Company Management

Each company in Queria represents a fully isolated environment. Documents, users, conversations and configurations are separated to ensure maximum confidentiality in the multi-tenant architecture.

Create a new company

From the admin panel, select New Company and fill in the requested information.

Required fields are company name and type. Description and logo can be added later.

Company types

Queria supports two operating modes, each optimized for a specific use case.

Document

Designed for organizations that need to upload, manage and query internal documents. Supports PDF, DOCX, XLSX and other common formats. Documents are processed by the Cog-RAG architecture for semantic indexing and intelligent search.

Main features:

• Document upload with automatic text extraction
• Built-in OCR for scanned documents and images
• Curated knowledge base for permanent content
• Temporary documents tied to individual conversations

Online

Designed for companies that operate with web content. Lets you index web pages, monitor sites and build a knowledge base from online sources.

Main features:

• Automatic web scraping with scheduling
• Indexing content from specific URLs
• Automatic content summarization with AI
• Source update monitoring

Company configuration

After creation, each company can be customized in detail.

Basic information

• Name: displayed name in the interface and in widgets
• Description: short description of activity or purpose
• Logo: identifying image, also used in embeddable widgets

System prompt

The system prompt defines the AI assistant behavior for the whole company. Here you can specify tone, domain of expertise, response rules and constraints. Every conversation inherits these instructions, unless overridden at topic level.

API key

Generate an API key to integrate Queria in external applications, widgets or automations. The key is associated with the company and respects all configurations and subscription limits.

Notice

The API key is shown only once at generation time. Store it safely. You can regenerate it at any time, invalidating the previous one.

Sector and sub-sector

From v3.5.0 every company is associated with a sector (LEGAL, FINANCE, HEALTHCARE, ...) and optionally a sub-sector. The choice determines:

• The default RAG preset (chunking, tags, domain terms).
• Domain terms applied to ingestion and retrieval.
• Suggested external sources (e.g. LEGAL -> Legal Sources).
• Access and confidentiality policies.
• The available AI Constructor, if any (e.g. TOURISM -> Tourism Constructor).

For the full mapping, see Sectors & YAML.

Sector change

If you change sector on a company with already-ingested documents, run a sector-fields backfill on existing chunks to stamp the new domain terms. See Ingestion DSL > Reingest.

Tags

Free labels associated with the company (e.g. enterprise, pilot, EU-East). They are used:

• As an admin-side filter (cross-tenant for SYSTEM_ADMIN).
• By the bot planner to disambiguate specific behaviors (e.g. enterprise vs SMB tone).
• To group companies in consumption dashboards.

Tags are entered separated by enter or tab. They can be edited at any time without impacting already-ingested documents.

Domain terms (company dictionary)

Terms specific to your company that don't fit the generic sector domain: internal acronyms, product names, technical jargon. They are merged with the sector's domainTerms when the system evaluates chunk relevance.

Examples:

• MOL-adj (adjusted EBITDA, internal acronym).
• Product Alpha, NACE code 62.01.
• Customer acronyms.

Add them in the company sheet > Domain terms, one per line. Additions and removals require a sector-fields backfill to apply to existing chunks.

AI pipeline (canvas DSL)

Unlike previous versions, the AI pipeline (chat, search, ingestion) is no longer configured in the company sheet. It must be edited in the respective DSL canvases:

• Chat: Chat DSL (Canvas)
• Ingestion: Ingestion DSL
• External integrations: OAuth & Service Canvas

The company inherits the system canvases; you can duplicate them and assign custom versions per topic.

Multi-tenant isolation

Isolation is guaranteed at all levels:

• Data: every company has separate vector collections and stores
• Users: no user can access data from other companies
• Pipelines: every company inherits or customizes the DSL canvases (chat, ingestion, service) independently
• OAuth connections: external connections are per (user, company, service); no cross-tenant leakage
• Consumption: token and channel monitoring is separated per company

Company deletion

Deleting a company is an irreversible operation that permanently removes all associated data: documents, conversations, users, configurations and vector indexes.

Irreversible operation

Before proceeding, make sure you have exported any needed data. The system requires double confirmation and typing the company name to proceed.

Best practices

• Choose the correct company type from the start: changing later is not supported
• Configure the system prompt before inviting users
• Generate the API key only when needed and limit its distribution
• Assign a dedicated Company Admin for each company
• Use descriptive names to ease management with many companies