Management & Settings
Overview
The Management section gives tenant administrators control over the platform-wide configuration that applies across all agents and users in the tenant. It is divided into three areas:
- System Settings — platform-wide defaults covering integrations, AI models, knowledge retrieval, and usage limits.
- User Management — add, update, and remove users from your tenant.
- Account Settings — per-user preferences for notifications and monitoring alerts.
Access: System Settings and User Management require the Tenant Admin or Settings Manager role. Account Settings are available to every user.
Table of Contents
System Settings
System Settings are the tenant-wide defaults and integration parameters that apply across the entire platform. They are accessible from Settings → System Settings.
Changes to system settings take effect immediately. Settings that serve as defaults can be overridden at the agent or model level.
Consumer Integration
These settings connect Evolvable.ai to your external application — the system that consumes the API and authenticates users.
| Setting | Description |
|---|---|
| API Key Header | The HTTP header name used to pass the API key to your application (e.g., Authorization) |
| API Key Value | The secret API key sent in that header (stored securely — not shown after saving) |
| Tenant Header | The HTTP header name used to identify the tenant in outgoing requests |
| Tenant Value | The value sent in the tenant header |
| Consumer Endpoint | Webhook URL called when a document operation completes |
| Auth Endpoint | URL of your application’s authentication endpoint, used to verify user identity |
Default Agent
| Setting | Description |
|---|---|
| Default Agent ID | The agent that handles conversations when no specific agent is specified in a request |
Chat Defaults
These values are used when an agent does not have its own model configuration set.
| Setting | Description |
|---|---|
| Chat Model | The default language model used to generate responses |
| Temperature | Controls response creativity — lower values produce more predictable answers, higher values more varied ones (range: 0.0 – 1.0) |
| Max Tokens | Maximum number of tokens the model can produce in a single response |
Knowledge Retrieval Defaults
These settings define how documents are chunked, embedded, and retrieved when knowledge retrieval (RAG) is enabled on an agent.
These defaults are applied when a new embedding model is configured. You can override them per model under LLM Models → Embedding Settings.
Model
| Setting | Description |
|---|---|
| Embedding Model | The default model used to generate vector embeddings for documents |
Retrieval
| Setting | Description |
|---|---|
| RAG Enabled | Whether knowledge retrieval is enabled by default on new agents |
| Top K | Number of document chunks retrieved per query |
| Distance Type | Similarity metric used when comparing embeddings: cosine, euclidean, or manhattan |
| Graph Creation Enabled | Whether to build a knowledge graph (for GraphRAG) when documents are processed |
Chunking
| Setting | Description |
|---|---|
| Chunk Size | Target number of tokens per document chunk |
| Min Chunk Size | Minimum character length — chunks shorter than this are discarded |
| Min Chunk Length to Embed | Minimum number of characters required before a chunk is embedded |
| Max Chunks | Hard cap on the number of chunks stored per document |
RAG Enrichment
Enrichment enhances document chunks with additional metadata at processing time. This improves retrieval quality at the cost of additional processing.
| Setting | Description |
|---|---|
| Keywords Enabled | Extract and store keywords for each chunk |
| Keywords per Chunk | Number of keywords to extract per chunk |
| Summary Enabled | Generate a short summary for each chunk |
| Adjacent Summary Enabled | Generate a summary that considers neighboring chunks for additional context |
Agent Feature Defaults
Default state for optional agent features. These can be overridden on individual agents.
| Setting | Description |
|---|---|
| Suggested Options | Whether reply suggestions are shown to users by default |
| Prompt Processing | Whether input pre-processing is applied to user messages by default |
| Events | Whether agents react to and emit platform events by default |
Widget
Settings for the embedded chat widget served to end users.
| Setting | Description |
|---|---|
| Greeting Message | The message shown to users when they open the widget |
| Greeting Delay (seconds) | How many seconds after opening before the greeting message appears |
External Authentication
Used when end users authenticate through an external identity provider before interacting with agents.
| Setting | Description |
|---|---|
| External Auth Endpoint | URL called to verify or exchange an external user’s credentials |
| External User Header | The HTTP header that carries the external user’s token or identifier |
| External User Key | The secret key sent alongside requests to the external auth endpoint (stored securely) |
Message Queue (ActiveMQ)
Evolvable.ai uses ActiveMQ for asynchronous document processing and event delivery. Configure the broker connection here.
| Setting | Description |
|---|---|
| Host | Address of the ActiveMQ broker |
| Port | Port the broker listens on |
| Username | Broker username |
| Password | Broker password (stored securely) |
| Virtual Host | ActiveMQ virtual host |
| Queue Name | The queue used for internal message delivery |
Monitoring
Controls the agent and memory store used to monitor platform health. These are typically set during initial setup.
| Setting | Description |
|---|---|
| Monitoring Agent ID | The agent responsible for monitoring memory events |
| Monitoring Memory ID | The memory store the monitoring agent reads from |
| Max View Window (seconds) | How far back the monitoring agent looks when evaluating events |
Usage Limits
Enforce rate limits on the number of messages users can send.
| Setting | Description |
|---|---|
| Max Messages per Hour | Maximum number of messages a user can send within any rolling hour |
| Max Messages per Day | Maximum number of messages a user can send within any rolling day |
Leave blank to disable the corresponding limit.
Voice
| Setting | Description |
|---|---|
| Voice Enabled | Whether voice input/output is available to agents |
| OpenAI API Key | API key used for voice transcription and synthesis via OpenAI |
User Management
User Management lets tenant admins control who has access to the tenant and with what permissions. It is accessible from Settings → Users.
Roles
Each user in the tenant is assigned one or more roles that control what they can see and do:
| Role | Permissions |
|---|---|
| Admin | Full access — manage all settings, users, agents, and agencies |
| Manager | Manage agencies, agents, and knowledge content |
| Settings Manager | Manage LLM servers, models, and system settings |
| Agencies Manager | Manage agencies, agents, and guardrails |
| Users Manager | Manage tenant users |
| User | Interact with agents via chat only |
Managing Users
To add a user:
- Go to Settings → Users.
- Click Add User.
- Enter the user’s email address and select their roles.
- Click Save — an invitation is sent to the address provided.
To update a user’s roles or status:
- Select the user from the list and edit their roles.
- Toggle Enabled / Disabled to grant or revoke access without removing the account.
Note: Disabling a user immediately prevents them from accessing the tenant. Their conversation history and data are preserved.
Account Settings
Account Settings are personal preferences that control how you receive notifications. They are accessible from your profile menu.
Notification Preferences
Configure where you receive monitoring alerts:
| Setting | Description |
|---|---|
| Email Notifications | Receive monitoring alerts via email |
| Slack Notifications | Receive monitoring alerts via a Slack bot |
| Slack Bot Token | The token for the Slack bot that posts your alerts |
| SMS Notifications | Receive monitoring alerts via SMS |
| In-Platform Notifications | Receive monitoring alerts inside the Evolvable.ai interface |
You can enable multiple channels simultaneously.
Structure Summary
Settings
├── System Settings
│ ├── Consumer Integration (webhook, API key, auth endpoint)
│ ├── Default Agent
│ ├── Chat Defaults (model, temperature, max tokens)
│ ├── Knowledge Retrieval (embedding model, chunking, retrieval params)
│ ├── RAG Enrichment (keywords, summaries)
│ ├── Agent Feature Defaults (options, prompt processing, events)
│ ├── Widget (greeting message, delay)
│ ├── External Authentication
│ ├── Message Queue / ActiveMQ
│ ├── Monitoring
│ ├── Usage Limits
│ └── Voice
├── User Management
│ ├── Roles (Admin, Manager, Settings Manager, ...)
│ └── User status (enabled / disabled)
└── Account Settings
└── Notification Preferences (email, Slack, SMS, in-platform)