Skip to main content
Arelis AI

Arelis Hub features

Everything in Arelis Hub, from chat to audit logs

What each part of the platform does, in plain terms. Every feature shares the same access rules, usage records and audit log.

Start free trialBook a demo

Chat

One chat workspace for every model your administrators enable, with the controls teams expect from a production tool.

Chat with any enabled model

Streamed answers from the model that fits the task, without leaving the conversation.

  • Token-by-token streaming responses
  • Switch models within a conversation
  • Markdown, code highlighting and LaTeX rendering
  • Attach images, PDFs, Word documents and spreadsheets

Conversation management

Keep, find and reuse work instead of starting from a blank prompt.

  • Searchable history with automatic titles and archiving
  • Regenerate, edit and resend messages
  • Pin and share individual messages
  • Export conversations as JSON or Markdown
  • Token usage and cost estimate per message and session

Knowledge bases

Upload the documents your teams rely on and let models answer from them, with sources attached to every answer.

Knowledge bases

Document collections scoped to the people who should see them.

  • Upload PDF, DOCX, TXT, CSV, Markdown and RTF files
  • Uploaded files scanned for personal data
  • Share with a team, a department or the whole organization

Retrieval-augmented chat (RAG)

Answers that point back to the passage they came from.

  • Semantic and hybrid (keyword plus vector) search
  • Source citations with every answer
  • Query several knowledge bases in one conversation
  • Configurable similarity thresholds

AI agents

Agents for recurring business tasks, made available the same governed way as models: requested, approved and scoped.

Agent catalog

Ready-made agents for defined business tasks, added to your plan as needed.

  • Pre-built agents for research, support, sales, compliance and more
  • Customize instructions and connect your own knowledge
  • Usage tracked per agent and per team

Governed roll-out

Nobody switches an agent on for the whole company by accident.

  • Request access with a business justification
  • Administrator approval before an agent goes live
  • Assign to a team, a department or the organization

Models & Prompt Lab

Administrators decide which models each team can use. Teams compare models on their own prompts before they commit to one.

One gateway to five providers

OpenAI, Anthropic, Google, Meta Llama and Mistral AI models, managed in one place.

  • Choose which models each team can use
  • Add, update and retire models centrally
  • Model fallback and routing on Professional and Enterprise plans

Prompt Lab

Test a prompt across models before it reaches production.

  • Prompt templates with variables, system prompts and examples
  • Run one prompt against several models side by side
  • Compare response time, token usage and cost
  • Versioned, shareable template library

Governance

The controls that let you show how AI is used in your organization: what models produce, what data they see and who did what.

Content safety and personal data

Visibility into what goes into models and what comes out.

  • Toxicity metrics on model outputs, tracked per model over time
  • Detection of personal data such as names, email addresses and card numbers
  • Redaction and a review queue for flagged content
  • User feedback on answers, by model and by team

Evaluation and audit

Evidence for model choices and an audit trail for every change.

  • Pointwise and pairwise model evaluation with a judge model
  • RAG quality metrics, including a hallucination score
  • Audit logs for sign-ins, role changes and configuration
  • Register AI systems and record their EU AI Act risk classification

Teams & access

Access follows your organization chart, and sign-in follows your identity provider.

Organization structure and roles

Permissions that match how your company is actually organized.

  • Organization, department and team hierarchy
  • Role-based access control at platform, organization and team level
  • Resources scoped as private, team, department or organization

Enterprise sign-in

Single sign-on with the identity provider you already run.

  • SAML 2.0 and OpenID Connect single sign-on
  • Domain verification and enforced SSO per domain
  • Just-in-time provisioning and role mapping from identity provider groups
  • Optional multi-factor authentication

Projects

Projects keep the conversations, knowledge bases, prompts and files for one piece of work together.

Project workspaces

A shared place for one initiative, with its own members and resources.

  • Link knowledge bases, chat sessions, prompt templates and files
  • Personal, team, department or organization visibility
  • Members, mentions, comments and activity notifications
  • Archive and restore projects when work finishes

Usage & cost

See where AI spend goes and set limits before an invoice surprises anyone.

Usage analytics

One view of consumption across every provider.

  • Token usage by user, team, model and agent
  • Interaction counts and adoption by team
  • Agent usage analytics

Cost control

Spend allocated to the teams that incur it.

  • Credit balance and full transaction history
  • Cost allocation per team and department
  • Low-balance and budget threshold alerts
  • Exportable period reports

MCP servers

Model Context Protocol (MCP) servers connect models to your data and tools. Administrators decide which servers exist and who can use them.

Managed MCP servers

Tool access configured once, by an administrator, and tested before release.

  • Register servers for search, databases, file systems, APIs and custom tools
  • Configure each tool: parameters, timeouts, retries and fallbacks
  • Test the connection and each tool before activating

Team assignment

The right tools for the right teams, with personal credentials where needed.

  • Assign servers to specific teams, with an access request flow
  • Per-user API keys and tokens
  • Invocation counts, error tracking and cost allocation per server

Keep exploring

More about Arelis Hub

Try every feature for 14 days

Starter and Professional include a free trial with full access to the plan's features. No credit card required.

Start free trialBook a demo