How I’m building an environmental impact application that connects everyday actions, scientific evidence, and collective change.
The problem: We act, but do we understand the impact?
Every day, people make small environmental choices. Someone recycles a plastic bottle, saves water while brushing their teeth, switches off unnecessary lights, or chooses public transport instead of a private vehicle.
These actions matter. But there is a gap between taking an action and understanding its environmental impact.
How much waste did I divert? How much water did I conserve? What does saving one kilowatt-hour of electricity mean for emissions? And what happens when thousands of people make similar choices?
These questions inspired me to build ONE EFFECT, an environmental impact application built around one simple idea:
Your action is small. Its impact isn't.
My goal is to make environmental contributions easier to record, understand, and measure without relying on arbitrary sustainability scores or unsupported claims.
Introducing ONE EFFECT
ONE EFFECT brings several environmental dimensions into one application:
- Waste: Recycling and composting.
- Water: Recording water conservation in litres.
- Energy: Measuring electricity savings in kilowatt-hours.
- Transport: Tracking distance and lower-emission travel choices.
- Nature: Recording trees planted, trees maintained, and areas restored.
The application is designed to help users move beyond simply collecting points. Instead, it focuses on understanding the different kinds of environmental contributions they make.
A user can explore available actions, provide relevant measurements, upload an image as evidence, and view the intended impact of the contribution. Personal history and community views help connect individual activity to a broader environmental picture.
Why image analysis is only one part of the solution
One of the central design decisions in ONE EFFECT is separating image analysis from environmental impact calculation.
Imagine a user selects the action “Recycle a plastic bottle” and uploads a photograph.
An image-analysis model may identify the object as a plastic bottle and estimate its likely material. However, recognizing a bottle does not prove that it was recycled, establish its weight, or determine the environmental benefit of the recycling process.
Those are different questions.
My intended workflow separates them into three stages:
1. Identify: Analyze the image to identify the object or material.
2. Validate: Check whether the available evidence is consistent with the selected action. Image recognition alone cannot establish that an action was completed.
3. Calculate: Use the relevant measurement and documented environmental factors to estimate the applicable impact.
This separation is important because an AI model should not invent scientific numbers. AI can help interpret evidence, while a deterministic calculation engine should handle the arithmetic.
Environmental impact needs evidence
A sustainability application can easily become misleading if it converts every action into one generic score.
For ONE EFFECT, I chose a different approach: use documented factors, retain their units, and explain what the resulting numbers actually mean.
For recycling, I researched the US Environmental Protection Agency's Waste Reduction Model, known as WARM. Its lifecycle-based factors can be used to estimate greenhouse-gas and energy impacts for specific materials under defined assumptions.
For electricity, I investigated India's electricity emissions data published through the Central Electricity Authority. A grid emissions factor can help estimate the emissions associated with a given quantity of electricity saved, provided the factor's year, geographic scope, and methodology are documented.
For water conservation, litres saved are a direct and understandable primary measure. I do not automatically convert those litres into carbon savings without an appropriate water-energy emissions factor.
For trees, I focus on recording trees planted or maintained and the area restored. Carbon sequestration depends on factors such as species, age, growth, location, and survival, so a universal carbon value per tree would be misleading.
These distinctions shape the methodology page in ONE EFFECT.
Each environmental factor should have a source, unit, applicable geography, reporting year, and version. Calculations should retain the factor used at the time so historical contributions can be interpreted correctly.
Modeled estimates should also be labelled as estimates rather than direct measurements.
From personal actions to collective impact
An individual contribution can feel insignificant when viewed in isolation. A community can tell a different story.
ONE EFFECT includes a community-impact concept designed to aggregate contributions across categories such as waste, water, energy, transport, and nature.
The goal is not to rank people or turn environmental action into a competition. It is to help users understand how individual efforts can contribute to collective progress.
For this to be meaningful, community totals must eventually come from actual stored contributions, with appropriate safeguards for privacy. Sample figures and simulated projections must remain distinguishable from real activity.
The same principle applies to the What If simulator: a projection can help users explore possible outcomes, but it should never be presented as a record of something that has already happened.
Where AWS fits into the architecture
Building ONE EFFECT also gives me an opportunity to explore how AWS services can support a modular environmental application.
My proposed architecture separates several responsibilities:
- Amazon Bedrock: Image analysis to help identify objects or materials in uploaded evidence.
- Amazon S3: Storage for uploaded images.
- AWS Lambda: Backend processing and coordination of application workflows.
- Amazon DynamoDB: Persistence for contribution records, measurements, timestamps, and impact-calculation metadata.
This separation allows image analysis, scientific calculations, and contribution storage to evolve independently.
For example, changing an image-analysis model should not require rewriting the environmental calculation engine. Likewise, adding a new environmental action should ideally involve defining its required measurements and supported impact factors rather than duplicating calculation logic across the application.
These services represent the architecture I am working toward. They should be considered implemented only after the corresponding integrations have been built and tested.
Challenges and lessons learned
One of the biggest lessons from this project is that a polished interface is not the same as a reliable product.
A dashboard can display attractive numbers, but those numbers are only meaningful when they come from traceable records and defensible calculations.
That means moving away from hardcoded counters, saving contributions through a real data layer, making action-specific inputs dynamic, and ensuring that every displayed metric is derived from the appropriate data.
Another lesson is that environmental calculations require careful communication. A number without a source, unit, or caveat can create a false impression of precision.
I am treating transparency and scientific restraint as product requirements, not features to add at the end.
What's next?
My next priorities for ONE EFFECT are:
- Connect the image-analysis workflow to a real AI service.
- Persist individual contributions rather than relying on fixed dashboard data.
- Make environmental actions and their measurement requirements configuration-driven.
- Calculate dashboard and community totals from recorded contributions.
- Validate environmental factors and make their sources accessible to users.
- Deploy and test the application with the appropriate AWS services.
The goal is to move from a promising prototype toward a working system that can explain not only what a user did, but also how its environmental impact was estimated.
Final thoughts
ONE EFFECT began with a simple question: what if people could see the significance of their everyday environmental choices?
Building it has shown me that answering this question requires more than a dashboard or an AI model. It requires reliable data, appropriate scientific factors, careful software architecture, and honest communication about uncertainty.
Environmental progress is collective, but understanding can begin with one action.
ONE EFFECT — Your action is small. Its impact isn't.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article