EMBEDDABLE AI COMPONENT

Knowledge infrastructure.
For your software.

Cortex adds document processing, semantic search and source-linked AI answers to your applications. An API component for software providers and enterprises building knowledge capabilities into their own solutions.

  • API-first
  • Traceable evidence
  • Configurable models
INTEGRATION ARCHITECTURE

YOUR APPLICATION LAYER

Business applicationCustomer portalAI assistant
CortexKNOWLEDGE COMPONENT
Ingestion
Retrieval
Evidence
Documents & sourcesPDF · TEXT · MD · HTML
AI modelsOpenAI-compatible · Ollama
Part of your architecture
Developed in SwitzerlandPDFTEXTMARKDOWNHTML

ONE COMPONENT. PART OF YOUR SOLUTION.

Integrate knowledge capabilities. Extend your application.

Cortex provides the knowledge processing behind your software. You define the interface, business logic and access controls. Cortex delivers processed content, retrieval results and answers with source references.

01

Extend existing applications

Add knowledge capabilities to products, portals or internal business applications through a common API.

02

Centralise knowledge processing

Use one component for document ingestion, language processing, retrieval and evidence.

03

Make results traceable

Connect statements to their source passages and document versions within your own application.

COMPONENT CAPABILITIES

Document processing, retrieval and evidence.

The core functions of a document-based knowledge layer, integrated into your software through defined interfaces.

Document ingestion

Accept PDF, UTF-8 text, Markdown and HTML. Extract text and apply OCR selectively to PDF pages whose content cannot be read directly.

Semantic retrieval

Retrieve relevant passages by meaning. Restrict the retrieval context to explicit knowledge scopes or a single document.

Source-linked answers

Generate AI answers from retrieved content. Individual claims remain connected to evidence paths and passages in the original.

Multilingual processing

Process German, English, French and Arabic. An English processing layer makes the content accessible while preserving its link to the original.

Explicit knowledge scopes

Organise documents in separate scopes. Your application specifies the request context; permissions are handled by your solution architecture.

Versioning & lifecycle

Manage processing status, active revisions, retention and legal holds. New revisions become active only after processing is complete.

PROCESSING MODEL

From ingestion to API response.

The connection between source document, processing stages and result is preserved.

  1. 01

    Ingestion

    Submit documents, assign knowledge scopes and initiate processing.

  2. 02

    Transformation

    Extract text, process languages and structure content semantically.

  3. 03

    Retrieval & query

    Retrieve relevant passages and optionally generate source-linked answers.

  4. 04

    Evidence

    Return results with document revision, source passage and provenance to the application.

RESULTS FOR YOUR APPLICATION

An answer. A verifiable evidence path.

Your application receives more than generated text. Cortex connects claims to document revisions and original passages. Integrate this evidence into your own review and approval processes.

Technical evaluation
Illustrative component query
Knowledge scope/ Contracts
QUERY

What notice period applies to the maintenance agreement?

CortexRESULTSource-linked result

The agreement can be terminated with 90 calendar days’ notice before the end of the contract period.

Answer → Passage → Original

Illustrative integration example with fictional documents. Not a live query.

APPLICATION AREAS

For products and platforms that process knowledge.

Cortex is embedded as application infrastructure within your solution. Users continue working in your organisation’s interfaces and processes.

SOFTWARE PROVIDERS

Add knowledge capabilities to your products.

Build document search and source-linked answers into your software. Your product logic and user experience remain the starting point.

ENTERPRISE PLATFORMS

Access knowledge through a shared component.

Connect business applications to a common knowledge-processing layer and organise document collections in explicit scopes.

BUSINESS SOLUTIONS & PORTALS

Integrate evidence into existing workflows.

Embed source-linked results in contract review, technical analysis or customer portals, with references back to the document.

ARCHITECTURE & INTEGRATION

Your architecture. Defined interfaces.

Cortex handles document-based knowledge processing. Your overall solution owns the interface, business processes, identities and access controls.

  • Documented REST API

    Access documents, processing status, retrieval results, answers and evidence from your applications.

  • Configurable model integration

    Connect OpenAI-compatible services or local models through Ollama. Define model choices and data flows for your use case.

  • Fit your operating model

    Evaluate container-based execution in your environment. Plan storage, access protection and production operations as part of the complete solution.

Your application
REST API
Cortex
OpenAI-compatible
Local models · Ollama

Currently for demos and pilot projects. Public production use requires additional security and operational approvals.

FAQ

Integration questions

Is Cortex a complete end-user application?

No. Cortex is an embeddable software component with an API. You integrate document processing, retrieval and source-linked answers into your own products. The user interface and business workflows belong to your application.

What does a query return?

Alongside relevant passages and optionally generated claims, Cortex returns their source references. The evidence path leads through processed content to the unchanged document revision, including page and text region for PDFs. Domain assessment and approval remain in your process.

Which formats and languages are available?

Processing supports PDF, UTF-8 text, Markdown and HTML, as well as German, English, French and Arabic. OCR is applied selectively to PDF pages. The English processing layer remains connected to the original; Arabic text preserves its logical order and direction.

What needs to be resolved before production use?

Cortex is in its non-public prototype phase and can be evaluated in a bounded proof of concept. Authentication, permissions, tenant mapping, platform choices and production approval must be defined within the complete solution. A scope alone does not provide access control.

TECHNICAL EVALUATION

Plan Cortex as part of your solution.

Discuss your use case, knowledge-processing requirements and integration architecture with us.

Proof of concept

Evaluate the component with a defined document collection, your models and measurable acceptance criteria.

Discuss a PoC

Integration planning

Define interfaces, access boundaries, data flows and operational responsibilities for your solution.

Plan integration