Guide: Understanding the NASA Demo Dashboards
Live demo is available at: bytelake.com/nasa-demo/
Understanding Our NASA Predictive Maintenance Demo: A Guided Tour of byteLAKE Cognitive Services
In Q2 2026 we released a major upgrade to byteLAKE Cognitive Services — a complete Edge AI (Private AI) platform designed for real-world industrial and back-office environments. It serves manufacturing, food production, automotive, paper mills, energy utilities and back-office teams in legal, sales and customer support. The platform handles predictive maintenance, waste tracking and reduction, production optimization, energy loss reduction, document processing automation and intelligent chatbots.
Key new capabilities include:
This article is a quick, screen-by-screen guideline to help you understand exactly what you’re seeing in the demo dashboards. If you have questions or want to explore how Cognitive Services can address challenges on your factory floor or in the back office, feel free to contact us.
RUL Prediction Dashboard
This dashboard shows how byteLAKE’s Cognitive Services automatically train and validate predictive models for Remaining Useful Life (RUL). You can see two lines almost perfectly overlapping: the teal “Actual RUL” and the yellow “Predicted RUL,” both declining smoothly over engine cycles. Below the graph are key performance metrics (MAE 13.9 cycles, RMSE 22.8, NASA score 2605) plus a Health Index degradation chart that confirms the engine is steadily losing health. Note the RUL clip slider — the system stops counting “extra healthy time” above a certain limit so operators see a clear “lots of life left” signal instead of misleading precision. In production, the system continuously learns from your data while letting operators embed human expertise for accurate, trustworthy predictions.
RUL Prediction Dashboard
Get an instant executive summary of your entire fleet. At the top you see high-level metrics: 249 engines in the fleet, 5 at immediate failure risk, average RUL of 82.8 cycles and preventive cost of $1.2M versus potential failure cost. The bottom table lists every engine with regime, predicted RUL, failure probability, Health Index and status (red “MAINTAIN”, yellow “INSPECT”, green “CONTINUE”). Most engines show red “MAINTAIN” buttons. Click any row and you instantly jump into detailed Explainable AI analysis. Maintenance and failure costs are displayed for quick ROI visibility.
Sensors & Health Dashboard
Dive deep into raw sensor behavior and health trends. This screen focuses on a single sensor (here Sensor 7 – HPC Eff) and visualizes raw readings with a smooth teal rolling mean, rolling standard deviation for variability and a color-coded scatter plot that clearly separates healthy, mid-life, and near-failure phases. You also see how individual sensors correlate with degradation, Health Index and RUL. Sensor_11_ema, for example, is the smoothed Exponential Moving Average version of the raw sensor_11 reading — like looking at the weather trend over the last few days instead of a single snapshot — so one weird measurement doesn’t confuse the model. This view serves as both a diagnostic tool and a way to validate insights from the Explainable AI module.
XAI / Root Cause Analysis Dashboard
Understand exactly why a failure is likely. This dashboard delivers both Global Root Cause Analysis (across the whole fleet) and Local Root Cause Analysis (for a specific engine) using clear SHAP waterfall charts. The left panel ranks the top sensors globally; the right panel shows exactly how each sensor pushes the prediction up or down for one engine. Some sensors have strong negative impact (degrading the engine), while others have positive impact (helping extend RUL). It clearly shows which sensors are driving risk — and which are performing well.
Counterfactual What-If Analysis
For any engine (example: ENG-021 in Regime 2), the system instantly runs counterfactual simulations. You see the original predicted RUL next to a new, improved RUL after hypothetical sensor changes. The bar chart visualizes Counterfactual Sensor Recommendations — some bars show how adjusting a sensor could increase RUL by +10 to +13 cycles. These What-If recommendations help maintenance teams evaluate strategies before taking action.
Maintenance Optimization & ROI Dashboard
Configure your real maintenance and failure costs, then let the AI simulate optimal strategies. This is the central ROI view: you see the large interactive policy curve at the bottom, key summary cards at the top (30 engines analyzed, 7 failures avoided, 9 preventive actions, $790K total cost), and sliders for alarm threshold, failure cost and service cost. At the current 70% threshold the pop-up tooltip breaks down preventive cost, failure cost and exact savings versus a pure reactive approach. The policy curve visualizes the sweet spot for your alarm threshold and shows the trade-off between preventive maintenance and reactive repairs.
Online Digital Twin – Live Simulation
Watch your fleet in real time. This Digital Twin streams live predictions, failure probabilities, and service decisions. You see the current cycle, predicted RUL (15.0 cycles), and failure risk rising to 61.1%. Two streaming graphs update dynamically and recent service decisions are listed with “INSPECT” or “CONTINUE” recommendations. You can adjust simulation speed and observe how engines degrade cycle by cycle — giving you a powerful playground to test scenarios and optimize operations.
Final Thoughts
If there’s one thing we would like to emphasize based on our experience at byteLAKE, it is that AI is far more than just data analytics. Of course you can connect AI algorithms to your MES, CMMS, SCADA, documents or accounting systems and get useful dashboards. However, if you want AI to actually achieve your business goals — reduce maintenance costs, optimize or eliminate waste, better predict demand, or turn a simple chatbot into a true agent that dialogues with partners and clients — you need two things: first, embed all the nuances and process-specific information into your AI models and augment them with your best people’s experience; second, create a model where AI and humans together deliver more than either could alone. That is exactly why we built advanced Feature Engineering, full Explainable AI with Global and Local Root Cause Analysis, counterfactual What-If scenarios, the Business Explainer, and the Industrial AI Agent into Cognitive Services. The result is a platform that doesn’t just tell you “what” is happening — it explains “why,” supports your teams when making decisions and becomes a vital part of your processes on the factory floor and in the back office. In short, you get an expert system that leverages your company data and team knowledge to suggest the right decisions and answer questions in plain language — exactly the kind of human-augmented intelligence to deliver the highest value.
We’d love to hear what you think after exploring the demo. Questions or ready to discuss your specific use case? Contact us here.