The Philosophy Behind Company Settings
elvex's company settings are designed around the principle of centralized control with user flexibility. Rather than restricting users, these settings establish organizational defaults and security baselines while allowing individual customization where appropriate.
Think of company settings as your organization's "foundation" - they provide the base configuration that ensures consistency, security, and branding across all user interactions with elvex, while still allowing users to build upon this foundation for their specific needs.
Company Branding and Identity
Logo Upload and Display
Maximum file size: 500 MB
Maximum file display: 38px by 140px.
We recommend you upload a logo that is at least twice those pixel dimensions (2x).
AI Provider Configuration - Understanding Default AI Providers
When you set a default AI provider, you're establishing the provider and model for the elvex Home Assistant, as well as the fallback option for every new custom assistant created by your organization.
This doesn't restrict users from choosing other providers for specific assistants, but rather ensures there's always a reliable option available.
β
Image Generation
elvex provides image generation through your OpenAI or Google Gemini API license. By selecting an image generation provider you are allowing your users to have access to image generation in the Home Assistant and in any custom assistant that they build.
β
PII Redaction in Assistants
The "Redact PII in assistants" setting represents elvex's approach to privacy by design. When enabled, this feature:
Automatically identifies and masks personally identifiable information in assistant conversations, including:
Email addresses
Phone numbers
Social Security numbers
Credit card numbers
We do not recommend checking this box at the administrator level. Our strong opinion is that they should be implemented at the assistant level and not at the organizational level.
β
When you centrally redact you also impact the assistant's ability to do anything important with things like phone numbers and emails to users. You are making the model less able to be intelligent.
This is what happens when you click that button:
Let's say you have meeting notes from a meeting with 5 people.
You want the AI to use those notes to create various outputs.
We preprocess the information before it gets sent to OpenAI, Anthropic, Google, etc., and replace all names, emails, and similar data with XXXXXXXXXXXXXXXX.
There is no longer any way for the model to distinguish who said what in these notes.
How Settings Cascade Through Your Organization
Understanding the hierarchy of settings helps you make informed configuration decisions:
Company settings establish the baseline
Individual user preferences can override where permitted
Assistant-specific configurations take precedence for that particular assistant
Conversation-level settings may temporarily override others
This cascade ensures that security and compliance settings (like PII redaction) remain enforced while allowing flexibility for productivity features (like AI provider choice).
