What would you like to ask your energy data?
Metry's MCP server is live! If your data is already with us, you can connect the AI tool you already use and ask your first question straight away.
For fifteen years we have collected energy data for property companies, quality assured it and passed it on to the most common systems for energy management and sustainability reporting. Throughout that time, one thing has looked the same: to do something of your own with the data, someone has had to build against our API, or export the data manually. Now there is another way. Metry’s MCP server is live, and it makes your energy data available directly in the AI tools that support the standard.
The route to your own answer used to go through a developer
Metry’s open API has been there all along and is included for all customers. Reading up on the documentation and building something that holds up in production is a craft of its own, though, and the people who can do it usually have a backlog stretching well into next year. A simple question like “which of our meters deliver hourly values?” therefore became either a small development project or a spreadsheet job someone had to sort out over an afternoon.
MCP (Model Context Protocol) is an open standard that lets AI tools fetch data directly from a source. Metry’s MCP server means your AI tool can fetch your energy data from Metry without anyone building an integration first. You log in with your usual Metry account and ask the question in the tool you already use.
Questions you can ask your AI today
These are questions that work directly against the data you have in Metry:
“What share of my meters has data for the full calendar year?”
“Which meters are missing values for January?”
“Which meters deliver hourly values and which give monthly values?”
“Compare electricity consumption in 2025 against 2024, per property.”
“Have all the sub meters in Ekbacken 2 reported values for February?”
The answer comes back in the chat, and you can keep asking from there. Ask for the same data presented in a table, ask for a summary per town, or ask the tool to take a closer look at the meters that look odd. A lot of the value sits in asking follow-up questions, and that’s something that’s been hard to do in ready-made systems.
“A few years ago customers wanted to buy one system that did everything. Today they run a separate procurement for collection alone, and when you see what ChatGPT, Copilot and Claude can do in terms of reports and analysis, another system is not necessarily needed. It will be fascinating to follow where this goes“, says Magnus Hornef, CEO and co-founder of Metry.
Start with one property
You don’t need a plan for the whole portfolio to try this out. Start small. Take one property, ask one question and see what comes back.
That is roughly how we see it being used today; someone test-runs a question ahead of an energy management meeting, discovers it saved half an hour of searching, and shows a colleague. Then it becomes a habit. Over time, property companies gain more control over their own reporting instead of depending on whatever the property system happens to offer.
The closer you work to the building, the easier it also is to scrutinise the answers: anyone who has managed the same portfolio for ten years can tell in a second whether a figure looks unreasonable.
We have told the AI what the data means
Just like you know your buildings, we know data. Metry’s almost fifteen years of experience in data collection is the foundation of Metry’s MCP server. A kilowatt hour is a kilowatt hour to anyone who has worked with energy data for twenty years. To an AI tool handed a long row of figures, it is not obvious. A value without a unit, a meter without an energy type or a time stamp without a stated period turns into guesswork, and guesswork is exactly what makes people stop trusting a tool.
“When I looked at a value myself, there was no unit stated anywhere. To those of us who have worked with this for 15 years it is obvious that we measure in kilowatt hours in Sweden, but it needs to be written down. That is part of the work we have put into the MCP server, getting it to understand what our data actually is“, says Magnus Hornef.
That is also why you can ask in plain English rather than looking up a meter point ID. The data in Metry is collected, validated and structured according to your portfolio, and it is described so that an AI tool knows what it has in front of it. Ask for the district heating in Main Street 2 and the tool finds the right meter itself.
The data you query is the data you have already collected
What makes the answers useful is the work you have already done. The data in Metry is collected, quality assured and structured according to your portfolio, and the meters’ names, addresses and energy types come along when the AI tool fetches the values.
If you are a customer today, you can start right away – nothing to order and no rollout to wait for! Follow our how-to guide for the AI tool you work in (you will find it under Integrations in the Metry portal), connect the tool and ask your first question.
If you’re not a customer, we start at the other end: get in touch and we will look at how your collection works today and what it takes to make it complete and quality assured.