Every agent on your team learns from every other. Namespaces, RBAC, pub/sub — so knowledge compounds instead of siloing.
Each agent reads and writes to a central memory. Namespaces and RBAC decide what each agent can see, write, or consolidate.
Every team we've onboarded ends up with a variant of the same shape: many specialized agents, one shared memory, scoped permissions.
Support, success, and billing agents all see the same customer context. A refund handled in billing is visible in the next support ticket automatically.
Literature, synthesis, and critique agents share findings on a pub/sub channel. One agent's discovery becomes another's citation in seconds, not weeks.
Five developers, five personal copilots, one shared project memory. Architectural decisions, debugging sessions, and code review context all compound.
Built into the memory primitive from day one, not bolted on later.
Three API calls: create a namespace, add agents, scope permissions. Everything else is automatic.
# Create team namespace curl -X POST https://api.remlabs.ai/v1/namespaces \ -H "Authorization: Bearer $REM_KEY" \ -d '{"name":"acme-support","description":"Customer success swarm"}'
# Add agents (one key per agent) curl -X POST https://api.remlabs.ai/v1/agents \ -H "Authorization: Bearer $REM_KEY" \ -d '{"namespace":"acme-support","name":"support-bot"}' curl -X POST https://api.remlabs.ai/v1/agents \ -H "Authorization: Bearer $REM_KEY" \ -d '{"namespace":"acme-support","name":"billing-bot"}'
# Scope billing-bot to read cx, write billing curl -X POST https://api.remlabs.ai/v1/agents/billing-bot/scopes \ -H "Authorization: Bearer $REM_KEY" \ -d '{"read":["cx/*"],"write":["billing/*"],"consolidate":["billing/*"]}'