Fingrip helps the finance teams of enterprise groups get their figures right before they are published. It reads your documents and workbooks, checks every figure against its source and runs the review and approval around the checks. Fingrip reads. Your people decide.
Veridia Group, a fictional company · synthetic figures
Net revenue for the quarter was €284.6 million. EBITDA reached €34.4 million, a margin of 12.4%. Personnel costs were €88.4 million and net debt closed at €842.1 million.
Follow one figure through a close at Veridia Group, a fictional company: PASSO reads it, QUADRA ties it to the workbook cell behind it, and REGIA runs the review and approval around the check.
In productionThe video arrives here
Each product functions on its own, and each builds on the ones before: QUADRA checks the figures PASSO reads, and REGIA runs the close around QUADRA’s checks. No double effort. Move from QUADRA to REGIA, and the rest of your QUADRA subscription is credited.
Turns Word, PowerPoint, PDF, Excel and XBRL files into figures that AI and code can read, each with its exact source.
Checks every figure in your commentary, decks and reports against the workbooks behind them, and routes what does not tie.
Runs the whole close around the checks: phases, workflow, calendar, checklists, reviews and the issue resolution log.
On your own One person can use it alone.
As a team The whole group joins.
PASSO by Fingrip
Finance-grade document parsingPASSO reads Word, PowerPoint, PDF and Excel files, the older .doc, .ppt and .xls formats and .xlsb workbooks, XML, XBRL, iXBRL and pictures: one file at a time, a ZIP or a whole folder. It returns every number with the text as written, its value, sign, unit, scale, label, period and place, and checks its reading against the original, so AI and code get a view of your documents they can trust.
Figure model one record per figure
Not needed: read by code from the file itself.
Word, PowerPoint, Excel and PDF, read natively with chart data and cell addresses, and the older .doc, .ppt and .xls formats and .xlsb workbooks too. XML, XBRL and iXBRL through an XBRL processor: the ESEF reports of listed companies and the supervisory reports of banks and insurers. Send one file, a ZIP or a whole folder.
A picture on its own, a scanned table, a chart pasted as an image or a photographed page is read with OCR or vision. Every figure from a picture carries a confidence score, and a person confirms the uncertain ones.
Every file gets its figure model and a Markdown text an AI can read. PASSO then checks that reading against the original file, and what does not match is flagged, never passed on. It compares nothing across files: that is QUADRA’s job.
Batch jobs for large packs, and tools to parse a file, list its figures and, when a file is uploaded again, see what changed since its last version. Ask your AI assistant to read a file: the connector gives you a one-time link to upload it, and its figures come back with their places, processed in the EU.
The figure model as JSON and the text as Markdown, one ZIP per batch. Kept 30 days by default, from none to a year, to download again at no cost. The original is deleted after parsing unless you keep it.
Every file is known by its content hash. The same file read again by the same person within the result window returns the stored result, free.
Two hundred pages of PDF are too long and too loose for an AI to read as they stand. PASSO gives each file back as Markdown: tables kept as tables, headings as headings, at a fraction of the size, and beside every table the page it came from, so each figure can be traced to its source.
# Consolidated income statement <!-- page 112 · table 1 · € million --> For the year ended 31 December 2025 | Line | Note | 2025 | 2024 | |--------------------------|-----:|--------:|--------:| | Net revenue | 4 | 1,138.4 | 1,071.9 | | Cost of sales | 5 | (612.7) | (583.0) | | **Gross profit** | | 525.7 | 488.9 | | Selling and distribution | 6 | (176.2) | (168.4) | | Administrative expenses | 6 | (98.5) | (94.1) | | Other operating income | 7 | 12.3 | 9.8 | | Other operating expenses | 7 | (21.9) | (19.6) | | **Operating profit** | | 241.4 | 216.6 | | Finance income | 8 | 6.8 | 5.1 | | Finance costs | 8 | (31.7) | (29.4) | | **Profit before tax** | | 216.5 | 192.3 | | Income tax | 9 | (54.1) | (48.6) | | **Profit for the year** | | 162.4 | 143.7 | | Basic EPS (€) | 11 | 1.62 | 1.44 | <!-- 28 figures · each with its page, table, row and column -->
QUADRA by Fingrip
The figure checkerBy the time the pack is final, one figure lives in six files. Someone retypes it, a chart keeps last month’s data, a commentary quotes an older version. QUADRA traces every figure in your Word, PowerPoint and PDF documents to the Excel workbooks behind them, and reports what does not tie.
PDFResults summary.pdf · page 1
| Q2 2026 | Q2 2025 | |
|---|---|---|
| EBITDA margin | 12.1% | 11.8% |
TTies to KPIs!D8
DOCManagement commentary.docx · page 2
EBITDA rose to €34.4 million, a margin of 12.1%, up from 11.8% a year earlier.
TTies to KPIs!D8
PPTBusiness review.pptx · slide 5
MMismatch: the workbook holds 12.1%, 0.3 pts apart
PDFBoard pack.pdf · page 7
EBITDA margin · Q2 2025, Q2 2026
TTies to KPIs!D8
XLSReporting pack.xlsx · KPIs
| B | C | D | |
|---|---|---|---|
| 4 | Q2 2025 | Q2 2026 | |
| 5 | Net revenue | 268.9 | 284.6 |
| 6 | EBITDA | 31.7 | 34.4 |
| 7 | Capex | 12.9 | 14.1 |
| 8 | EBITDA margin | 11.8% | 12.1% |
KPIs!D8 = D6 / D5 = 12.09%
Synthetic example: Veridia Group is a fictional company, and every name and figure is invented.
After each run the right people are told, one message each. Notifications report differences computed, so they go out without waiting for approval.
From the summary to every line. A finding opens the figure on its page or slide, in Word, PowerPoint, PDF or a picture, beside the workbook cell it matched.
Interactive online. Offline, one HTML file that works without a connection, and an Excel workbook with one row per figure.
A match needs the same label, period, unit and scale, never the number alone. Code compares each pair against your tolerance. The AI’s confidence, high, medium or low, is about the match, never about whether the numbers agree. By default documents are the targets and workbooks the sources; switch any file’s role, for example to check a workbook against the published PDF.
The matches your people confirm are kept per lane, with who confirmed them and when, and the next cycle starts from them. Controllers see, correct and remove them, a page at a time.
Every change to QUADRA’s checking engine is run against a fixed regression set of report packs, and must give the same outcome and cell for every figure. A difference is shown figure by figure, and only a person accepts it.
QUADRA runs in the Fingrip web app with only its parts switched on: cycles and lanes, runs, findings, reports and settings. It also works through your AI assistant, with the Fingrip connector, and an Excel add-in follows. Move to REGIA and its modules appear in the same place, with no migration.
REGIA by Fingrip
Runs the closeREGIA runs the whole reporting cycle around the checks, from consolidated data to released reports. It includes QUADRA and PASSO. Draw each phase on a canvas: objects linked by arrows and yes/no decisions, green for yes and red for no, each with an owner and a deadline.
COFinance operationsContributors, controllers, lead controller
Contributors upload in the browser, email a Fingrip mailbox or publish to an exchange folder. Deadlines come from your calendar; reminders run themselves.
QUADRA’s reconciliation and the anomaly checks review the material automatically, against the baselines each check names.
Your controllers’ checks, each with an ID, a prompt, evidence examples and a threshold. Grip proposes, code applies the threshold, the controller decides.
The controller’s review, then, where the workflow asks for it, the lead controller’s. The same person never submits and approves.
Upload next year’s calendar as the Excel or Word file it already is. REGIA reads every date exactly as written, you confirm the result, and the timeline schedules the kick-offs, file requests and reminders. Milestones reach your contacts’ Outlook calendars.
The issue resolution log holds open mismatches, failed checks, rejections and questions for the whole cycle. An open item can send the lane back to File upload for a new version.
Deadlines reach your contacts’ Outlook calendars through a calendar feed, or are booked with reminders where you allow it. Replies to notifications are filed on the item they answer.
Administrator, contributor, controller, lead controller, analyst and auditor, set per lane and phase. A new person changes nothing in the process, and every approval is signed by a named person.
A disclosure management tool is where a group writes its annual report: authors work in sections, and the figures are linked to the systems behind them. Fingrip does a different job. It reads the finished files, whichever tool made them, and checks every figure against its source.
CCH Tagetik Intelligent Disclosure Wolters Kluwer
Oracle Narrative Reporting Oracle Cloud EPM
Workiva financial reporting
SAP Disclosure Management SAP
OneStream Narrative Reporting OneStream
Certent Disclosure Management insightsoftware| Disclosure managementThe products named above | FingripPASSO, QUADRA and REGIA | |
|---|---|---|
| The job | Writing the report. Authors work on sections of one document, its figures linked to the data behind them. Most also tag and file it as XBRL. | Checking what was written. Every figure in the finished files is traced to the workbook cell, XBRL fact or report it comes from, and what does not tie is reported. |
| What it covers | The documents written in it. Linked figures follow their source, and several now check consistency inside those documents. We found none that describes checking a finished file made in another tool. | Every file in the pack, however it was made. The designed PDF, the press release, the investor deck, a unit’s commentary, a scanned table. |
| Where the figures come from | The systems connected to it. CCH Tagetik’s, Oracle’s and OneStream’s are part of their vendor’s finance platform. Workiva and Certent connect to many systems. | Any system that produces a file. SAP, Oracle, CCH Tagetik, OneStream or plain Excel: Fingrip reads the export. No connector, nothing inside your network. |
| How your people work | Inside the tool. Authors write through its Office add-ins or in its own editor, on sections that are checked out and in or shared live. | As they do today. In Word, PowerPoint and Excel as they are. Files arrive by upload, by email, through the API or an exchange folder. |
| Getting started | The report is set up first. Its structure, its data links and who writes what. Vendors offer implementation services, partners and courses for it. | Your next close. Nothing to install or connect. A pilot is set up with you in its first week and checks your real pack. |
| The licence | It differs, and is seldom published. Oracle licenses by named user, with a minimum number of users. Workiva licenses by solution, with unlimited seats. We found no published metric for the others. | By use, at published prices. Pages and lanes, with unlimited users: everyone who submits, checks or approves joins at no extra cost. See the pricing |
| The evidence | The link, and review. A linked figure follows its source. Validation rules and, in some, AI agents review the rest. | A computed tie for every figure. Its file, page and cell, the tolerance applied and the person who decided, on a trail chained by hashes. Whether two numbers agree is never left to a model. |
Fingrip does not write, tag or file your report: that stays with your disclosure tool. Fingrip checks what it produces, the published PDF and the iXBRL file included, against the workbooks behind it and against every document it never sees.
And it is not only for the annual report. The monthly pack, the board deck and each unit’s submission go through the same checks, every close.
Based on each vendor’s public product pages, documentation, price lists and annual reports, read on 11 November 2026. The products differ from one another and change: if we have yours wrong, tell us at info@fingrip.ai and we will correct it. Product names and logos are trademarks of their owners and are used only to identify the products; Fingrip is not affiliated with, or endorsed by, any of them.
Grip is the assistant in PASSO, QUADRA and REGIA. On every screen it says what changed and what to do next, in a few short sentences. Ask it anything about your files or your close: it answers from what Fingrip holds, cites the source and never invents a figure.
Two figures on slide 5 come from a picture. One is below 95% confidence: please confirm it.
The EBITDA margin on slide 5 does not tie. KPIs!D8 is the matched cell, confirmed last cycle.
I propose a fail on GC-02: the intercompany balance differs by 60,000, above the threshold.
You asked for total revenue. I don’t add figures up on my own. Calculate it, labelled as calculated?
Two to four sentences. Facts first, then one action you can take with a click.
Quoted with the file and the cell they came from. Totals only when you ask, labelled and logged.
Proposals and explanations stay apart from your decisions, in the record and on screen.
On every screen, with a side panel and a command bar: press Ctrl K and ask.
Synthetic examples from Veridia Group, a fictional company.
Code reads, matches and ties. AI explains. People decide. Built for regulated finance from the first line.
Fingrip quotes figures exactly as your files give them. It calculates only when a person, a scheduled rule or an agreed check asks, and labels the result.
Every AI output is a draft until a person accepts, amends or rejects it. No agent closes its own finding, and AI and human content are kept apart.
Fingrip never changes an uploaded file or a source system, and needs no access into your network. Each customer is isolated in the database itself.
Storage and processing stay in the EU. You choose the AI model, EU options included, and see where it runs. System logs record actions, never your figures.
Fingrip remembers the matches your people confirm, so recurring packs need less checking each period. That memory is for your organisation only. No model is ever trained on your data.
Every upload, check, decision and notification, chained by hashes and held with your data. Change one event and every later hash breaks. Auditors export it with the evidence.
Everyone who submits, checks or approves joins at no extra cost. PASSO is priced by pages, QUADRA by lanes and pages, and REGIA by lanes and AI capacity. All prices exclude VAT.
For partners and developers. Developers pay after each month’s use; the partner plan is billed monthly in advance.
Talk to us about PASSOThe same file, known by its content hash, read again by the same person within the result window costs nothing. Results download again for free.
What counts. A lane counts once, however many cycles it reports in. A page is a document page or slide that QUADRA checks: source workbooks are not counted, and a page already checked is free.
Extra pages €0.30 each, cheaper by volume.
Start a QUADRA pilotLanes count as in QUADRA; phases do not change the price. Large groups can bring their own model contract through an EU route and pay the REGIA fee only.
Start a REGIA pilot| A document page reconciled | 1 |
| A checklist check evaluated | 2 |
| A question answered | 1 |
| An agent task, such as drafting IRL follow-ups | 5 |
| A page that did not change since the last run | 0 |
Extra credits come in packs of 25,000 at €7,500, cheaper by volume. We never stop a close: we alert at 75% and 90% of the pool and bill packs at quarter-end.
Start with QUADRA if one team wants every figure in its documents checked before the board sees them. Choose REGIA to run the whole close around the checks: phases, workflow, calendar, checklists and reviews. PASSO is for partners and developers who build on financial documents; inside Fingrip, it powers the other two.
Yes. REGIA includes QUADRA and PASSO, and its modules appear in the same workspace, with no migration. The rest of your QUADRA subscription is credited.
No. Your files are read-only. Fingrip quotes figures exactly as they appear and calculates only when a person, a scheduled rule or an agreed check asks for it. The result is labelled as calculated and logged.
No. Fingrip checks the documents your tools produce. It never connects into your network or writes to your consolidation system: files arrive by upload, by email, through the API or through an exchange folder you control.
Because the two do different jobs. Your disclosure tool writes the report and keeps its linked figures in step. Fingrip checks the finished files, those your tool never sees included: the designed PDF, the press release, the investor deck, each unit’s pack. See the two side by side.
Word, PowerPoint, PDF and Excel, the older .doc, .ppt and .xls formats and .xlsb workbooks, and XML, XBRL and iXBRL: running text, tables and the data behind charts. Send one file, a ZIP or a whole folder. Pictures, on their own or inside documents, such as a scanned table or a pasted chart, are read with OCR or vision and carry a confidence score.
Storage and processing are in the EU. The AI model is your organisation’s choice, with EU options, and Fingrip shows where each one runs, including any outside the EU. System logs record actions, never your figures, and no model is trained on your data. The audit trail is your own record, held with your data.
Because a close involves every unit, function, reviewer and auditor. Charging per seat would keep people out. You pay per product instead: pages, lanes and AI capacity.
A pilot runs your next real close. In the first week we set up your lanes, cycles, tolerances and source areas with you, and for REGIA your workflows, phases, checklists and calendar.
Yes. Notifications come from Fingrip by default, or from your own name and domain once your IT has authorised it.
To be the integrity layer of financial reporting in Europe: where every figure is traced to its source, and every check signed off by a person, before it reaches a board, a regulator or the market.
Fingrip gives finance operations, FP&A and investor relations teams a firm grip on their figures. Code reads, matches and ties; AI explains; people decide.
Figures as given, storage and processing in the EU, and a named person behind every decision.
We are looking for pilot customers: groups that bring their next real close to Fingrip, with our team beside yours. A pilot is credited to your first year.