| AI models | Anthropic, OpenAI, Google, xAI and open-source, chosen per agent | Anthropic models only |
|---|
| Mix models in one task | Yes. For example, a Claude writing agent using an OpenAI or Google image tool | Claude models only |
|---|
| Image and video generation | Built in through OpenAI Image, Nano Banana Pro, Veo and Grok Imagine | No native image-generation model |
|---|
| Seats | Unlimited on every plan, including Free | Per seat. Team plan from US$20 per member per month, up to 150 members |
|---|
| People and agents together | Group chats with colleagues and several agents; agents can join in without being @mentioned | Individual chats and shared projects; Claude in Slack |
|---|
| Scheduled and background work | Scheduled jobs that belong to the workspace | Scheduled Cowork tasks that run in the cloud |
|---|
| Step-by-step workflows | Visual workflows with triggers, loops, spreadsheet rows and approval steps | Agentic tasks; no visual workflow builder |
|---|
| External triggers | Inbound webhooks start agents and workflows | Not available |
|---|
| Where your team reaches agents | Web, desktop and mobile apps, plus Slack, Microsoft Teams and Discord | Claude web, desktop and mobile apps, Chrome side panel and Slack |
|---|
| Chat on your website | Agent Widget for Web: a chat pop-up on your site where visitors talk to your agent, added with one snippet | Not available |
|---|
| Integrations | 140+, including Xero, HubSpot, Microsoft 365, NZBN and LINZ, plus any MCP server | Connectors directory, including Microsoft 365, plus custom MCP connectors |
|---|
| Approvals and audit | Approval per action, audit log of people and agents, SOC 2 Type II | Admin tool permissions; audit logs and OpenTelemetry export on Enterprise |
|---|
| Seeing the value | Hours saved and estimated ROI by agent, person and workspace | Usage analytics |
|---|
| Coding | Agents write and run code in their own VM. Autohive isn't a coding tool | Claude Code, a dedicated tool for software teams |
|---|