Our public carbon & water statement
A smaller footprint, honestly measured
Medicine in Practice is digital by design, but digital does not mean impact-free. We publish what we have measured, distinguish estimates from facts, include water as well as carbon, reduce first and review this plan at least annually.
Net zero by 2050 or earlier
Reviewed every year
Ranges published, not hidden
AI water use acknowledged
Straight about the numbers
Our footprint, honestly stated
Assessed August 2026 using published conversion factors and stated assumptions. These are estimates, not audited measurements, and the ranges show where uncertainty remains. Water is separated because a defensible service-specific total is not yet available.
Commitment
Our net zero commitment
We commit to reaching net zero greenhouse gas emissions by 2050 or earlier, in line with the expectations the NHS places on its suppliers. In practice we expect to reach it far sooner: our footprint is small, digital and largely within our control.
We will measure our emissions annually, reduce them first, and compensate for the small residual using high-quality, verifiable UK carbon removal, such as Woodland Carbon Code units—not cheap unverifiable credits.
Reduction first
How we keep it down
Six practical choices shape how the service is built and delivered.
Digital by design
No training venues, printed packs or delivery travel. Learners practise where they already work.
UK hosting
The service runs in the AWS London region; Amazon reports matching its electricity use with renewable-energy purchases.
Remote-first
We operate without commuting or business flights by default.
Efficient engineering
A lightweight, aggressively cached platform reduces compute and data transfer per learner.
Longer-life hardware
Devices stay in service for as long as practical and are replaced refurbished where possible.
Water-aware AI
Live AI is optional, metered and bounded. Reviewed scripted scenarios are used whenever they can deliver the learning safely without a model call.
The footprint people cannot see
AI also uses water
Live AI work creates heat. Data centres may use water directly for cooling, and electricity generation can have an indirect water footprint too. The amount varies materially by data-centre location, weather, cooling design, power source and model, so a generic “litres per prompt” figure would be misleading.
AI requestcompute is performed
Heat & powerservers and facilities
Water impactdirect cooling + electricity
Our hosting withdrawals
AWS now makes attributable water withdrawals available by account, service and region. We will capture the London-region figure alongside our annual carbon review.
Water per AI patient call
Medicine in Practice does not currently receive metered water use or inference-location data from Anthropic for each response. We will not manufacture a litres-per-call claim without those inputs.
Unnecessary model calls
The service records authorised AI calls and turns, caps each patient reply and conversation, and falls back to reviewed scripted content. AI is used where variation adds learning value, not as decoration.
Our 2026 action: pair the AI call-and-turn counts already visible in the owner dashboard with AWS water-withdrawal reporting; request usable water data from our AI provider; then publish a service intensity figure only when its boundary and assumptions can be explained.
The bigger number
The travel our training can avoid
Using UK government conversion factors (about 0.17 kg CO2e per km for an average petrol car), a typical 50-mile round trip to an external study day emits roughly 14 kg CO2e per learner.
A 30-learner cohort that practises in the simulator instead of attending one external study day avoids around 400 kg CO2e of travel alone, before venue energy, catering and printing.
30 learners×one study day=about 400 kg CO2e avoided
Every garden needs a tea break
Gail has claimed the best seat
Carbon plans can become a page of numbers, so we have given ours a slightly less serious corner: a bright garden, a reusable mug and somewhere to put your feet up after doing the watering.
Gail's Seat: available between gardening jobs, subject to mug ownership.
For commissioners
For NHS and university commissioners
This page constitutes our public net zero commitment and carbon-and-water reduction plan, aligned with the requirements of the NHS net zero supplier roadmap for contracts of our size. We are happy to complete buyer-specific sustainability documentation, including the Evergreen Sustainable Supplier Assessment, as part of contracting.
It sits within our procurement evidence register alongside our data protection, security and accessibility documentation.
How the estimates were made
Carbon estimates use the UK Government greenhouse gas conversion factors, published cloud-provider sustainability reporting, and published research ranges for AI inference energy—the largest source of uncertainty, and stated as a range for that reason.
For water, we distinguish withdrawal (water drawn) from consumption (water not returned for reuse). Our reporting basis follows the AWS water system boundary for hosting and the primary research paper Making AI Less “Thirsty” for the direct-cooling and indirect-electricity distinction. AWS account and regional withdrawal data are described in its Sustainability Console.
Free tools we recommend to other small businesses starting out: the SME Climate Hub and the UK Business Climate Hub.
Questions or challenge on any figure: [email protected]. We would rather correct a number than defend it.