Drone installation
Mounted on energized lines by drone or hot stick. No outage, no customer interruption.
Hardware · AI platform · OptimizationObservability and intelligence
Observability and intelligence
for the power grid.
We build smart sensors for the power grid, artificial intelligence platforms and optimization solutions that turn distribution and transmission networks into observable, predictable and optimized systems.
Medium voltage to 500 kVSelf-poweredGNSS-synchronizedEmbedded AI
Fault locationHigh-impedance faultsDynamic Line RatingPower qualityPartial dischargeSurges and transientsState estimationDistributed generationAsset monitoringAnomaly detectionFault locationHigh-impedance faultsDynamic Line RatingPower qualityPartial dischargeSurges and transientsState estimationDistributed generationAsset monitoringAnomaly detection
What we do
From measurement to decision.
A technology company for the power sector: we generate the missing data, build the intelligence that interprets it and deliver solutions that improve operation and planning.
01
Observability hardware
Smart conductor-mounted sensors, self-powered and synchronized, for distribution and transmission.
02AI platform
Event analysis, diagnostics and forecasting with machine learning, integrated with utility systems.
03Optimization & R&D
Research and development projects with utilities: models, algorithms and tools tailored to real grid problems.
Why now
The grid has changed. Instrumentation hasn't.
Distributed generation, renewables and electrification made power flows volatile and bidirectional. Yet most feeders and lines are still measured only at the substation.
01
Blind spots
Between substations lie kilometers of network with no measurement at all. That is where faults, losses and overloads happen.
02
Instrumentation is expensive
Conventional metering requires outages, cabinets, CTs, VTs and communication infrastructure at every point.
03
Data that never arrives
Even where sensors exist, data volume and field communications limit what actually reaches the control room.
The device
A complete measurement point, the size of a sphere.
Sensing, power, timing, processing and communication integrated into a single device mounted on the conductor — no outage, no additional infrastructure.
- 1
Spherical housing
Geometry designed for high voltage, weather and sealing. Side opening to clamp onto the cable.
- 2
Timing and radio
GNSS time for synchronized measurements across distant points. Local radio between phases and long-range uplink.
- 3
Embedded processing
Continuous analysis at the edge, with AI models that detect and classify events inside the sensor.
- 4
High-fidelity sensing
Current, voltage and fast transients measured directly on the conductor, with event-triggered oscillography.
- 5
Conductor mounting
Clamped directly onto the energized cable, at line potential. Measurement is taken on the conductor itself, without opening the circuit.
- 6
Energy harvesting
Draws energy from the magnetic field of the current in the cable, powering the sensor and recharging its internal battery.
- 7
Internal battery
Keeps the sensor measuring and recording during faults, outages and periods of low line current.
- 8
Clamp and installation
Closes and locks onto the cable in a single operation, by drone or hot stick.
Built for the line
Installs in minutes. Runs for years.
Self-powered
Energy harvested from the conductor itself, with a reserve to keep operating during blackouts.
GNSS-synchronized
All sensors share the same time base, so events can be compared across points kilometers apart.
Embedded AI
Models running on the sensor identify faults, disturbances and anomalies as they happen.
Three phases, one point
One sensor per phase. The three talk to each other and deliver a complete picture of the point.
Intelligent communication
Sends information, not raw data. Full oscillography on demand or when an event occurs.
Voltage classes
From the feeder to the transmission line.
One architecture, adapted to each voltage class: distribution networks, sub-transmission and extra-high-voltage transmission lines.
- 13.8 kVDistribution
- 23 kVDistribution
- 34.5 kVDistribution
- 69 kVSub-transmission
- 138 kVSub-transmission
- 230 kVTransmission
- 500 kVTransmission
Applications
What changes when the grid is observable.
Each application depends on measurements that don't exist outside the substation today. With distributed, synchronized sensors, they become possible.
Distribution · 01
Fault location
Pinpoints the faulted section within moments, cutting patrol and restoration time.
Distribution · 02
High-impedance faults
Detects downed conductors and vegetation contact that conventional protection misses.
Distribution · 03
Power quality
Sags, harmonics, unbalance and transients measured along the entire feeder.
Read the technical noteDistribution · 04
Distributed generation
Real-time visibility of reverse flows and DER impact, point by point.
Distribution · 05
State estimation
Real synchronized measurements instead of pseudo-measurements, for precise grid operation.
Read the technical noteDistribution · 06
Incipient failures
Recurring event signatures reveal equipment and connections before they fail.
Differentiator · Latent telemetry with AI
The sensor transmits the latent space. The platform reconstructs the signal.
Embedded AI models compress each window of the six waveforms (voltages and currents) into a compact latent representation of just a few bytes, sent over low-bandwidth links such as LoRa or cellular. On the platform, another model reconstructs the waveform, with phase and harmonic content, and delivers the information each application needs.
Edge AILatent spaceLearned compressionEvents preserved
- 01
AI on the sensor
A model running on the device learns to represent electrical behavior compactly.
- 02
Latent space
Each measurement window becomes a latent representation of a few bytes instead of thousands of samples.
- 03
Low-bandwidth link
A few bytes per window fit long-range, low-power radios.
- 04
Reconstruction on the platform
The waveform is rebuilt in the cloud, with direct extraction of phasors, harmonics and events.
Sensor → Platform
In steady state the sensor periodically sends only the latent representation. The platform reconstructs the waveform from it.
Raw samples · 1 window
512 samples × 6 channels × 16 bits = 6,144 bytes
Latent representation · 1 window
≈ 16 bytes
Concept illustration. The byte comparison is representation arithmetic; reconstruction quality is the subject of ongoing research.
AI & optimization
Intelligence to operate and plan the grid.
Beyond hardware, we develop machine learning and optimization models for power systems. They work with our sensors, with existing measurements and with the data utilities already have.
AI state estimation
AI models that estimate grid state from few measurements and update in real time.
Fault location and classification
Models trained on oscillography and electromagnetic simulations to identify fault type, cause and location.
Load and DER forecasting
Feeder- and point-level forecasts to support operation, energy procurement and planning.
Dynamic line capacity
Thermal and weather models that compute real ampacity and unlock transmission margin.
Optimal sensor placement
Optimization that decides how many sensors to install and where, reaching the grid observability target at minimum cost.
Grid digital twin
Continuous grid simulation fed by real measurements to test scenarios before acting in the field.
Platform
From sensor to decision.
Sensors form a layer of synchronized data across the grid. The platform turns that layer into operational information and integrates with the systems utilities already use.
- 01
Sensors
One per phase, at each point
- 02
Local link
Phases talk to each other
- 03
Long range
Cellular or LoRa
- 04
G4W Platform
Events, maps and analytics
- 05
Your systems
SCADA, ADMS, GIS
- Geo-electrical map with the state of every point
- Waveforms and event oscillography
- Automatic alarms and diagnostics
- Power quality reports
- Integration APIs
- Remote device and firmware management
R&D and innovation
Projects with utilities, from problem to field-proven solution.
We take part in R&D and open innovation projects with distribution and transmission utilities. We start from the operational problem and deliver technology validated on the grid.
- 101
Diagnosis
We start from the operational problem: recurring faults, losses, power quality, line capacity or planning.
- 202
Modeling and simulation
Grid models and historical data define the solution: algorithms, required sensors and expected results.
- 303
Field pilot
Deployment on real feeders or lines, with sensors, platform and KPIs agreed with your team.
- 404
Scale
From proof of concept to operation: systems integration, training and continuous model improvement.
Company
Power systems engineering with artificial intelligence.
G4W is a Brazilian technology company for the power sector. We combine power systems, instrumentation, machine learning and optimization to build the hardware, platforms and algorithms that bring observability and intelligence to the grid.
“The next generation of power grids needs more than sensors. It needs a new observability infrastructure.”
Mission
Transform power system observability by democratizing access to high-quality measurements through innovative, accessible and intelligent technologies.
Vision
Make it possible to observe, understand and optimize every relevant point of transmission and distribution networks.
Engineering rigor
Physics, simulation and testing behind every decision.
Hardware + software
We build the complete solution: from the sensor on the cable to the software that turns measurements into decisions.
Applied science
Research and close collaboration with academia.
Total cost
We optimize the system, not just the part.
Contact
Let's make your grid observable.
We work with utilities, transmission operators, research centers and industrial partners.
Talk to us
Tell us about your grid's challenge. We'll reply with a technical conversation.
Send an emailDistribution utilities
R&D projects, open innovation and feeder pilots.
Transmission operators
Line monitoring, DLR and fault location.
Research centers
Joint projects, data and publications.
Manufacturers & integrators
Integrating sensors and platform into existing portfolios.
Investors
Deep tech in critical infrastructure.