Document intelligence

Every number, traced back to exactly where it came from.

BuiltQuery reads your documents — PDFs, Word, Excel — and returns clean, structured data against schemas you define. Every value stays linked to its source, so your team verifies with a click instead of re-reading the document.

Any document typeAny schema you defineSource-verified outputAPI-first & SharePoint-native

Request a demo Create a free account

Source-verified

See it work

Click a value. Land on its source.

This is what sets BuiltQuery apart. Extraction alone is easy — trusting it is the hard part. Every value BuiltQuery returns keeps a link to the exact place it came from. Pick a field on the right and watch it point straight back into the document.

Cost Plan — Rev C.pdfpage 4

Area schedule

Ground floor — NIA1,940 m²

Level 1 — NIA2,110 m²

Total GIA12,480 m²

Structured outputJSON

Every value carries this same link — in the interface and over the API.

How it works

One pipeline, from raw file to trusted data

The same six stages run whether a person drags in a single file or a SharePoint site feeds thousands of documents automatically.

Stage 1

Connect & ingest

Upload directly, or connect SharePoint so documents flow in on their own.

Rev C.pdf

Stage 2

Auto-classify

Labelled against a tag vocabulary you define, with a confidence score.

Cost Plan · 0.96

Stage 3

Extract

One or many templates run in a single pass — parsed only once.

{ 24 fields }

Stage 4

Ground

Every value pinned to its page, coordinates or spreadsheet cell.

page 4 · C18

Stage 5

Validate

Reconciled against your master data and schema rules; issues flagged.

matched ✓

Stage 6

Deliver

Lands in the review queue and is available over the partner API.

GET /v1 · 200

Capabilities

What makes BuiltQuery different

Five capabilities that hold together whether you extract one document by hand or migrate an entire archive.

Beta

Dynamic templates

Any document, any schema

You describe the fields you want as a JSON schema and a prompt — BuiltQuery does the rest. No fixed forms, no per-document setup.

Four ready-made templates (project, areas, cost plan, programme) plus unlimited templates of your own.
Quick-select from 25+ common metrics, or define custom fields with type and description.
Run many templates in one job — the document is parsed once and reused, so more data costs almost no extra time.
Works across PDF, Word and Excel, including cell-level reading of spreadsheets.

Beta

Provenance & review

Data you can verify, not just trust

Every extracted value carries the exact place it came from, so a reviewer confirms in seconds instead of hunting through the source.

Click-to-source highlighting — page, coordinates, and text snippet for documents; sheet and cell reference for spreadsheets.
A confidence score on every value guides where humans should look first.
A full review workflow: approve, correct fields, add notes, and compare or revert a re-run.
Schema checks flag type, format and missing-field issues as warnings — never silently dropped.

Beta

Partner API

Built for developers

Everything the interface does is available programmatically, so BuiltQuery drops into the systems your team already runs.

Organisation-scoped API keys, self-managed in settings, with strict tenant isolation.
Submit a document and template set, then poll for status — fully asynchronous.
Results return the structured data and the full source map — the same provenance the UI shows.
For connected SharePoint drives, submit by reference — no file bytes are ever copied or stored.

Beta

Master-data validation

Connected to your systems

Extracted values don't sit in isolation — they resolve against your own records, so the output speaks your organisation's language.

Link template fields to your live systems with secure, per-organisation credentials.
Placeholders like %Project% resolve to a real record and ID, with AI disambiguation from context.
Validation screens show each value matched to your master data, ready to confirm or correct.
Nothing commits until a human signs off against systems you control.

Beta

Bulk & always-on

From one document to an entire archive

The same engine powers a one-off review and a standing pipeline that keeps a document library structured with no one watching.

Bulk migration — point BuiltQuery at an existing corpus, filter by tag and location, and fan out extraction across every matching document.
Always-on rules — connect a SharePoint site and any new document is auto-classified, then the extractions you've mapped to that classification run on their own.
Delta sync & change notifications keep the index current as files are added, edited or removed — without copying them.
Low-confidence results route to a “needs review” queue so people spend time only where the machine is unsure.

Developer experience

Provenance, over HTTP

Submit by reference, poll for the result, and receive both the data and a source map for every field — ready to drop straight into your own review tooling.

# 1 — submit a SharePoint document against your templates
POST /api/v1/extractions
Authorization: Bearer bq_live_…
{ "driveId": "b!x…", "itemId": "01ABC…",
  "templates": ["leasing-hot"] }
→ 202  { jobId, status, statusUrl }

# 2 — fetch the grounded result when complete
GET /api/v1/extractions/{jobId}?include_result=true
{ "results": { "leasing-hot": {
    "data":      { "totalGIA": 12480 },
    "sourceMap": { "totalGIA": [{ "page": 4,
      "cell_range": "C18", "confidence": 0.98 }] }
} } }

At scale

A library that structures itself

Define the rules once. From then on, documents arriving in SharePoint are classified and extracted automatically — the loop closes without anyone lifting a finger.

Integrates with

SharePointMicrosoft 365

Document arrives

A file lands in a connected SharePoint site; a change notification indexes it by reference.

Auto-classified

It's tagged against your vocabulary. High confidence tags it outright; the rest go to review.

Rules match

Its tags and location are checked against your standing extraction rules.

Extraction fires

The templates you mapped to that classification run automatically and land grounded.

Continuous — every new or changed document runs the loop again

Honest status

In beta, and what's next

Everything above is currently in beta and ready for prospective customers. A few more capabilities are coming next — some in active development, others planned — flagged here so the picture is complete.

Currently in beta — ready for prospective customers In development Planned — on the roadmap

In development

Enrichment system

Automatically enrich extractions with data from third-party sources — so each record arrives not just extracted, but augmented with context from beyond the document itself.

Planned

More connectors

SharePoint is the connector available today. Google Drive and Dropbox use the same ingestion foundations and are next in line.

Planned

OpenAPI import

Point BuiltQuery at an OpenAPI spec to wire up master-data sources automatically, instead of configuring each by hand.

Planned

Custom org endpoints

Organisation-specific API endpoints, so bespoke workflows can be exposed cleanly per customer.

Also in the platform

Beyond extraction

Chat with documents

Ask questions across one or many files, side by side, with retrieval over your library.

Interactive data grids

Structured results in editable grids with automatic unit standardisation.

Extraction Test Studio

Measure template accuracy and catch regressions before they reach production.

Multi-tenant controls

Feature-based access, roles and usage metering per organisation.