Extracting data from utility bills
A utility bills every household and every site it serves, monthly, on a layout of its own design. Banks take them as proof of address and energy managers take them for consumption data, so one document serves two unrelated jobs. Both handle thousands a month, as portal downloads and phone photographs.
A verifier needs the account holder, the service address and the bill date; an energy team needs the meter readings, the usage and the charge totals. KYC onboarding checks the address against the application, and an energy management platform tracks cost per unit across sites.
What makes utility bills hard
One usage figure, several rates
A month's kWh is priced in blocks: the first 103 at one rate, the next 135 at another, each its own line.
Two services, two meters
Electric and gas share one bill, each with its own meter, readings, usage, charge blocks and total to reconcile.
A year of usage is drawn
The consumption history is a bar chart with a temperature line over it, holding a year of daily averages.
Parse, extract, verify
Parse
Parse turns the document into reading-order Markdown and a hierarchy of blocks (text, tables, figures), and grounds each block to its exact location on the page.
Extract
You define the fields in a JSON schema, and Extract returns their values in a consistent, structured format, each linked back to where it appears in the Markdown.
Verify
Each value links back to where it appears, down to the character range in the Markdown, and the page and bounding box in the source document, so every result can be verified.
One utility bill, end to end
About this document: A PSE&G residential bill covering a single billing period. Four portrait pages: a summary with a payment stub, a charges overview, then a detail page each for gas and electric, every detail page carrying a usage chart. This is a real customer's bill. The name, address, account, invoice, meter and point-of-delivery numbers were replaced with plausible stand-ins before publication.
Processing objective: The objective is to reconcile one service's charges against its meter. Page 4 bills 238 kWh, the difference between two readings, through a service charge and two rate tiers. Delivery of $16.71 plus supply of $29.71 makes the $46.42 total.
Demonstration fields: Account number, Meter number, Delivery charges, Service charges, Total electric charges
The document
4 of 4 pages
Page 1
Page 2
Page 3
Page 4shown below
Where each value came from
account_info.account_number
{
"value": "6873630651",
"page": 4,
"range": {
"end": 10202,
"start": 10171
},
"box": {
"xmin": 0.68822,
"ymin": 0.04242,
"xmax": 0.91347,
"ymax": 0.05355
}
}electric_charges.meter_number
{
"value": "245603755",
"page": 4,
"range": {
"end": 10344,
"start": 10327
},
"box": {
"xmin": 0.08519,
"ymin": 0.13796,
"xmax": 0.38551,
"ymax": 0.1558
}
}electric_charges.total_delivery_charges
{
"value": "$16.71",
"page": 4,
"range": {
"end": 10969,
"start": 10963
},
"box": {
"xmin": 0.472,
"ymin": 0.30622,
"xmax": 0.56866,
"ymax": 0.32274
}
}electric_charges.monthly_service_charge
{
"value": "$4.95",
"page": 4,
"range": {
"end": 10687,
"start": 10682
},
"box": {
"xmin": 0.47196,
"ymin": 0.23462,
"xmax": 0.5686,
"ymax": 0.25159
}
}electric_charges.total_electric_charges
{
"value": "$46.42",
"page": 4,
"range": {
"end": 11587,
"start": 11330
},
"box": {
"xmin": 0.0814,
"ymin": 0.4216,
"xmax": 0.61031,
"ymax": 0.68859
}
}How to try it yourself
Everything on this page is reproducible. The sample, its schema, and the Parse and Extract API outputs are published. You can repeat the same two calls from whichever surface you already build on.
- Claude Skilldescribe the job; the agent writes it
- CLIno code at all
- MCPexpose ADE docs as an agent tool
- REST APIany language
- Pythonlandingai-ade
- TypeScriptlandingai-ade
Documents into data
Accurate
The best accuracy on the difficult documents other tools get wrong, such as messy scans, complicated tables, checkboxes and handwriting.
Auditable
Every answer cites its sources. Values come back with an audit trail (page, bounding box, character range) so a human can review them.
Agentic
Applies multi-step reasoning, and plans before it acts. Handles documents it has never seen before and adapts to variations.
API-first
Composable APIs that drop into any tech stack and can be called from anywhere. Easy to build on and embed in your pipelines and products.
Affordable
Cost-competitive rates with complete price transparency. Spend controls before you run, and every job reports its exact cost.
Enterprise-grade
Built for production and scale from day one. Fully observable and monitored, running securely in our cloud or yours.
Built for high-volume, high-throughput production
- Zero data retention
- Priority rate limits
- Enhanced support
- VPC and on-prem deployment
Available on Team and Enterprise plans.
