THE REGISTRY LAYER FOR UNSTRUCTURED DATA
PDFs, scans, spreadsheets, emails. Talonic reads them once and returns typed, schema-validated rows with evidence. Every workflow, system, and agent queries the same database from then on.
Response within 1 business day · Or start free: 5,000 credits/mo, no credit card
Ask the document estate
↳ ▋
| counterparty | value | renews | source · confidence |
|---|---|---|---|
| Northern Utilities | €1.24M | Sep 04 | p.12 · 0.96 |
| Municipal Heat Ltd | €640K | Sep 11 | p.08 · 0.94 |
| East Industrial Park | €2.08M | Sep 13 | p.21 · 0.97 |
3 rows · 0.38 s · every value traces to its source line
Illustrative query. On your corpus the answers come from your documents, extracted once and never re-read.
GETECPHOENIX PharmahandelBridgewayMaruti SuzukiWZBBosch Rexroth
live enterprise deployments · three continents
01 / production
contract_value · illustrative011M+
source observations, live tenant, 26 August 2026The migration needed 75 fields per contract dataset. One read of the estate holds 1,042,665 source observations and 52,937 reusable concepts, live tenant, 26 August 2026.
Read the case study →
supplier_name · illustrative021,130
contracts delivered, one row each, Ivalua import-readyStructured once, delivered in the shape Ivalua imports. No re-extraction between systems.
Read the case study →
lane_rate · illustrative0396% vs 78%
same documents, same labels, same scoringThe same hand labelled documents, two fields, both systems scored the same way. Replacing a six figure incumbent.
Read the case study →Automotive, research, and industrial operations. Different document estates, the same governed registry.
metric_value · illustrativeNineteen months of head office reporting from India’s largest automaker, resolved across 24 verticals into one queryable form with provenance to the page.
Read the case study →
hours · illustrativeGerman school timetable law from 1996 to 2021, transcribed exactly as printed. Every row carries its source page, a confidence score, and a note on anything ambiguous.
Read the case study →
tolerance_band · illustrativePerformance data existed only as printed curves, read into numbers. Every curve sampled to 100 points, with the reading conventions written onto the sheet.
Read the case study →529 document types, zero templates · every number on this page names its measurement
02 / the false economy
Every RAG answer re-reads the same corpus and re-spends the same tokens. The cost scales with documents × systems × queries, and the answers still change between runs. A database is read once and queried forever.
Every question re-extracts the same contracts, re-spends the same tokens, and can return a different answer than yesterday.
When the answer feeds an ERP entry or a filing, “the vector was close” isn’t provenance. Every Talonic value traces to the exact line, scan region, confidence, and reasoning.
Talonic resolved 529 document types with zero templates. Your documents already define the schema. The Schema Graph finds it.
03 / one pass
Any source: file shares, email, scans, RPA feeds, agent outputs. 25+ formats, four languages at production quality.
PDFs · scans · handwriting · spreadsheets · emailsThe Schema Graph finds the fields, entities, and relationships on its own. Canonical fields compound across every document you add; humans review only low-confidence values.
vendor_name · contract_value · governing_lawSQL, natural language, REST, or straight into your agents over MCP. Ask once, and the answer comes from the registry, not a re-read of the corpus.
WHERE auto_renew = true AND notice_period_d < 30
04 / put it to work
A Spec describes exactly what a downstream system needs: the fields, the shape, the format. Pipelines deliver it automatically into Dynamics, Ivalua, Salesforce, or any REST endpoint. Every delivery is signed, retried, and audit-logged. The workflows you used to script with RPA run on AI instead.
Everything above is available over MCP. An agent can create the Spec, wire the Pipeline, and watch the deliveries. Your team describes the outcome and the agent configures Talonic.
npx -y @talonic/mcp@latest05 / receipts
Same corpus. Side by side. Run it yourself.
Same 53 SEC filings, same model, 39 questions that need every document at once. Talonic answered 23 of 24 ranking questions. Hybrid RAG answered 3.
Read the benchmark →Retrieval pays on every question, forever. Structuring pays once: $10.14 to read 53 documents, then $0.00 per 1,000 questions on the structured path.
Read the benchmark →Ask the same question ten times. Hybrid RAG gives the same answer all ten times on 2 of 39 questions. Talonic does it on all 39, byte for byte, across 390 turns.
Read the benchmark →REST API, Node SDK, or MCP server. Every response carries typed fields, confidence, and provenance, so your agent knows when not to trust an answer. Free tier: 5,000 credits a month, no credit card.
06 / your documents
Send a representative sample of contracts, scans, case files, or operational documents. We run them through Talonic and send back your extracted data within five business days: field-level coverage, confidence, and provenance. If the database doesn’t beat what you have, you’ll know exactly why, field by field.
Every company we walk into is sitting on the same two things: decades of documents nobody can query, and an AI budget being spent re-reading them. We think that’s backwards. Your documents already contain a database. Someone just has to find it once, prove every value, and keep it current. That’s the entire company.
So we won’t pitch you with adjectives. Send us your ugliest pile of files: the scans, the handwriting, the folder nobody opens. Then judge the output field by field.
— Nikolas Adamopoulos & Holger Nordsiek · Founders, TalonicPhotography: A.Savin (FAL) · U.S. Air Force, public domain · Marcin Wichary (CC BY 2.0) · Prime Minister’s Office, India (GODL) · Coenen (CC BY-SA 3.0) · Kleuske (CC BY-SA 3.0)