AI-READY ENERGY DATA

Talk to your energy data

Metry’s MCP server connects your data directly to the AI tool you already use. Ask in plain language and get the answer from your own data.

Goodbye manual exports, hello energy data!

Stay on top of data collection

Which meters have stopped reporting? Where are the gaps in last year’s data? And where do the values look implausible? Spot anomalies in your data and find the gaps before they turn into a reporting problem.

Follow-up and analysis

Compare energy use between periods, find the buildings that stand out and ask for an energy signature while you’re at it. It used to take an export and a pivot table to get done. Now, you just ask.

A basis for reporting

Pull together annual consumption by utility across the whole portfolio ahead of CSRD or GRESB. Check coverage before you start reporting, produce the figures without opening a spreadsheet, and compare against last year with ease.

Getting your data into a system of your own has always meant a developer building an integration against our API. Now you click the Metry connector and start asking questions in plain language.

Sebastian Levander, Head of Design, Metry

Stay on top of data collection

From reactive to proactive

Build flows that fit the way you work

Which meters have not reported for two months? What share of the portfolio has complete data collection for the full year? Are there properties where we are missing data for part of the period? With the MCP server, you can ask your data directly and keep track of collection.

Move from reactive to proactive by setting up your own automated alerts and flows in the AI tool, drawing the answers from your actual Metry data.

Freedom to choose

Your data in the AI tool you already use

Mix and match as you please

Connect the AI tool you already use. The MCP standard is open, which makes your data available in tools that supports MCP.

Instead of exporting a file and pasting it into a chat, the AI tool fetches the data straight from Metry when it needs it. The standard works like a universal connector, so the same connection works across several tools. If you switch AI vendor in two years, your energy data comes with you.

IT and security

No new permissions, no new keys

Your IT department can rest assured

The connection is read-only, so nothing can be changed or deleted through it. You sign in with your existing Metry account, and the AI tool sees exactly the data that account already has access to. No service accounts, no API keys, no shared passwords.

The server runs on AWS in the EU and stores neither credentials nor conversations.

Read more about IT and security

Up and running before your coffee gets cold

Add the MCP server
Sign in to Metry
Ask your data

One address is all it takes

Open the connections in your AI tool and add https://mcp.metry.io. In most tools that is one field and one button. If you work in a CLI, a single command will do.

Get started

If you already use Metry

Your data is already in the portal. All that is left is connecting the server in your AI tool. Sign in and follow the instructions for the tool you use.

Sign in to the Metry portal

If you are not a customer yet

An AI tool is never better than the data it is fed. If your energy data is spread across grid operators, submeters and spreadsheets, that is where the work starts. Get in touch and we will look at how your data collection works today.

Book a demo

Down to the details

You ask, we answer

Convince the sustainability manager that the answers can be trusted. Give IT what they need to say yes. Show the operations team that they can carry on working exactly as before. These are the questions we get most often. If you have more, book a meeting with us and we will go through them together.

What is MCP, exactly?

MCP stands for Model Context Protocol and is an open standard for how AI tools connect to data sources. Instead of exporting a file and pasting it into a chat, the AI tool fetches the data straight from Metry when it needs it. The standard works rather like a universal connector, which means the same connection works across many different tools.

Which AI tools does it work in?

Tools that support MCP. Today that includes Claude Code, ChatGPT, GitHub Copilot in your development environment, Mistral Vibe, Gemini CLI and other terminals. Beyond those there are several other MCP clients, and most of them connect in the same way.

Or you can run local models in Ollama or AnythingLLM, for instance. Yes, the MCP server works there too!

What data can the AI tool access?

Exactly the data your Metry account already has access to. That means consumption and readings for electricity, heating, cooling, water and gas, meter metadata such as name, address, EAN and energy type, and property and building structure. No new permissions are granted, and no personal data beyond the account used to sign in.

Can the AI change or delete anything?

No. The connection is read-only. Data can be read, but not created, changed or deleted.

What does our IT department need to do?

Allow outbound HTTPS to mcp.metry.io. No inbound rules are needed, no VPN and no IP allowlisting. Authentication uses OAuth against your existing Metry account, so there is no need for service accounts or API keys. The server runs on AWS in the EU and stores neither credentials nor conversation data.

The data the AI tool receives is handled under your own agreement with the AI vendor. We recommend using a business or enterprise tier account, as personal accounts often lack a data processing agreement.

A more detailed basis for security review is available here.

What happens if we remove someone’s access?

The connection to the MCP server has no permissions of its own. It uses that person’s Metry account. If you disable the account in Metry, access ends, and nothing needs to be done in the AI tool.

Can we trust the answers the AI tool gives us?

Language models are generally poor at arithmetic, and an answer is never better than the data beneath it. That is why we at Metry have put all our effort into the data layer, with quality assurance that gives the model the right figures to work with from the start, and full traceability of where the data came from. Read more about our quality assurance.

Our advice is the same as the advice we give ourselves. Start with questions you can verify, have someone who knows the portfolio look over the answers early on, and treat the AI tool as a fast colleague who is sometimes wrong rather than as the final word.

Do we need a developer to get started?

No. With the MCP server, all it takes is adding an address in the tool you already use. Our non-developer colleagues have managed it too.

Our API is still there and works exactly as before for anyone using ready-made integrations or who has built their own against it.

Can we combine Metry data with data from other systems?

Yes, provided the other sources are also connected in your AI tool. This is one of the bigger changes MCP brings. Combining energy data with finance and property information used to be an integration project running over several months. Now it is a question in a conversation.

What would you ask your energy data?

Book a demo and see it in action
Part of the EVORA Group

© 2026 Metry | Kämpegatan 6, 411 04 Gothenburg, Sweden