EQ Syntropy
Physics-Grounded Intelligence for Real-Time Power Systems
EQ Syntropy™ finds what is going wrong in your power system, explains why, and shows you the evidence. It is the intelligence of the Energy Quotient platform: AI that reasons across every device’s record in EQ Coherence™, from live data to weeks of full-resolution waveform history, to find precursors, correlations, and root causes.
Ask a question in plain language in EQ Sight™, and EQ Syntropy runs the investigation an engineer would: it pulls the right windows of data, runs the analyses, and answers with every value cited to the measurement it came from.
The Intelligence Runs Beside Your Data
EQ Syntropy runs on our servers or on edge AI hardware at your site, beside EQ Coherence, and does the heavy lifting there. Its orchestration routes each question by how strictly it must be handled: rule-defined pipelines for work that has to follow the standard, such as compliance checks, and model-driven investigation with tool calling for open questions. Either way, the model works from a fixed set of analyses and must cite every value, while the analysis tools and internal models work through the full-resolution waveform record where it lives, and its context engine keeps the graph of events and system state. It reaches any event in the stored history, down to the individual waveform, in under a second, querying the recorded Parquet files directly with DuckDB. Only compact, cited results reach the language model on top: your choice of a frontier model or a self-hosted open-weight one. Raw waveforms never leave the system, and you never spend tokens pushing them through a model.
Answers With the Evidence Attached
- “What caused last Thursday’s voltage sag?”
- “Have we seen this harmonic pattern before?”
- “Which circuit’s current distortion rose this week, and when did it start?”
An agent works each question through deterministic analysis tools that run on the measured record itself:
- Waveforms and metrics: any window of the 32 ksps waveform or the cycle-by-cycle metrics, from any device
- Spectrum and distortion: harmonics through the 50th, and total harmonic distortion
- Events: sag and swell detection, and segmentation of operating sessions
- Voltage quality: voltage distribution against ANSI C84.1
- Energy: energy use over any period
- History and context: queries across the live and historical event and state graphs, with a scratch graph for each session
The same tools are open to you: through an MCP server, your own software and AI agents can call them on the same measured record.
Every answer cites the values, time windows, and analyses behind it, so an engineer can check each claim against the waveform. Results arrive as tables, spectra, and waveform views that open in the EQ Sight explorer, and EQ Syntropy recommends next steps for your equipment and process.
Continuous Condition Monitoring
EQ Syntropy watches every circuit cycle by cycle, flagging RMS threshold crossings, shifts in power factor and harmonics, and load steps as they happen. A signal floor and damping keep idle or noisy circuits from raising false events. Each event lands in the event graph the agent queries, so the next question starts with it.
Analysis on Demand
When the built-in analyses are not enough, EQ Syntropy writes and runs its own Python against the full-resolution data: a custom KPI for your process, a comparison across circuits, or a plot for a report. You see the code it ran and the result.
Standards checks run against IEEE 519, SEMI F47, and the ITIC (CBEMA) curve, and every figure in a compliance report traces back to the waveform behind it.
Reports are written with you: EQ Syntropy drafts the findings, you edit them in the same document, and every version is saved.
Grounded, Checked, and Traceable
Language models alone are not enough for critical power systems. EQ Syntropy uses the language model to investigate and leaves the measuring to deterministic code:
- Grounded in measurement: every number comes from a tool that computed it from the waveform record, not from the model’s memory.
- Checked continuously: answers are scored against reference power quality scenarios, and an independent evaluator checks live answers for grounding in their cited evidence.
- Fully traceable: every analysis the agent runs is recorded, so any conclusion can be audited after the fact.
Where It Runs
- Hosted by us or on your hardware: our servers, or NVIDIA Jetson Thor, NVIDIA DGX Spark, or an x86 computer with a discrete GPU on the same network as EQ Coherence.
- Entirely on your premises: with a self-hosted model, the whole system runs on site, including on isolated networks, with no per-query fees or usage limits.
- The model is a choice, not a dependency: the analyses and condition monitoring run without a language model attached.
Whether you run a semiconductor fab, a defense microgrid, or a medical imaging facility, EQ Syntropy turns your waveform record into answers your team can act on.
Book a demo
Already have access? Sign in at demo.eq.systems.