DiginexESG: using AI to make sustainability reporting doable for a small team.
Sustainability reporting means answering hundreds of detailed questions and turning them into a polished report that an auditor will trust. DiginexESG uses AI to do the heavy lifting. I led the design on this one: I worked with the business team to shape the ideas into flows, wrote the user stories for the developers, and owned the design system across the whole product.
AI-assisted disclosures
AI drafts, people check.
Report data selection
Pick what goes in the report.
AI report builder
Shape the report, step by step.
Document Center
Every document in one place.
Chat with AI
Ask the documents a question.
AI-assisted disclosures
Each reporting question sits beside its answer editor and supporting documents. AI prepares a draft, guidance and comments stay alongside, and a completion summary reminds users to review AI answers before moving on.
Sustainability reporting is a huge amount of work, and the teams doing it are usually small. That gap is the whole problem.
A report means answering hundreds of questions (called disclosures) and writing them up in a way an auditor accepts. DiginexESG uses AI to speed this up. The PM brought the ideas and requirements; I worked with the business team to turn them into flows and user stories, and designed the three features below.
The hard part isn't the writing. It's finding the right evidence in a stack of company documents, and making sure every AI draft is checked before it goes into a report. So the design keeps people in control: AI prepares answers and summaries, and the screens put the question, the draft and the supporting documents side by side for review.
What I owned.
This was the most complex product I've worked on, and I led the design across all of it. The PM brought the ideas; I owned turning them into something real:
The analysis
I sat with the business team to pressure-test the ideas and agree on what we were actually building.
The flows & user stories
I turned each agreed direction into flows and into user stories the developers could pick up directly.
The design system
I owned the design system across the whole product, which kept a genuinely complex tool consistent and faster to build.
The thread running through every feature was the same: let AI do the slow, heavy work, but keep a person in control of anything that ends up in a published, audited report. Getting that balance right, in the design and in the stories, was the core of the job.
From company documents
to report drafts
These three examples show how I organised the reporting process and placed AI assistance alongside the information users need to review.
Drafting and reviewing disclosure answers
Teams need to answer detailed reporting questions, called disclosures, using information from company documents. AI can prepare drafts, but users still need to check the answers and evidence.
- 01Placed the question, answer editor, and supporting documents together so users can review them in context.
- 02Kept guidance and comments beside the answer, with actions to request data or upload evidence.
- 03Showed completion by category in the disclosure navigation.
- 04Added a completion summary with a reminder to review AI-generated answers before continuing.
Choosing report data and settings
Users need to decide what information belongs in a report and how to present it for different audiences.
- 01Started the report flow with time frame and data selection before moving into report configuration.
- 02Placed reporting frameworks beside a searchable metrics table, with codes, source reports, and available data points.
- 03Grouped audience, purpose, scope, and company branding within the configuration step.
- 04Used sliders for tone, detail, and length, with helpful tips beside the settings.
- 05Kept the report stages visible at the top, from data selection through to preview and export.
Finding information in company documents
Teams need a place to manage their reporting documents and a way to explore their contents without reading every file from start to finish.
- 01Created a Document Center with upload, search, file categories, and uploader details.
- 02Made the document name visible above each AI conversation so users know which file they are discussing.
- 03Added follow-up actions for a detailed breakdown, a comparison with the previous year, or a custom question.
- 04Kept past conversations in a dated sidebar so users can return to earlier discussions.
From documents
to a report.
How a reporting team turns company documents into a finished report, with AI drafting the answers and people checking them before anything goes out.
Flow for: Reporting teamUpload company documents
StartFiles with categories, uploader details and dates in one list.
Document CenterAsk the documents a question
The file name stays above the chat, with follow-ups ready to pick.
Chat with AIAnswer the disclosures
AI drafts each answer beside the question and its evidence.
DisclosuresHas every AI answer been reviewed?
The completion summary asks for a review before moving on
The completion summary asks for a review before moving on
Choose the report data
Time frame first, then frameworks and metrics to include.
Report data selectionSet up the report
Audience, scope, branding, and sliders for tone, detail and length.
AI report builderPreview and export
DoneThe stages stay visible at the top until the report is out.
Preview and exportHKMA award, 2023 (ESG disclosure)
From hundreds of manual disclosures to AI drafts a small team can trust.
- Dev Sinha · Product Manager
- Abraham Joseph · Business Analyst