ReDiX Labs | ReDiX Informatica

Introduction

We are a consulting and software development company established in 1990 and formed by professionals who are passionate about their work.
During 2023 we began our experiments and developments in the world of machine learning and GenAI 🤗
Today that work comes together in Alembic, our document intelligence platform.

Alembic - Your knowledge, distilled.

Alembic is an AI assistant that works on your organisation's own documents, not on the open web. Users upload their files into a workspace, ask questions in natural language, and get answers with citations pointing to the exact passage they came from. When the answer isn't in the documents, Alembic says so instead of making it up.

Built for data that can't leave the building. Alembic is designed to run on-premise, down to fully offline on local hardware with open-weight models. It can also run on managed infrastructure or on the customer's own cloud: same product, three deployment options.

What it does

• Search with sources - semantic search ("what happens if we withdraw early?") and exact, exhaustive search (every occurrence of a code or reference), with citations on every claim
• Any document - PDF, DOCX, PPTX, web pages, CSV; scanned PDFs are handled automatically with OCR
• Version comparison - what was added, removed or changed between two versions of a contract or policy
• Real calculations - numbers in spreadsheets are computed by running code in a sandbox, not estimated by the model
• Italian legislation - current text of laws and articles, the text in force at a given date, and amendment history
• Visible work - users see the tools used, the work plan for complex requests, and where each answer comes from
• Model-agnostic - works with multiple LLMs, swappable via configuration with no vendor lock-in

➡️ Learn more: www.redixlabs.it/alembic

Services

We offer the following specialised AI and ML services:

• Italian Dataset Creation
  Design and construction of high-quality, domain-specific datasets in Italian for training and evaluation of language models.
• LLM Fine-tuning
  Adaptation and specialisation of large language models for vertical use cases, industry-specific terminology, and proprietary workflows.
• RAG & Function Calling Pipelines
  Architecture and development of production-grade applications integrating Retrieval-Augmented Generation and external tool-calling - the same expertise behind Alembic.
• Unstructured Data Extraction
  Custom projects that transform raw documents, PDFs, emails, and heterogeneous sources into structured, queryable information.

Contacts

We welcome enquiries about projects, collaborations, consulting engagements, and Alembic demos.

• Website: redixlabs.ai · redix.it · Alembic
• Email: redix.ai@redix.it
• Phone: +39 039 2458044