The AI auto-filling system is designed to drastically accelerate your data collection process. It scans your uploaded ESG documentation to automatically populate answers, justifications, page references, and attachments, allowing you to bypass manual data entry entirely.
Understand the AI Processing Rules
To ensure strict data integrity across your workspace, the AI extraction engine follows a set of "Golden Rules" when interacting with your project.
Existing Data Protection: The system targets only questions that are currently empty. It will never overwrite an existing answer, whether that answer was input manually by a human or generated during a previous AI run.
Direct Input: There is no intermediate "suggestion box." When the AI finds a high-confidence match, it directly inputs the data and automatically updates the indicator's status to In Progress.
Passive Protocol for Missing Data: If the AI cannot find an answer within your documents, it remains silent. It will not auto-fill "Not Available" or spam your workspace with N/A statuses, ensuring your dashboard remains clean for human review.
Note: The user-facing Document Management interface is currently under active development. This feature is operating as a Greenly back-office tool and will be released to the client side by the end of Q2 2026.
Upload Your Corporate Documents
Before the AI can extract answers, you must feed the system with your relevant ESG documents, such as PDFs, spreadsheets, policies, and sustainability reports.
Navigate to the ESG Data menu located in the main left-hand sidebar.
Click on the Document Management tab to open your centralized library.
Verify that you are working within the correct entity and reporting period using the top navigation filters.
Click the Upload files button to select and import your documents.
Monitor the ESG Files list on your screen. The documents are successfully processed when their status label changes to Ready.
Launch the AI Analysis
Once your documents are uploaded and marked as ready, you must manually trigger the extraction engine to begin reading the text.
Locate the Run Analysis panel within the Document Management interface.
Click the Language drop-down menu and select the primary language of your uploaded documents.
Click the Start analysis button to initiate the scan.
Tip: You can run the analysis as many times as you like. If you acquire a new policy document halfway through your project, simply upload it and click Start analysis again. The AI will only attempt to answer the remaining empty questions using this new context.
Review and Validate the Populated Data
While the AI handles the heavy lifting, human oversight is a mandatory step in the compliance workflow. You must review the auto-filled forms to ensure complete accuracy.
Navigate back to the Data Collection tab to view your main list of indicators.
Locate the indicators that have automatically transitioned to the In Progress status.
Click on an indicator row to open its dedicated datapoint form.
Review the auto-populated data to verify the AI extracted the correct narrative text or numerical values.
Scroll down to verify the AI correctly populated the Justification text, attached the relevant source document, and noted the specific Page reference.
Click the Mark as completed button if the information is accurate, or manually edit the fields if corrections are required.
Frequently Asked Questions
How long does the analysis take?
The duration depends on your project's completion level. The very first run takes the longest, as the AI must analyze every single data point against all uploaded documents. Subsequent runs are significantly faster because the system instantly filters out already-filled answers before engaging the AI.
Can I run the analysis on one specific document only?
No. To ensure maximum accuracy and context, the AI considers the text content of all currently uploaded documents during every run.
Does the AI fill out complex tables?
Yes. The AI generates structured outputs and is capable of filling narrative text, Boolean toggles (Yes/No), and complex multi-row table structures depending on the specific indicator requirements.