Energy Under Control: How Integrating Energy Meters with the ASI Biont AI Agent Automates Metering and Reduces Costs by Up to 35%

Introduction

Electricity is one of the largest expense items for any enterprise. According to the International Energy Agency (IEA), industrial facilities spend up to 30% of their operational budgets on electricity, and in office buildings this figure can reach 20%. Traditional Building Management Systems (BMS) collect data but cannot analyze it in real time or predict future peaks. As a result, companies overpay for capacity, lose energy due to suboptimal equipment operation modes, and fail to notice anomalies until a bill with an exorbitant amount arrives.

Integrating the Energy Meters service with the ASI Biont AI agent solves this problem. Instead of manual meter reading and monthly reports, you get an automated system that continuously monitors consumption, builds forecasts based on machine learning, and offers specific actions for savings. And most importantly, you don't need to wait for platform updates to connect. All you need is an API key from your meter and a couple of minutes of conversation with the AI in the chat.

What Are Energy Meters and Why Connect Them to an AI Agent

Energy Meters are a class of IoT devices that measure electricity consumption in real time. They range from simple household meters with Wi-Fi (e.g., Shelly EM or Sonoff POW) to industrial systems with three-phase metering (e.g., Siemens PAC or Schneider Electric PowerLogic). Each device collects data on voltage, current, power, frequency, and accumulated consumption, then transmits it via an API to a cloud server.

The problem is that raw data is just numbers. To turn it into useful insights, you need analytics. The traditional approach: you manually export a CSV file once a month, upload it to Excel, and build charts. This takes hours, and most importantly, you react to problems after the fact.

The ASI Biont AI agent, by connecting to the Energy Meters API, takes over all the routine: it automatically fetches data every 5-15 minutes, cleans it of noise and outliers, builds consumption models, and alerts you to any deviations. This is not just automation—it's a shift from reactive to proactive management.

How ASI Biont Connects to Energy Meters: No Panels, Just Chat

Many IoT platforms promise "one-click integration," but in reality require filling out long forms, selecting a device type from a dropdown list, and waiting for developers to add support for your model. ASI Biont works differently.

Connection happens through a dialogue with the AI agent in the chat. You simply send a message: "Connect my Energy Meter from manufacturer X, here's the API key: sk-xxx." The AI independently reads the API documentation of your device (from a link you provide or from open specifications), writes Python integration code, tests it, and starts collecting data. All this takes a few minutes, not days.

Moreover, ASI Biont can connect to any service that has a REST API or WebSocket. You don't need to wait for the asibiont.com team to add official support—you decide which devices to use. This provides flexibility: you can combine data from different manufacturers, even if their APIs differ.

What Tasks Does the Integration Automate

After connection, the AI agent begins solving several key tasks:

1. Automatic Data Collection and Visualization

Instead of logging into the meter's app every day, you get a unified dashboard in the chat. The AI can build consumption graphs for any period, showing hourly, daily, and monthly trends. For example, a request: "Show energy consumption for the last week broken down by day"—and the AI returns a visualization with explanations.

2. Peak Load Forecasting

Based on historical data and machine learning, ASI Biont builds a model that predicts when peak loads will occur in the coming hours or days. This is critical for enterprises that pay for capacity under a two-rate plan or are fined for exceeding limits. The AI can suggest in advance: "Tomorrow from 2:00 PM to 4:00 PM, a peak is expected. I recommend turning off non-essential equipment or shifting ventilation startup to nighttime." According to practical cases, such forecasting can reduce capacity charges by 15–20%.

3. Anomaly Detection

Energy consumption is rarely perfectly uniform. Sharp spikes may indicate equipment malfunction (e.g., a jammed motor or current leakage) or unauthorized connection. The AI agent analyzes data in real time and, upon detecting an anomaly (e.g., consumption increased by 40% in 10 minutes under normal conditions), sends a notification to the chat: "Anomaly detected on line 3. Current power: 12.5 kW, expected: 8.2 kW. I recommend checking the compressor in workshop #2." This allows you to respond in minutes, not days.

4. Cost Optimization

The most valuable feature is recommendations for reducing expenses. The AI analyzes the tariff plan, equipment operating schedule, and external factors (e.g., outdoor temperature) and suggests specific actions: "Shift the production line startup 1 hour earlier—this will avoid the daytime tariff and save 3,400 rubles per month." In the long term, according to experiments on commercial facilities, savings reach 25–35%.

Example Use Cases

Scenario 1: Plastic Products Factory

The factory has 15 three-phase PowerLogic meters on each machine. Previously, an engineer walked around the workshops once a week, recorded readings, and entered them into Excel. This took 4 hours per week. After connecting to ASI Biont, data is collected automatically. The AI noticed that one machine consumes 20% more energy on Fridays and identified that the operator forgets to turn off cooling after the shift. The notification led to a change in instructions, saving 12,000 rubles per month.

Scenario 2: Class A Office Center

The building has Smart Shelly EM meters on each floor. The AI agent analyzed data over a year and found that air conditioners run at full power an hour before employees arrive, even though the outdoor temperature is already comfortable. ASI Biont suggested adjusting the climate control schedule, reducing electricity bills by 18% without sacrificing comfort. Additionally, the AI warned about a fan malfunction on the 5th floor 3 days before it broke down—preventing downtime and a 50,000 ruble repair.

Scenario 3: Data Center

For data centers, energy consumption is the main operational expense. ASI Biont connected to the PDUs (Power Distribution Units) via API and began monitoring the load on each server. When one unit showed abnormally high consumption (30% above normal), the AI sent a notification to the administrator. It turned out that a cryptocurrency mining process had been running on the rack without authorization. Removing the process saved 8,000 rubles per month.

Comparison with Traditional BMS Systems

Many companies use Building Management Systems (BMS) from Siemens, Johnson Controls, or Schneider Electric. These systems are powerful but expensive (implementation costs can exceed 1 million rubles for a medium-sized facility) and require an integrator for setup. Moreover, BMS rarely include ML models for forecasting—they only visualize data.

How to Get Started: Step-by-Step Instructions

1. Ensure your Energy Meter has an API. Most modern devices (Shelly, Sonoff, PowerLogic, Siemens) provide a REST API or MQTT broker. If not, you can use a Modbus gateway with an HTTP interface.
2. Get an API key. In the manufacturer's app, find the "API Access" section and generate a token. Usually, it's a string like sk_1234567890abcdef.
3. Open the chat with the ASI Biont AI agent at asibiont.com. Send a message: "Connect my Energy Meter. Model: Shelly EM. API key: sk_..." Attach a link to the API documentation if available.
4. The AI will write code and test the connection. Within 1–2 minutes, you'll receive confirmation: "Integration successful. Collecting data every 10 minutes."
5. Configure scenarios. Ask the AI: "Notify me if consumption exceeds 10 kW per hour" or "Build a forecast for tomorrow and send it in the morning." Everything is configured through dialogue.

Conclusion

Integrating Energy Meters with the ASI Biont AI agent is not just about automating metering—it's a transition to intelligent energy management. You stop being a hostage to paper reports and unexpected bills. Instead, you get a system that monitors every kilowatt-hour, predicts risks, and finds hidden savings reserves.

The main advantage of the ASI Biont approach is universality. You don't need to wait for developers to add support for your specific meter. If the device has an API, you can connect it right now, just by talking to the AI. Try it yourself: go to asibiont.com, open the chat, and connect your Energy Meter. In 5 minutes, you'll see how much energy you're actually using—and how much you can save.