Transparency report
What Klara costs the planet
Klara runs on AI, and AI uses energy. Here is exactly how much, measured, not guessed, and what we do to keep it small.
of CO2 for our whole company, over a full year
of CO2 per customer per year, everything included
of driving. That is your yearly share.
Yes, AI uses energy. Here is how much ours uses.
You might be wondering if using an AI assistant in your practice comes with a cost to the planet. The short answer is yes, but it is a small one, and we measure it.
Over the last 12 months, running all of Klarify produced about 0.72 tonnes of CO2. That is every server, every stored note, and every task Klara did for every therapist on the platform. It is less than one person’s round-trip flight from Toronto to London.
Split across our paying customers, your share is about half a kilogram of CO2 per year. That is roughly one mile of driving, or charging your phone every night for three months.
One more thing that matters here. We never train AI models on your notes or client data. Training is the most energy-hungry part of AI, and we add nothing to it. Klara only runs models that already exist.
Measured, not estimated
Most AI footprint pages are built from academic estimates. Ours is different. Klara runs on Amazon Web Services, and AWS reports the actual emissions tied to our account every month. The method is checked by an independent third party. We are sharing our real numbers, not a model of what they might be.
Three layers of AI energy
AI has three kinds of energy cost. It helps to know which ones we control.
Training the models
The AI models Klara uses were built once, by companies like Anthropic. Training them took a lot of energy, but that cost is shared across millions of people and thousands of products. We do not train models, and we never train on your notes or client data. We add nothing to this layer.
Building Klarify
Our team uses AI to help write and test the product itself. This is a one-time cost per feature. Once a feature ships, that energy is already spent.
Running Klara
Every note, letter, and reflection Klara drafts uses a small amount of energy. This is the ongoing cost, the part we measure every month, and the part we work to keep small.
What a year of Klarify costs the planet
These numbers cover our entire cloud footprint from June 2025 to May 2026. Not just Klara’s AI work, but our servers, databases, and every note stored on the platform.
Our total, before clean energy is counted
Our total after AWS’s clean energy purchases
Per customer, per year
What does half a kilogram look like?
Driving: about one mile in a gas car
Phone: charging it nightly for about three months
Kettle: boiling water for about 35 cups of tea
Streaming: around nine hours of video
Your yearly share of Klarify is smaller than one trip to the grocery store.
What these numbers include, and what they don’t
Everything is in there. These numbers cover all of Klarify, including Klara’s AI work, our servers, and storage.
Most of it is hardware, not electricity. Over 90% of our footprint comes from building the computers we run on, spread over their six-year life. The electricity to actually run Klarify for a year is a small slice, about 0.04 tonnes.
Your data lives in Montreal. Client data is stored in AWS’s Canadian region, where the grid runs mostly on hydro power.
It is still an estimate. AWS measures and allocates these numbers, and an independent third party checks the method. But no cloud footprint is exact, and reporting runs about two months behind. We will update this page as new data comes in.
How we keep it small
01
We use the smallest model that does the job
Not every task needs the most powerful AI. Klara uses lighter models for simple work and saves the big ones for tasks that need real depth.
02
We never do the same work twice
Once Klara drafts a note or letter, it is saved. Opening it again uses no new AI energy at all.
03
We never train on your data
This protects your clients’ privacy, and it also means your sessions never feed the most energy-hungry part of AI.
Small footprint, real responsibility
Using Klarify has a real cost to the planet. We think it is an honest and modest one, about one mile of driving per therapist per year. We are sharing it because you deserve to make informed choices about the tools you bring into your practice.
AI is also getting more efficient every year, so the energy behind each note keeps going down. We will keep measuring, keep improving, and keep this page up to date.
Questions about any of this? Email us. A real person will answer.
Sources and method
Emissions data: AWS Sustainability console (Customer Carbon Footprint Tool), our account, June 2025 to May 2026. Methodology verified by an independent third party. Link: https://sustainability.aboutamazon.com/aws-customer-carbon-footprint-methodology.pdf
The 0.72 tonne figure uses the location-based method, which does not count clean energy purchases. Counting AWS’s clean energy purchases, the figure is 0.68 tonnes.
Driving comparison assumes about 400 grams of CO2 per mile for an average gas car (US EPA).
Last updated: July 2026.