elvex Platform Features

Sidebar displays active Agent and conversation for easier navigation with new resource creation.

Configure web crawling with regex rules, max depth, and custom authentication headers for protected sites.

Programmatically merge, split, watermark, and fill complex PDF forms, or convert documents to print-ready PDFs.

Programmatic creation and editing of Word docs, including professional styling, redlining, and format conversion.

See who uses Agents, usage frequency, and positive/negative feedback.

Agents can perform up to 40 tool interactions autonomously before checking in.

Quickly copy settings from one Agent to another.

Provide Agents with specific knowledge like internal company jargon or acronyms.

Make Agents visually recognizable with icons or emoji avatars.

Propose rule updates to existing agents based on patterns of user feedback, with human-in-the-loop diff review.

Instruct the Agent on what to do and how to do it with editable context and rule version history.

Search for Agents by name with improved semantic search when sorting by relevance.

Find Agents tagged with your team or use case.

Scope generic sub-agents with specific toolkits (web, integrations, datasources, code) for delegated work.

Every edit creates a new agent version with one current version, so changes never destroy prior configurations.

Control whether Agents are public or private to you and shared users.

Access to firewalled Amazon instance powering Cohere Command R/R+, Anthropic Claude, and AI21 Labs Jamba models.

Support for Amazon Redshift databases including public and private connections.

Watch Agents generate answers in real-time with smooth scrolling and transitions.

Access to Claude model family including Claude 3.5, 3.7, 4 Opus, 4 Sonnet, Haiku 4.5, Sonnet 4.5, and Opus 4.5 for advanced AI capabilities.

Generate, name, and revoke API keys for programmatic access.

Artifacts are read-only resources with standard permissions; regenerating a file overwrites it in place.

PDFs and previews generated on demand for artifacts; view, download, and share from the artifacts page.

Files produced by agents become first-class, permissioned resources with their own page, search presence, and sharing.

Get helpful suggestions for rules, use cases, and conversation starters during building.

Conversational agent-building mode that proposes full agent configurations for review and one-click creation.

Full usage history including user, date, duration, model provider, prompt, response, and full LLM request with JSON rendering.

Spaces can automatically extract durable facts from conversations into context, with approval or manual modes.

Access to the firewalled Azure instance for Large Language Models with enterprise security.

Direct BigQuery integration with native SQL syntax support and secure JSON credentials.

Bookmark favorite Agents for easy discovery.

Unified Box integration for file access and management.

Guides you through the agent building process.

Agents can be designated as callable so other agents (including the home agent) can delegate work to them.

Render interactive Apache ECharts visualizations from data directly in conversations.

Conversational interface with full history, file attachments, and unlimited input length.

Support for citations with footnotes section for better source attribution in drafts.

Support for Coda documents as data sources.

Support for js, tsx, py, and java code files as data sources.

Access to the Command family of Large Language Models.

Metrics on users, Agents, and conversations to understand what's driving value.

Company-wide background that every thread can rely on, managed by admins.

Use Confluence pages as datasources with automatic updates.

Warnings when conversation context is too long and may affect Agent behavior.

Give users a prebuilt prompt for interacting with AI Agents.

Search for specific content within conversations with highlighted matches.

Export tables from drafts for use in Excel or Sheets.

Create and manage custom roles to define precise access controls for your team.

Full dark mode support with improved rendering consistency.

Analyze raw datasets to extract insights, generate trends, and create visualization charts automatically.

Clear error indicators when file reading fails with real-time progress updates.

Connect AI interactions to company data with support for files, folders, and live sync.

Status display for datasource processing progress.

Add branching logic to flows for conditional execution.

Per-message selector between fast (quick, no code), thoughtful (extended thinking), and deep (full capabilities) modes.

Create, edit, and refine persistent content in a dedicated workspace separate from chat, with formatting retention.

Connect Dropbox files and folders to Agents with automatic sync.

Edit conversation names for easier organization and management.

Receive email notifications when granted access to resources.

Data is encrypted in transit and at rest using enterprise-grade standards (AWS-hosted).

Animated placeholders, tab-to-complete suggestions, and improved resizing.

Model selector displays multiple models with descriptions; legacy models in submenu.

Text extraction from PDFs including images and improved table parsing.

Code runs in a secure, isolated container that is destroyed immediately after use, ensuring zero data persistence.

Drag and drop files into conversations with support for csv, doc, docx, htm, html, jsonl, md, pdf, ppt, pptx, rtf, txt, xls, xlsx.

See the time and date for when a given file was updated.

Automatically detect and redact credit card numbers.

Use flows as nodes within other flows for complex automation scenarios.

Add attachments to flow runs at execution time.

Warnings when deleting entities used in flows to avoid disrupting automations.

Support for loops within flows for batch operations and iterative tasks.

Delete past flow runs for better data management.

Complete run history for flows including drafts and step-by-step results.

Execute multi-step workflows with a user-friendly interface supporting different models, tools, datasources, decision steps, and loops.

Tag flows for easier organization and discovery.

Query an elvex Agent to generate text responses via API.

Spin up scoped worker agents with only the toolkits granted for the task at hand.

Connect Google Drive files and folders to Agents with automatic sync.

Access to Gemini model family including 2.5 Flash, 2.5 Pro, and Flash Lite for versatile AI interactions.

Latest OpenAI image generation models with transparent background support and multi-image editing.

Compliant with HIPAA standards for handling protected health information (PHI).

Starting point for engaging with elvex with direct LLM interaction and guided recommendations.

Generate interactive HTML files with charts and UI components rendered in an isolated viewer.

Consequential integration actions require explicit user confirmation in the thread before they run.

Upload, paste, and use image file attachments including screenshots in conversations.

Edit previously generated images with prompts, including compositing multiple source images together.

Generate images directly within elvex using GPT Image 1, DALL-E 2, and DALL-E 3 models.

For advanced users, fine-tune the exact prompt engineering powering the Agent.

Agents take action in the world (send email, update Salesforce, create Notion pages) via callable integration actions.

Admins control which integrations are authorized; users connect on demand when first using an authorized integration.

Let LLMs determine what and how to search through broad sets of data.

More efficient handling of large payloads from integrations.

Files imported from cloud storage are updated automatically with status tracking.

Switch existing apps to new models in seconds without breaking functionality.

Standardized prompting layer that ensures the same prompt works across different models.

Dates respect local timezone settings throughout elvex.

Filter logs by Agent for targeted analysis.

Add Model Context Protocol servers as integrations; their tools appear as integration actions (MCP_ prefix).

Guided setup for connecting MCP servers, including OAuth app registration, scopes, and redirect URI configuration.

Edit any message in a conversation, creating a fork with both versions accessible.