Skip to content
APFerrer
P-DAT-01 · ProductCross-industry (SMEs and mid-sized companies)

Natural language questions on company data, with pseudonymization

An assistant that answers questions in Spanish about company data, with per-profile permissions, identifying data replaced by codes before it reaches the AI, and the source of every figure.

Source codeExclusivity availableInstalledOEM

A company's data is usually spread across spreadsheets, the CRM and a database, and every question ends up as a request to IT. This software gives you natural language query for business data: you ask in Spanish and get an answer with figures, tables and charts, each with its source. Before anything reaches the AI provider, identifying data is replaced by codes, and each profile sees only what it is allowed to see.

Who it is for

  • Companies that want to give management and middle managers access to their data without preparing a report for every question.
  • IT managers and data protection officers who want to control what goes out to the AI and who sees what.
  • Consultancies and integrators that want to offer this service to their clients.

What natural language query for business data solves

  • Scattered data. Sources are copied into a database of its own, without putting load on the source systems.
  • Personal data going out to the AI. The provider only receives totals, or rows with the sensitive fields replaced by codes. Whoever has permission sees the real value on screen.
  • Figures with nothing behind them. If the AI writes a number that is not in the results, the system rejects it. Every figure points to its source, or to the document and page.
  • Permissions that are hard to maintain. Rules are defined per profile and applied before every query.

What you can do with it

Connect your data

  • Load Excel workbooks with several sheets and CSV files. Data types and total rows are detected.
  • Read contacts, companies and deals from HubSpot CRM.
  • Read the tables you choose from your own PostgreSQL database.
  • Schedule loads and see the date of the data in every answer.

Ask and get answers

  • Ask in Spanish. The AI does not query your database freely: it proposes a plan that the system checks and runs.
  • Get a headline, an explanation and up to two charts and two tables, with a notice if the answer is partial because of permissions or if the question is out of scope.
  • Review the history, follow conversation threads and use suggested questions.

Protect personal data

  • Decide what happens to each field: keep it, replace it with a code, remove it or never let it out.
  • Keep the same code for the same value, so analyses add up.
  • Mask IBANs, email addresses, Spanish ID numbers (DNI and NIE) and phone numbers in document text before it is sent to the AI.
  • Set how long the log is kept.

Control who sees what

  • Assign owner, administrator, permissions manager, auditor and read-only roles.
  • Restrict each profile by rows, columns, document groups or period, or let it see only totals with a minimum group size.

Query documents

  • Upload PDF, Word, PowerPoint and image files, with text recognition for scanned documents too.
  • Search the text in Spanish, ignoring accents, with the document and page cited.

Run the installation

  • Choose Anthropic, OpenAI or Google Gemini with your key, stored encrypted, and see every request sent to the AI with an estimate of its cost.
  • Start with a guided setup, manage users and profiles, and add your name, colors and logo.
  • Export to CSV the log of questions, requests sent to the AI, answers and admin actions.
  • Limit failed sign-in attempts.

Benefits

  • Management gets figures on the spot, without waiting for a report.
  • You use AI without handing a third party the identifying data of customers or employees.
  • Every figure carries its source and can be checked.
  • Permissions are defined once and apply to every question.

Industry context

General industry data with its public source. These are not results of this product.

What you get with each license

Price upon request, depending on license type and scope. Maintenance and support are contracted separately for every license type.

  • The software installed on your server or in your cloud, under a license to use.
  • Setup of your sources, profiles and the treatment of each field.
  • Initial training for administrators.

Technical specifications

Language and runtime: TypeScript 5.9 on Node.js. Minimum supported version: Node.js 20.19 (the one Next.js 16 and Prisma 7 require in their configuration).

Frameworks: Next.js 16.4 and React 19.3, with Prisma 7.10 for data access and Tailwind CSS 4.3.

Database: PostgreSQL, with the unaccent and pg_trgm extensions.

Operating systems: A server with Node.js.

Interface languages: Spanish.

Required infrastructure:

  • PostgreSQL.
  • File storage on local disk or in an Amazon S3-compatible service.
  • An hourly scheduled task (cron) that runs maintenance.
  • A master key in the server configuration.
  • AI provider: Anthropic, OpenAI or Google Gemini (requires your own account).
  • Datalab, to extract text from documents, only if you query documents (requires your own account).
  • HubSpot CRM, only if you connect it as a source (requires your own account).
  • An SMTP server, optional.

Private deployment: Installs on your infrastructure, but needs a connection to the AI provider and, if you query documents, to Datalab.

Main libraries:

Library Version License
next 16.4.2 MIT
react 19.3.1 MIT
@prisma/client 7.10.1 Apache-2.0
@anthropic-ai/sdk 0.137.0 MIT
openai 6.49.0 Apache-2.0
@google/genai 2.31.0 Apache-2.0
@aws-sdk/client-s3 3.1156.0 Apache-2.0
exceljs 4.4.0 MIT
papaparse 5.7.0 MIT
nodemailer 10.2.1 MIT-0
zod 4.6.7 MIT

Dependencies with specific license terms:

  • jszip 3.10 (a dependency of exceljs): MIT OR GPL-3.0-or-later.
  • @img/sharp-libvips-* (libvips binaries, a dependency of Next.js): LGPL-3.0-or-later. @img/sharp-win32-* and @img/sharp-wasm32: Apache-2.0 AND LGPL-3.0-or-later.
  • buffers 0.1.1 (a dependency of exceljs): MIT/X11. The license does not appear in the package published on npm: the author declared it in the original repository (commit 1b745ee).

Compatibility and integrations: Anthropic API, OpenAI API and Google Gemini API through their official libraries, HubSpot CRM API v3, Datalab API v1, Amazon S3-compatible storage, PostgreSQL, Excel and CSV, and SMTP email.

Redistribution terms

The asset's own code can be embedded in the licensee's commercial products under the source code, exclusive and OEM licenses, on the terms of the contract. With the installed license, the software is used and not redistributed. Specific license terms apply to jszip (MIT OR GPL-3.0-or-later) and the libvips binaries (LGPL-3.0-or-later); the details are in the Technical specifications.

Installation requirements

Installed on your server or in your cloud, one installation per company, with its own PostgreSQL database. Files are stored on disk or in Amazon S3-compatible storage. You need your own account with an AI provider and, to query documents, another with the text extraction service. Outgoing SMTP email is optional. Questions, answers and document search work in Spanish; other languages need adaptation.

Integrations

Anthropic, OpenAI and Google Gemini with your key, HubSpot CRM, PostgreSQL, Excel and CSV, Datalab to read documents, and SMTP email.

FAQ

Frequently asked questions

  • What data exactly reaches the AI provider?

    The question, the description of the tables the profile can see and the results, with sensitive fields replaced by codes or removed. In document text, IBANs, email addresses, DNI and NIE numbers and phone numbers are masked first.

  • Can the AI make up figures?

    Figures are checked against the results obtained. If a number appears with nothing to back it, the answer is corrected or only the data is shown.

  • Can we use our own AI account?

    Yes. Anthropic, OpenAI or Google Gemini, with your company's key; your company contracts and pays its provider directly.

  • Does it connect to our ERP or to Microsoft 365?

    The included connectors are spreadsheets, HubSpot and PostgreSQL. If your ERP exports to Excel or CSV, or uses PostgreSQL, it works through them. Other connectors are assessed separately.

  • Is maintenance included?

    No. Maintenance (updates and bug fixes) and support (questions, training and adaptations) are contracted separately.

Request

Request information on P-DAT-01

Fill in the form with the asset reference and the license type you're considering. I reply in writing. If you need to review code, tick the NDA box and we'll sign one before anything is shared.

Controller: APFerrer. Purpose: to answer your request. You can exercise your rights as described in the privacy policy.

Price upon request, depending on license type and scope. Maintenance and support are contracted separately.