Solar, battery, and EV charging,
in one home interface.

Introduction

Most solar homeowners juggle three or four apps just to understand where their energy is going. Powerbeat is a concept for what happens when you combine the best ideas from those tools into one interface, designed around how homeowners actually think about energy, not how inverter engineers wire their systems.

The idea took shape after analyzing a full year of solar data and visiting the Intersolar trade fair in June 2024. A Home Energy Manager should not feel like a collection of disconnected utility apps.

The Concept

Powerbeat is a working-name UI prototype. It does not mirror a single existing app. The goal is to combine the most useful ideas from today’s solar, battery, and EV tools into one product surface.

What Solar Owners Actually Need

Four patterns emerged from a year of daily solar data:

• Too many apps. Tesla for the Powerwall, SMA for production forecasting, EVcc for charging and dynamic pricing. The core problem is fragmentation.
• Summer mode vs winter mode. A typical system is oversized in summer and undersized in winter. The HEMS has to adapt automatically.
• Unpredictable consumption. The car is the biggest wildcard. A single charge can consume an entire day’s production.
• Daily production is the key question. Below 20 kWh? You are buying from the grid. Above 60? Charge the car for free.

Daily Production	Day Category	Category Label
0 - 20 kWh	Grid Import Needed	Red
20 - 40 kWh	House Only	Orange
40 - 60 kWh	House Plus	Blue
60 - 100 kWh	Charge Car	Green
100+ kWh	Full Charge	Yellow

How Home Energy Got Complicated

• Grid only (Year 2000). One fixed price per kWh. No personal optimization needed.
• Solar system (Year 2012). Use more energy when the sun is shining, and less when it is not.
• Home battery (Year 2018). Store daytime energy for nighttime use.
• Dynamic power plan (Year 2022). Hourly market prices turn timing into an optimization problem.
• Battery charging from grid (Year 2024). Charging from the grid becomes attractive on cheap winter nights.
• Battery grid export (Year 2026). Home batteries stop being passive storage and become active market participants.
• Bidirectional car charging (Year 2026). The car battery becomes part of the home’s energy system.

User Interface

In my view, a strong Home Energy Manager needs four main surfaces:

Powerbeat Now	Powerbeat Today	Powerbeat Devices	Powerbeat Insights
Energy flow in real-time	Today’s data and forecast	Charging planner	GenAI-powered reports
Status quo of the energy flow right now.	Show today’s solar production, forecast, and pricing context.	Decide how devices and cars should be charged.	Use LLMs to produce helpful energy reports.
Data in kW: Solar power, battery charging, home usage, grid export, and more.	Data in kWh: Generated today, forecast for the rest of the day, comparisons, and more.	Charging modes: Solar only, fast charging, minimum charging, disabled, and more.	Reports: Top production days, monthly usage, charge cycles, and other analysis.

Now

How much power is the roof generating right now, and where is it going? The Now screen answers that question in real time: solar generation, home consumption, battery charge rate, and grid flow, all in one view.

The Tesla app uses a house animation for Powerwall users:

EVcc makes the proportions between energy flows easy to read:

Powerbeat uses a Sankey diagram instead. It shows relative flow volumes and makes devices like the home battery feel like first-class participants in the energy system, not just numbers on a dashboard:

House visuals can still be nice, but only if they are truly personal. If an app shows my house, I want it to be my house, not a generic icon.

Today

The daily production number drives every energy decision. It determines whether the battery charges from the sun, whether the car gets a free charge, or whether the household buys from the grid.

The main goal of the Today section is to decide which bucket the current day is heading toward and what that means operationally.

The Tesla app only shows today’s generated kWh:

The SMA inverter app goes further and shows a forecast for the next hours and days, but the forecast data does not update throughout the day:

Powerbeat should treat forecast uncertainty more explicitly. A better visualization is a range between best-case and worst-case production:

Devices

The Devices screen is where energy strategy becomes action. The hard part is not the UI, it is the policy underneath: what should the car do when production is 35 kWh and a cold week is forecast? The answer depends on tariffs, battery state, and a household’s willingness to pay for convenience.

Charging Modes

Every vendor invents its own vocabulary for the same handful of behaviors. EVcc uses Off, Min+Solar, Solar, and Now. Tesla’s wallbox app offers Self-Powered, Time-Based Control, and Storm Watch. SMA, KEBA, and Easee each ship variants of the same set. The shared underlying question is simple: should the car wait for the sun, take whatever is on the wire right now, or fall back to a guaranteed minimum?

Powerbeat treats charging as a policy decision, not a per-session toggle. Set the rules once, let the system follow them. Four operational questions drive the design:

Device priority. When solar is scarce, which load wins? Heat pump for the morning shower, car for the school run, or battery for the evening peak? The answer changes by season, time of day, and weekday vs weekend. Those defaults need to be visible on the Devices screen, not buried in a settings menu.

Solar-only vs grid thresholds. The decision to top up from the grid should be driven by the forecast from Today. If tomorrow looks under 20 kWh and tariffs are favorable tonight, charge from the grid. Otherwise wait for the sun. This is the link between the daily forecast and the operational plan.

Automation vs manual override. Defaults should not surprise. If the household plugs in at 18:00 and expects a full battery by 07:00, the system should hit that target without anyone touching a slider. Overrides exist for the edge cases (an unexpected long drive tomorrow), but the resting state is hands-off.

Seasonal switches. Summer surplus and winter scarcity are different operating modes, not the same logic with different inputs. In summer, top up everything: house battery, car, hot water. In winter, arbitrage cheap grid hours and reserve the battery for the evening price peak. The screen should make the seasonal mode explicit, not hidden in fine-tuning.

Screenshots will follow once the charging-mode panel is in a presentable state.

Insights

Energy data is a time series. The value is in the patterns, not the rows. Traditional dashboards show charts you have to design upfront, which means every new question becomes an engineering task. An LLM-backed surface lets you ask the question first and pick the visualization second.

Natural-language queries replace SQL and notebook glue. The bar for a useful report drops from “build a dashboard” to “ask a question.” Exploratory analysis becomes cheap, and the questions a homeowner actually has are not the questions a dashboard designer anticipated.

As a starting point, check out the Custom GPT for my solar data from 2023. The following prompts are a useful baseline:

Plot daily solar production

This is the base view: it shows the annual shape and highlights the months where production drops sharply.

Other prompts for Powerbeat Insights

• List the top 10 production days in a sorted table.
  This surfaces the system’s production ceiling, shows when it happens, and estimates how many “Full Charge” days the system can realistically deliver.
• Calculate monthly battery cycles using a 13.5 kWh battery capacity.
  This converts daily energy flowing in and out of the battery into cycle wear, the metric that matters for manufacturer warranties.
• Create a monthly solar production bar chart.
  This shows the seasonal envelope, supports sizing decisions, and gives a factual answer to anyone claiming solar does not work in winter.
• Find every streak of 2 or more consecutive days below 20 kWh of solar production.
  Return a table with the first day and streak length. This shows how often the household is forced onto the grid for back-to-back days and informs battery sizing.
• Identify likely snow days in 2023 from the production data.
  This connects production anomalies with implied weather: a near-zero production day in March, when comparable days produced far more, is a clear signal.

Limits and Concerns

Cloud LLMs see your home telemetry, which is more revealing than most people realize: production patterns leak occupancy, charging habits leak driving routines, and battery state-of-charge leaks evening behavior. Long histories blow past context windows, so multi-year analysis needs aggregation or function-calling. Numerical accuracy needs verification, because an LLM will confidently report a wrong average. A local LLM via Ollama is the obvious next step for the privacy-conscious and the offline-first.

Powerbeat is the conceptual surface; PowerNight is the production tool that puts these energy ideas into practice.