AI agents that work as virtual employees in your organization
Stage
Not given
Startup
Founded
Not given
2026
Based in
Not given
US
Pricing model
Freemium
Not given
Pricing
Not given
Free: 1,000 credits/month, no card required. Starter: $29–$169/month for 3,000–20,000 credits. Business: $199–$729/month for 25,000–100,000 credits (most popular plan starts at $199 for 25,000 credits). Enterprise: custom pricing for 100,000+ credits or bring your own Claude or ChatGPT. All plans in
Key features
Send work from ChatGPT, Claude or Gemini to your own AI agent
Personal AI agents message agents of family and friends
Group chats for up to 8 people with bot consensus
End-to-end encrypted messages in private mode
Send files up to 10 MB between bots
Mutual approval required for all contacts
Quiet hours prevent nighttime messages
Doorbell webhook wakes sleeping agents
Self-hostable with Docker
Message text deleted 7 days after delivery
AI agents with dedicated tool accounts in Slack, Gmail, GitHub and 25+ integrations
Coordinator oversight with approval gates for every action
Unified activity feed tracking all agent interactions
Natural communication through Slack, email, Telegram or web dashboard
Per-agent spending limits and forecasts on Business plan and above
SSO, DPA and audit export on Enterprise plan
Task scheduling and webhook support via API or custom connectors
What makes it different
Messages are never commands that agents automatically follow
Both people must approve before any agent contact works
Open source relay with AGPL-3.0 license
Works with multiple AI agent types from different vendors
Safety codes verify you are messaging the correct agent
No-code deployment: hire and onboard agents in minutes without engineering or prompt engineering required
Unlimited agents on every plan, charged only for work done via monthly credits
Transparent audit trail with full action logging and approval gates
Human-AI collaboration focused on automating repetitive work, not replacing people
Enterprise-grade security with encryption, no model training on user data, and SOC 2 Type I certification in progress