Case study · Climate tech · AI module

Upload a bill. AI does the data entry.

The AI Document Manager is where Bharat Carbon users upload bills and compliance documents. AI reads them, fills in the numbers for carbon accounting and ESG reports, and a checker signs them off.

Book a free call See the screens

Client: Bharat Carbon
Industry: Climate tech · Carbon accounting & ESG
Product: AI module in the Bharat Carbon web app
Timeframe: 2024
My role: UX + UI design, interaction design, prototyping
Tools: Figma, FigJam

01 · The short version

The project in a minute

The problem

Data entry from paper

Carbon accounting runs on bills and filings. Typing twelve months of electricity readings by hand is slow and easy to get wrong.

What I did

Upload once, check, save

A two‑step upload drawer, an AI view that shows the bill next to the numbers it found, and a review flow with clear statuses.

The result

Check, don't type

Users review numbers instead of typing them, and every figure can be traced back to the document it came from.

2steps to upload: add details, then the file

12months of readings pulled from a single bill

3review states: in review, approved, rejected

2products fed: Carbon Accounting and ESG Reporting

02 · Who it's for

Built for teams who report emissions

Sustainability, finance and facility teams collect the bills. Admins and checkers make sure the numbers are right before they reach a report.

Primary users

Sustainability teams · Facility managers · Checkers

“Let me drop in the bill and just check the numbers.”The brief in one line, as I framed it

Less typingAI reads the bill and fills the table.
Numbers I can trustSee where on the bill each value came from.
Stay in controlEdit any row before saving.
Know what's pendingIn review, approved or rejected, at a glance.
One place for every documentBills for carbon accounting, filings like AOC‑4 for ESG.

03 · How it came together

Map, design, review

The hard part was trust: AI can read a bill, but people have to believe the numbers.

1
Step 1
Map the flow
Which documents feed which product: utility bills for carbon accounting, filings like AOC‑4 and Form CRA‑2 for ESG and BRSR.
- Flows
- IA
2
Step 2
Design the upload
A two‑step drawer with drag and drop, file preview, upload progress and an AI processing state.
- Interaction design
- States
3
Step 3
Design the review
The side‑by‑side extraction view, row‑by‑row edits, assigning a checker, and statuses in the document list.
- Visual design
- Prototype

The four steps of the upload flow — The upload flow: add details, upload the bill, let AI read it while you check the values, then assign a checker.

04 · Key decisions

The choices that made it work

Decision 01

Show the bill next to the numbers

Why: People only trust AI output they can check against the source.
What I did: The bill stays on the left with the source highlighted. Month, kWh and cost sit on the right, and every row can be edited.

AI extraction view with the bill and the data table

Decision 02

Ask for context before the file

Why: The same upload means different things for carbon accounting and for ESG.
What I did: Step one asks for product, year, facility and bill type (or framework and document type), so AI knows what to look for.

Decision 03

Make every status visible

Why: Documents pass between makers and checkers. Nobody should wonder what's stuck.
What I did: The list shows in review, approved and rejected, with comments like “Re‑upload document” and the fix one click away.