At Ecologi, we use AI carefully, deliberately, and only where it helps us deliver a better experience for customers.
Ecologi Zero handles business-critical carbon accounting data, including source documents, consumption data, financial information, and other commercially sensitive records. We know customers place significant trust in us when they use the platform. Our approach to AI is designed to protect that trust.
We use AI to support specific workflows, not to replace our carbon accounting methodology, human oversight, or responsibility for data security.
Our approach
Our use of AI in Zero is guided by the same principles that shape how we build the platform.
Fundamentally, we don’t risk breaking customer trust. Protecting customer data, preserving data integrity, and maintaining confidence in the numbers that Zero produces are of the utmost importance to us. This is a foundation of integrity that we won’t compromise.
We favour simple, bounded uses of AI. We don’t give AI systems broad or unnecessary access to customer data. Where AI is used, we aim to keep the purpose clear, the data inputs limited, and the outputs reviewable.
We test and measure before expanding AI use. We do not assume that an AI feature is useful, accurate, or safe simply because it is technically possible. We assess whether it improves the customer experience and whether its outputs are good enough for the purpose they serve.
We take considered risks to make progress. AI can help us move faster and build better tools, but we introduce it in controlled ways, with clear safeguards and a preference for small, measurable improvements.
Where AI is used in Zero
Utility bill processing
Customers can upload utility bill PDFs to Zero to save time during the data collection process.
Uploaded documents are stored securely in our platform infrastructure and linked to the relevant customer profile. They are not publicly accessible and are only available to authorised users.
Zero can process the file to extract structured data. This helps turn uploaded PDF files into fields that can be reviewed and mapped, such as:
emissions source
consumption amount
unit
start and end dates
description
category
tags
This process extracts data for customer review, and is applied to a customer’s footprint using carbon accounting methodologies that have been agreed and confirmed with our Specialists. All data goes through a rigorous validation process by our Carbon Accountants as part of report generation.
We do not use AI in Zero to:
train public or general-purpose AI models on customer data
sell customer data
How we protect customer data
AI features sit within the wider security controls of the Zero platform. These include authenticated access, role-based permissions, profile-level data separation, and controlled access to uploaded documents. Customer data is not stored within the platform unnecessarily, and we regularly check to ensure data no longer being used is deleted.
Where a third-party AI service is used, we choose enterprise-grade services with appropriate privacy and security controls. Customer inputs and outputs processed through Amazon Bedrock are not used to train base foundation models.
We also aim to minimise the data sent to AI systems. Where an AI feature only needs summarised emissions context, we send summarised context rather than raw source documents.
Accuracy and human oversight
AI can be useful, but it can also make mistakes. We design AI-supported features so outputs can be checked, corrected, or ignored.
Ongoing review
AI technology is changing quickly. We review how we use it, whether it is delivering value, and whether our safeguards remain appropriate.
Questions
If you have questions about how Ecologi uses AI in Zero, or how your data is protected, please contact us at zero@ecologi.com, or via your Customer Support team.