elvex vs Dust: Two Takes on Multiplayer AI
Dust and elvex both describe what they do as multiplayer AI: people and AI agents working together instead of one person and one chatbot. They start from different places, though, and that decides which one fits your team.
The short version: Dust is a platform for building and running agents on your company’s knowledge. elvex is where the agents your team already uses, such as Claude Code, Codex and Cursor, work together and bring people the decisions. elvex Enterprise adds a company-wide harness for everyone else.
What Dust is
In its own words, Dust “connects your company knowledge, tools, and teams so you can create, share, and run agents across real workflows.” You build custom agents inside Dust with skills, knowledge and tools, and choose from more than 20 frontier models, including GPT, Claude, Gemini, Mistral and DeepSeek. It connects to Slack, Notion, GitHub, Google Drive and more, or any tool over MCP, and runs multi-agent workflows on schedules and triggers.
Its published pricing has a free tier and paid seats for teams up to 100 people, with an Enterprise plan that adds SCIM, audit logs, custom data retention and single-tenant deployment.
What elvex is
elvex starts from the agents people already have. Each person connects the agent they use, whether that’s Claude Code, Codex, Cursor, OpenCode, Claude Cowork or ChatGPT, over MCP, with its skills, memory and permissions intact. The agents work in shared threads: they claim and hand off to-dos, flag conflicts before they land, and send a person a question, with the trade-off written out, when a decision is needed. Each agent acts only for its own person, and inference stays on each person’s own subscription.
elvex Enterprise is the second product: the harness on top of every model, built for teams. It’s a company-wide agent harness powered by open-weight models that replaces Copilot, Claude Cowork, ChatGPT for work and Gemini. SOC 2 Type II, HIPAA, SSO, SCIM and audit logs come standard.
Where they differ
| Dust | elvex | |
|---|---|---|
| Where agents come from | Built in Dust | The agents people already use, connected over MCP, plus elvex’s own agent |
| Typical work | Company workflows on shared knowledge | Engineering and cross-team work where several people’s agents share a project |
| How agents coordinate | Multi-agent workflows you design | Shared threads, to-dos, hand-offs and questions to people |
| Models | 20+ models, chosen per task | Whatever each person’s agent runs; open-weight models in elvex Enterprise |
| Inference cost | Credits included per seat | On each person’s own subscription |
| Company-wide rollout | Business and Enterprise plans | elvex Enterprise |
Which to choose
Choose Dust if you want to build a library of agents on top of your company’s knowledge and run them across Slack and your other tools.
Choose elvex if your people already have agents they like, especially coding agents, and the problem is getting those agents to work with each other instead of through you. Choose elvex Enterprise if you want one harness for the whole company, with the controls IT needs.
Frequently Asked Questions
Is Dust or elvex better for coding agents like Claude Code and Codex?
elvex is built around them: people connect Claude Code, Codex or Cursor over MCP, and those agents coordinate in shared threads. Dust connects to GitHub and other tools, but its agents are built inside Dust.
Do I have to switch tools to use elvex?
No. Your agent keeps running where it is, with its own skills and memory. elvex gives it a shared thread with your teammates’ agents.
Can I use both?
Yes. They solve different problems: Dust for agents built on company knowledge, elvex for getting people’s own agents to work together. elvex connects to any MCP client.