PRIVATE BETAJoin the waitlist →
QuantyQuantyBETA
PRODUCT

A spreadsheet that does the reading.

Columns are prompts, cells carry proof, and email keeps the sheet fed. This is how it fits together.

HOW IT WORKS

Just a spreadsheet. Until you add a column.

This is the actual grid: rows are documents, columns are properties you define, everything one click deep.

Home / Invoice matcherEmail-connected · watching +invoices@
Chat with AIExport
+ Add row+ Add columnFilter rows…
1,204 rows · 8 columnsCompletedRecompute stale (12)Review3PipelineRun
File
T Counterparty AI
# Amount AI
Invoice date AI
Payment terms AI
Edit column
Recompute all cells
Duplicate column
Compare models (A/B)
Pin column
Hide column
Delete column
# Δ vs PO PY
+
1
Invoice_88101.pdf
Semtech Corp.
$12,400.00
2026-07-02
Net 30 p.2
0.00
2
CloudProvider_July.pdf
CloudProvider
$3,180.50
2026-07-04
Net 60 p.4stale
+120.50
3
Freight_inv_0921.pdf
Nordhaul AS
€1,890.00
2026-07-05
Net 14edited
0.00
4
Vector_amendment.pdf
Vector GmbH
$18,220.00
2026-07-08
processing
5
Scan_receipt_003.pdf
Baltika Sp. z o.o.
zł4,520.00
2026-07-09
Net 30 p.1
0.00
Page 1Q3 audit+ Add Page
or email +invoices@
1
Every row is a documentDrop in PDFs and scans, or let email feed the sheet. Rows that arrived on their own carry a small bolt.
2
Every column is a propertyPick a type: text, number, date, select, JSON. Write a prompt in plain language and AI, Python or Web fills it for every row.
3
Click a column to work itRecompute stale cells, compare models A/B, pin or hide. Uncertain values queue for human review instead of writing themselves.
WHO IT'S FOR

Hours of reading. One clear picture.

Wherever documents pile up, the pattern is the same: Quanty does the reading, you present the conclusion.

Explore all industries →
Logistics45× faster
1$1,890matches rate card
2$2,340over rate card
3$980matches rate card
4$1,120no shipment
Freight invoices matched to shipments
Invoices, delivery notes and rate cards meet in one sheet that checks itself.
Accounting38× faster
Inbox
212 invoices
Coding
GL accounts
VAT check
whitelist match
Review
7 exceptions
Export
ledger / CSV
A month of invoices coded in minutes
Coding, VAT registers and completeness checks without retyping anything.
Finance30× faster
1$12,400reconciled
2$3,180no statement
3€1,890reconciled
4$7,250reconciled
Reconciliations with cited proof
Bank statements against the ledger, every match with its source line.
Legal32× faster
01Auto-renewal · 90 days
02Uplift capped at 5%
03Unlimited liability
Clause review across contract stacks
Hundreds of contracts screened for the clauses that carry risk.
Recruiting69× faster
1M. Nowakfit 92
2J. Weberfit 88
3A. Kimfit 61
4P. Silvafit 44
CV screening against one profile
Every candidate scored the same way, each score with its evidence.
Real estate36× faster
Rent_roll_2026.xlsxLease_abstracts.pdfBreak_options.csv
Lease terms across a portfolio
Every lease abstracted into one searchable, exportable table.
45×
faster invoice matching
69×
faster CV screening
0
values without a source
AI COLUMNS

The column is the interface.

A column is a prompt in plain language: pick an output type, write the question, test on one row, run on every document. A column can also pull live data from the internet, with sources.

Invoice matcher
Run all
▤ File
# Amount
◎ Payment terms ←
1
Semtech_MSA_2026.pdf
$12,400
Net 30 p.2
2
CloudProvider_TOS.pdf
$3,180
Net 60 · flagged p.4
3
Nordhaul_contract.pdf
€1,890
Net 14 p.1
4
Vector_amendment_v2.pdf
$18,220
→ human review · 0.61 ·
1,204 rows · avg 0.8s per row
Payment terms
Type
TextSelectNumberDate
Model
claude-haiku
Source
Documents + web
Prompt@ references other columns
Extract the payment terms from @File. Flag anything longer than Net 45.
Grounding · citations
links every value to its source page
Confidence gate< 0.90 → human review
TEST ON ONE ROW
◰ CloudProvider_TOS.pdfNet 60 · flaggedp.40.94
Save · run on 1,204 rows
same prompt, same format, same model · on row 1 and row 10,000
EVERY VALUE HAS A SOURCE

Click a cell. See the proof.

Every AI-filled cell opens the exact quote, its page and a confidence score. Approve, fix or rerun, one click each.

▤ File
◎ Doc type
# Value
1
◰ MSA_CloudProv_2026.pdf
Master Agreement
$1,240,000 p.4
#Value· MSA_CloudProv_2026.pdf
$1,240,000conf 0.94
p.4 §4.2 · Fees and Payment
"…total fees payable under this Agreement shall not exceed USD 1,240,000 over the Initial Term…"
Open in source viewer →
✓ Approve✎ Edit↻ Rerun
COMING TO QUANTY

Type a NIP. Get registry proof.

Soon every counterparty is checked against live state registries, not just the document. Try it now: one NIP, three registries and an official evidence ID you keep.

Verify
or try an example: 7740001454
Biała lista VAT
Ministry of Finance · wl-api.mf.gov.pl
·
VIES
European Commission · ec.europa.eu
·
KRS
National Court Register · api-krs.ms.gov.pl
·
ALSO ON THE WAY

Registries prove them. The open web completes them.

Verification is just the start. Describe the companies you need and Quanty fills the sheet from the open web: size, locations, people, news, funding. A full profile for every counterparty, not just a VAT status.

Food wholesalers in Wielkopolska, 20 to 200 employees
Complete · 24 companies · 6 columns
COMPANY
CITY
EMPLOYEES
LATEST SIGNAL
Agrohurt Wielkopolska
Poznań
120
opened a cold storage site
FreshLine Trade
Kalisz
45
expanding EU exports
Polfood Serwis
Leszno
80
won a public tender
Danex Dystrybucja
Konin
35
new contract in Germany
+ 20 more rows

In Quanty this runs as a column: every invoice and every counterparty verified automatically, every result logged with proof.

Join the waitlist
live demo · real registries · nothing you type is stored
YOUR AI TEAM

Agents with one job: knowing yours.

Build agents that specialize. Forecasts, margins, invoices, contracts. Trained on your company, working around the clock. Click one.

AGENTS
QUERY
Check June invoices against contracts and flag overcharges
Invoice agentlive
Invoice
Supplier
Amount
vs contract
FV 2026/612
Semtech
€12,400
✓ in line
FV 2026/619
CloudPro
€3,180
+18% · flagged
FV 2026/633
Nordhaul
€1,890
✓ in line
FV 2026/641
TransEu
€7,420
duplicate · flagged
ASK YOUR BUSINESS

Ask anything.
It has read everything.

Plain questions. Cited answers. And the next question worth asking.

Good morning, Dawid.

Describe a job, or pick up where you left off.

Ask anything, or describe a workflow, e.g. "check June invoices from @Invoices 2026 against the VAT whitelist"
+ files▣ library▤ templates
⏎ shows a plan first · nothing runs unreviewed
CHAT DRIVES THE SHEET

Say it. The sheet does it.

In the app, the chat slides in over any sheet. Ask, and columns appear, fill in, flag themselves and total up in the formula row. Ask for data your files cannot have, and a column fills itself from the live web, sources included.

Fleet Q3Chat with AI
Truck
Distance
Fuel cost
WGM 4412
92,400 km
€28,940
WGM 7031
88,100 km
€34,620
WGM 1288
61,700 km
€17,850
WGM 8821
101,300 km
€38,150
ƒx
AVG€0.34
4 rows · 3 columns
Chat with AI
add_column · Cost/kmset_formula_row · avg
Done. Fleet average €0.34 in the formula row, 2 trucks over the line.
THE WORLD, IN YOUR SHEET

Ask for the world. It appears.

Live data from the internet, straight into your spreadsheet or chat.

Ask
Live from the web · every value cited
Diesel · EU averageup 4% this week → your route costs +€3,100 a monthsource
German toll rise · Dec 1reroute 2 lanes → save €910 a monthsource
EUR/PLN weakerrefuel across the border → save €0.06 per litersource
live web research under the hood
QUANTY CONNECT

Deep research can also query premium databases alongside the live web: company registries, traffic analytics, financial fundamentals. One run, one cited answer.

Fiber.aiSimilarwebBaselayerAffiliate.comParticleFinancial DatasetsJinko

More sources on request: Crunchbase, Harmonic, ZoomInfo and others.

IT LEARNS YOUR COMPANY

The longer it runs, the sharper it gets.

Every document feeds one growing model of your company, plugged into the world outside. Each month, sharper answers.

Invoice 88123
Semtech Corp.
PO-2214
MSA 2026
Email thread
Payment
Credit note 114
Shipment SH-88
Rate card Q3
GL 4400
Bank statement 07
Delivery note 4411
VAT register Q3
Invoice 88124
issued bymatchesgoverned byarrived viasettled byfulfilled bypriced bycorrectsposted tosigned withreconciled inconfirmed byreported infollowsprices
THE LIBRARY

One library. Every document, indexed.

Everything you drop in or email lands here: read by OCR and vector-indexed, ready for every sheet and chat.

Library212 items
+ New FolderUpload
Search files and folders...
List View
Name
Type
Source
Pipeline
Added
Invoices 2026
Folder · 212
Today
Legal 2026
Folder · 87
Yesterday
Invoice_88124.pdf
212 KB
email
Ready
Today
MSA_2026.pdf
1.4 MB
upload
Ready
3 days ago
Rent_roll_2026.xlsx
890 KB
upload
Processing
Today
Scan_receipt_04.pdf
96 KB
email
Queued
Today
files arriving via ⚡ keep the original email attached as provenance
Drop files anywhere · PDF, DOCX, XLSX, CSV, images (OCR)
AUTOMATIONS

Forward an email. That's the whole setup.

Rules read like sentences: mail arrives, decide what it is, act. Anything uncertain waits for a human. Nothing writes on its own.

← Automations
Invoice intake● Active
Send test emailSave changes
TRIGGER · email arrives
Inbound email
+invoices@in.quanty.app
Only allowlisted senders · PDF attached
CLASSIFY · AI decision
What is this email? AI
Decide: invoice, credit note, something else.
Confidence < 0.8 → human review
BRANCHES · then do
invoicesave file → append row → run AI columns → reply "received"
credit noteappend row · flag negative → notify #finance
something elsehold in Needs review · no writes
dry run passed · nothing is written until you activate
INTEGRATIONS

Email and Google Sheets today. Your whole stack next.

The beta starts with inbound email and Google Sheets: every sheet gets its own address, and any Google spreadsheet can sync rows in or out. More connectors are on the roadmap.

Emailin betaGoogle Sheetsin betaGmailcoming soonOutlookcoming soonMicrosoft Teamscoming soonSlackcoming soonGoogle Drivecoming soon
DUE DILIGENCE

Every source in. Every deliverable out.

Point Quanty at the data room and it reads everything at once. Out come models, decks and alerts in the systems you already use.

Data room
SharePoint
Email inbox
invoices & threads
CRM export
counterparties
Bank statements
CSV / XLSX
Contract stack
MSAs & amendments
Quanty
cross-document reasoning
Financial_model.xlsx
created
Deal_summary.pptx
drafted
Red-flag report
saved to Drive
Email to counsel
sent
#deal-room
notified
WHY NOT JUST A CHATBOT?

Ask for a forecast. Two very different answers.

Same numbers, same question, asked twice. On the left, a plain chatbot. On the right, the same model inside Quanty.

The full story: why LLMs need this →
What will Q4 revenue be, based on ◰ Actuals_Jan_Sep.xlsx ?
Without Quanty · plain chatbot
RUN 1
"Q4 revenue should land around €2.4M, roughly 8% ahead of Q3."
RUN 2 · same question again
"Expect about €2.55M in Q4 if current trends hold."
A different number every run. Nothing your CFO can check.The arithmetic happens inside the prose. There is no formula to inspect.Next month it invents another method, so nothing compares.
With Quanty
RUN 1
€2.41M⚑ seasonal run rate · 9 months of actuals9 mth actuals0.94
RUN 2 · same question again
€2.41Midentical · same method, same inputs
method · Actuals to September, seasonal index per client, open pipeline weighted. The formula sits in the column, not inside the answer.
The number comes from code you can read. The model chooses the method and explains it.Every input traces back to the invoice or statement it came from.Same columns next month, so July and August sit side by side.
WHY QUANTY

AI can already read your paperwork.
Quanty makes the numbers defensible.

Asking a chatbot for a forecast fails in three predictable ways. Quanty keeps the reading ability and engineers away the rest.

01 · THE INPUTS

A forecast is only as good as the reading underneath it.

Paste 300 invoices into a chat and attention thins out: document 214 blends with 209, a credit note slips by, and the total is quietly wrong.

✕ ONE GIANT PROMPT
Document 3 · read correctly
Document 214 · blended with 209
Document 288 · credit note missed
All 300 documents compete for the same attention. The error ends up in your total.
✓ ONE ROW, ONE DOCUMENT
Row 3 · its own focused read
Row 214 · its own focused read
Row 288 · its own focused read
Quanty gives the model one document and one question at a time. The actuals your forecast stands on are actually right.
02 · THE METHOD

A model should not be doing your arithmetic.

Ask a chat for the same forecast three months running and you get three methods, none of them written down. In Quanty the method is a column: the model picks it and explains it, code computes it, and it stays pinned until you change it.

✕ IMPROVISED EVERY MONTH
JUL "about €2.4M"
AUG "roughly 8% up"
SEP "€2.6M if trends hold"
Three months, three methods. There is no variance to explain, only new numbers.
✓ SAVED AS A COLUMNseasonal run rate · pinned
JUL NET_30
AUG NET_30
SEP NET_30
Same method every month, so this month and last month sit side by side.
03 · THE BOARDROOM

Someone will ask where the number came from.

A chatbot hands you a deck with no lineage, so the honest answer is that you will check and come back to them. In Quanty every figure on a slide stays wired to the sheet, and every cell to its document.

✕ A FLUENT GUESS
"Q4 revenue will reach €2.55M, up 8% on Q3."
No source. No method. Wrong, and it reads exactly like the truth.
✓ A CITED CELL
Q4 forecast €2.41M · margin 11.2%
◰ Invoice_88124.pdf · p.1confidence 0.98
Click the slide, open the sheet, open the invoice. Three clicks, in the meeting.
IN SHORT

Same model. Different discipline.

A CHATBOT
QUANTY
INPUTS
All your documents in one prompt, competing for attention
One document, one question per cell
METHOD
Arithmetic improvised inside the answer
A formula you can read, run on every row
OUTPUT
Prose you retype into slides
Sheet, dashboard and deck from the same numbers
PROOF
None. You reread the documents to verify
Every figure opens its page, live in the meeting
WHEN UNSURE
Guesses confidently anyway
Goes to a human review queue, not into your data
EVERY MONTH
A new shape each time, nothing to compare
Same columns, same method, month over month
LAUNCHING FALL 2026

Be there before everyone else.

Join the waitlist and get first access as seats open. Whatever we charge later, beta members always keep the better offer.

By submitting this form I confirm I have read and accepted Quanty’s Privacy Policy.

First in lineAlways our best offerDirect line to the team
no spam · one email when your seat opens
QuantyQuanty© 2026 Pluscode Sp. z o.o.
Quanty is a product of Pluscode Sp. z o.o., Poznań, Poland
ProductIndustriesPricingWhy QuantyPrivacy PolicyTerms of ServiceGoogle user data