Choose the LLM provider used by every Nessyflow AI agent. Applied on the next iteration — no restart required.
Connect Nessyflow AI to your external tools — Linear, Slack, HubSpot, Stripe, ProfitWell and more — via one-click OAuth or an API key. Tokens are encrypted at rest and used by every agent.
Manage integrations →Loading…
Keys are stored in .kirasa/config/provider.json (chmod 0600) alongside the selected provider. Environment variables, when set, always win over persisted keys — so you can still manage credentials from .env if you prefer.
Control how frequently the CEO agent runs its autonomous work cycle. Shorter intervals (1-2 min) mean faster task dispatch; longer intervals (10-30 min) reduce resource usage.
Changes apply on the CEO's next heartbeat cycle (≤ 5 minutes). No container restart required.
Control how many agents can run simultaneously. Changes to semaphore count take effect immediately in the 3D view; pool and lane changes apply on the next CEO cycle.
Max simultaneous agent executions (1–32)
Reusable SDK subprocess pool (1–32)
Per-domain parallel agents (1–16)
Overrides are stored in .kirasa/company/concurrency.json. To make changes permanent across restarts, update KIRASA_MAX_CONCURRENT, KIRASA_SESSION_POOL_SIZE, and KIRASA_LANE_CONCURRENCY in your .env.