Inside Datakrew's mobility intelligence tech stack

Datakrew's SOH prediction methodology is proprietary and represents the core of the company's technical differentiation. The exact algorithms and model architecture are confidential.

What can be shared is that SOH prediction draws on a large number of parameters assessed continuously across each vehicle — covering dimensions such as cell-level voltage and current behaviour, thermal history, charging and discharging patterns, depth of discharge trends, usage intensity, and climate of operation, among others. The approach combines real-world telemetry with physics-informed modelling to account for the non-linear nature of battery degradation, which conventional BMS systems are not designed to capture.

The result: Datakrew can predict vehicle SOH up to 26 months in advance with up to 97% accuracy. This capability is field-tested in partnership with some of the leading OEMs in the industry, and is grounded in over 105 million kilometres of labelled, real-world telemetry data collected across commercial EV fleets operating in diverse climates and duty cycles.

For fleet operators, this translates into defensible, forward-looking battery health forecasts that support warranty provisioning, resale valuation, maintenance scheduling, and regulatory compliance — without relying on lab simulations or manufacturer-provided degradation curves.

Datakrew's ITUS Max hardware and OXRED MyFleet platform are compatible with 35+ OEM brands and over 100 vehicle models spanning passenger EVs, light commercial vehicles, electric buses, heavy trucks, and specialised industrial equipment.

Compatible manufacturers include BYD, Tata Motors, Hyundai, Kia, Tesla, MG Motors, Mercedes-Benz, BMW, Audi, Toyota, Vinfast, Xpeng, Foton, Sany, Maxus, and many others. The full compatibility list covers both established global OEMs and regional manufacturers across Asia and Europe.

ITUS Max is powertrain-agnostic and OBD-II compatible, making it deployable across mixed fleets that combine EVs, ICE vehicles, and hybrids from different brands — without requiring separate hardware configurations per make or model.

For the complete and up-to-date compatibility list, visit Datakrew's OEM Compatibility page.

State of Charge (SoC) and State of Health (SoH) are both expressed as percentages, but they measure entirely different things — and confusing the two is one of the most common and costly mistakes in EV fleet operations.

SoC is a real-time operational metric. It tells you how much energy is available in the battery right now — essentially the fuel gauge equivalent for an EV. SoC fluctuates constantly as the vehicle drives and charges. It determines whether a vehicle can complete its assigned route today. A dispatcher relies on SoC every morning before vehicles leave the depot.

SoH is a long-term asset health metric. It measures how much of the battery's original capacity remains compared to when it was new. A battery at 85% SoH can only store and deliver 85% of its original energy — meaning a vehicle that once had a 300 km range will now achieve around 255 km under the same conditions. SoH changes slowly over months and years, driven by charging patterns, thermal history, depth of discharge, and usage intensity.

The relationship between the two matters operationally: a battery at 80% SoC and 90% SoH behaves very differently from a battery at 80% SoC and 70% SoH. The first vehicle is well-charged and healthy. The second is well-charged but degraded — it will deliver significantly less range than the dashboard implies.

For fleet managers, the distinction maps to two different decision horizons:

SoC informs daily dispatch decisions — which vehicles go on which routes, which need charging before the next shift.

SoH informs strategic asset decisions — when to reassign vehicles to shorter routes, when warranty claims are warranted, when to plan replacements, and what a vehicle is worth at lease return or resale.

Datakrew's OXRED MyFleet platform tracks and presents both metrics in a unified dashboard, giving operations teams the daily readout they need and fleet managers the long-term visibility to protect asset value and plan ahead.

Most commercial fleets in transition are not fully electric — they operate a combination of EVs, ICE vehicles, and increasingly hybrids across the same depots, routes, and teams. Managing these on separate platforms creates data fragmentation, additional overhead, and blind spots precisely where operational decisions matter most.

Datakrew's OXRED MyFleet is built to handle mixed fleet environments on a single platform. Fleet operators access both EV and ICE vehicles through the same interface, with a dropdown that switches between vehicle types — presenting each with the parameters most relevant to its powertrain. EV assets surface battery-first intelligence: State of Charge, State of Health, cell-level anomalies, charging behaviour, and range prediction. ICE assets surface the diagnostics, utilisation, and maintenance signals that matter for combustion vehicles. The underlying platform, data pipeline, and AI layer are shared across both.

What this means in practice:

Fleet managers get a holistic view of total fleet health, utilisation, and operational readiness — without switching platforms, reconciling separate reports, or maintaining parallel systems. Strategic decisions like replacement planning, route assignment, and maintenance scheduling can be made with visibility across the entire fleet, not just the EV portion.

Driver behaviour monitoring, predictive maintenance triggers, and GuardianAI's automated workflows apply across all vehicle types. A fault on an ICE vehicle and a battery anomaly on an EV are both surfaced, prioritised, and actioned through the same intelligence layer.

For operators managing fleets in active transition — adding EVs progressively while running existing ICE assets — this means one deployment, one operational platform, and no disruption to existing fleet management processes as electrification scales.

Most fleet management platforms are built around alerts. They monitor vehicle data, detect a condition outside a defined threshold, and send a notification to a human — who then decides what to do, coordinates with the relevant team, and manually initiates the appropriate action. The intelligence stops at the notification.

GuardianAI is Datakrew's proprietary AI agent layer, and it is architected differently. Rather than stopping at the alert, GuardianAI is designed to act — autonomously initiating the next step in the workflow based on predefined operating procedures, without requiring manual intervention to bridge the gap between detection and resolution.

The practical difference:

A standard alert system tells a fleet manager: "Vehicle 047 has a battery anomaly."

GuardianAI identifies the anomaly, assesses its severity, cross-references the vehicle's upcoming schedule, and can automatically trigger a maintenance work order, notify the service team, flag a replacement vehicle for the affected route, and update the driver — all from a single detected condition.

Why this matters for fleet operations:

Alert fatigue is a well-documented problem in fleet management. Platforms that generate high volumes of notifications but require humans to act on each one create backlogs, missed actions, and delayed responses — particularly in large fleets or operations with lean teams. GuardianAI reduces the human coordination burden by automating the response workflow, not just the detection.

GuardianAI operates continuously across the entire connected fleet, monitoring battery performance, driving behaviour, vehicle health, and operational KPIs simultaneously. It applies Datakrew's AI and real-world telemetry data to prioritise actions by severity, ensuring that critical interventions are never buried under lower-priority notifications.

This positions GuardianAI not as an alert system with more features, but as a fundamentally different class of tool — an autonomous AI fleet manager that acts on data rather than simply reporting it.

The battery is the single most valuable component in any electric vehicle, representing a significant share of the asset's total cost. As a result, battery State of Health (SoH) has become the defining factor in how EVs are priced at resale, how warranty claims are assessed, and how OEMs provision for lifecycle obligations.

Residual value

Industry data indicates that each percentage point of SoH loss reduces the resale value of a mainstream commercial EV meaningfully, with the effect accelerating below 85% SoH. Two vehicles of the same make, model, age, and mileage can carry materially different market values depending on documented battery condition.

For fleet operators, this creates a direct financial case for active battery intelligence. Datakrew customers who follow Datakrew's battery management recommendations — optimising charging behaviour, avoiding thermal stress, and acting on early anomaly alerts — have observed approximately 20% longer battery life compared to unmanaged operation. This extended battery life translates to an approximate 10% improvement in salvage value at end-of-life or defleet, though the precise figure varies by vehicle make and model, market, and country of resale.

The compounding effect across a fleet is significant. A 10% improvement in salvage value on a fleet of 200 vehicles represents a material return — one that is directly traceable to the operational decisions Datakrew's intelligence makes possible.

Beyond the salvage figure, actively managed battery health slows the degradation curve that erodes vehicle value throughout the asset's life — not just at the point of disposal. Charging optimisation reduces thermal stress from fast charging sessions. Driver behaviour insights reduce harsh patterns that stress the battery under load. Early cell-level anomaly detection prevents minor issues from escalating into pack-level damage. Together, these interventions keep SoH higher for longer, which is where residual value is protected in practical terms.

Datakrew's SoH records also provide the certified, data-backed documentation that buyers, lessors, and financiers increasingly require to price used EVs with confidence — moving valuation from assumption to evidence.

Warranty management

Standard OEM battery warranties are typically structured around an SoH retention threshold — commonly 70% over an 8-year or 100,000 km period. Without continuous SoH monitoring, fleet operators frequently miss the window to initiate warranty claims before the coverage period lapses or degradation crosses the threshold undetected.

Datakrew's platform tracks SoH continuously against each vehicle's warranty parameters, enabling proactive identification of claim-eligible vehicles before the window closes. For OEMs, this same longitudinal data supports warranty provisioning decisions — knowing the actual degradation trajectory of deployed assets, stratified by model, duty cycle, climate, and charging behaviour, enables far more precise reserve calculations than lab-based degradation assumptions allow.

Datakrew's SOH prediction capability — forecasting battery health up to 26 months in advance with up to 97% accuracy, field-tested with leading OEMs — gives both fleet operators and manufacturers the forward visibility to plan warranty actions, schedule replacements, and manage asset lifecycle decisions with confidence rather than reacting to degradation after it has already affected operations or value.