pixels.

UnboxLogic
How is SOC estimated?
AI
Just now
SOC is estimated using an Extended Kalman Filter (EKF).

The logic is located in the SOC Subsystem
Ask a follow up...
BMS / Battery Management / SOX / SOC
I_meas
V_meas
Temp_C
Subsystem
EKF_Core State Estimation
(Discrete)
SOC_%

Interactive
Logic Mapping

Don't just search for text strings. Search for logic.

When you ask UnboxLogic a question like "How is SOC calculated?", it doesn't just give you a text answer. It opens the exact subsystem in a lightweight, browser-based viewer.

One-Click Navigation

Chat responses contain deep-links to specific blocks.

Visual Context

See the inputs, outputs, and signal flow immediately.

Automated
Documentation

UnboxLogic acts as an autonomous agent in your CI/CD pipeline. It detects changes in your Simulink models and automatically updates the corresponding functional documentation.

No more stale wiki pages. The AI explains complex logic changes—like the switch to Kalman Filter estimation—in plain English, stored separately from the code.

Natural Language Docs
Commit-Linked Updates
CHANGED: soc_core.slx
I_meas
V_meas
Temp_C
Subsystem
EKF_Core State Estimation
(Discrete)
SOC_%
git commit -m "feat: replace coulomb counting with EKF"
[master 8f2a1b] feat: replace coulomb counting with EKF
unbox-doc generate --target=models/soc_core.slx
docs/bms_design.md M
Updating Docs...

SOC Estimation Strategy

### Change Log (Auto-Generated)
- **Updated:** Estimation logic switched to Extended Kalman Filter.
### Methodology

The previous Coulomb Counting method has been replaced.

The new **EKF Core** block fuses current integration

(prediction step) with voltage-based correction

(update step) using the OCV-SOC curve.

Note: This improves estimation accuracy by rejecting current sensor noise and correcting SOC drift over long drive cycles.
UnboxLogic Terminal
>
Explain the logic for the battery thermal management override.
search_documentation("thermal override")
read_file("bms_thermal.cpp")
*
Found logic in bms_thermal.cpp lines 402-450.
if (temp_cell_max > TEMP_CRITICAL && coolant_flow_rate < MIN_FLOW) {
  trigger_derate(DERATE_LEVEL_2);
  activate_pump_override(true);
}
The system overrides pump control when cell temp exceeds critical threshold while flow is low.
>

Context-Aware
Engineering

Ask questions in plain English. The AI maintains full repository context, indexing your entire codebase including .slx, .c, .cpp, and .m files.

It traces deep dependencies across languages to answer complex queries.

"Where is the EGR valve failsafe defined?"

UnboxLogic scans 3 repos, tracing signal path from egr_control.slx to driver in powertrain_core.c.

Automated
Triage

Diagnose issues instantly. Upload experimental data (logs, CAN traces) and UnboxLogic analyzes it against your entire repository.

Watch the AI agent autonomously work—searching code, reading files, and checking git history—to pinpoint the root cause in seconds.

Experimental Data Analysis
Interactive Consultation
Issue Triage
/
openecu/stm32-firmware
Analysis Complete
Issue Description
Idle speed instability & stalls when T_eng < 10°C.
Investigation Steps 15 Calls
semantic_search("idle surge cold")
120ms
read_file("src/control/pid.c")
45ms
View 13 more calls
git_blame("pid.c")
80ms
read_json("config/idle_map.json")
20ms
find_refs("PID_CAL_TABLE")
210ms
Root Cause Identified

Improper tuning of idle_pid_config.

Controller lacks temperature-based RPM scaling, causing integral windup.

Suggested Fixes

  • 1 Validate `idle_temp_scale` to ensure smooth transitions.
  • 2 Add hysteresis to filtering to reduce sensor noise.
50%
Engineering Time Lost

Tracing undocumented spaghetti logic.

45d
Avg Resolution Time

For complex legacy software bugs.

500+
Monthly Issues

Piling up due to knowledge gridlock.

Accelerate Engineering with AI.

Don't let software and calibration hold back your launch. Reduce debugging cycles from weeks to minutes and ensure Start of Production (SOP) stays on track.

ENTER