The routing brains behind the OS. Every request goes to the cheapest agent that can handle it; top tier is restricted so day-to-day questions never burn premium credits.
Repeatable sequences that run exactly the same every time.
D-Flows run as deterministic pipelines — no LLM randomness. Wire up data sources, transformations, and outputs into reliable automations.
Structured data store — like a spreadsheet that agents and flows can read and write. Updated via polls, cron jobs, or manual entry.
| Client | Contact | Status | Monthly | Last activity |
|---|---|---|---|---|
| Diamond Corp | Sarah Mitchell | ● Active | £1,200 | Today |
| Hartley & Sons | James Hartley | ● Active | £850 | Yesterday |
| Apex Cleaning | — | ○ Paused | £0 | 2 weeks ago |
Tables are accessible by chat and flows. Poll external sources on a schedule, or let agents write directly. Data stays structured and queryable.
Monitor and orchestrate agents, flows, and multiflows. Supervisors watch what's running and can trigger chains on schedule.
Agents registered here can be monitored, paused, and restarted from one place.
Flows promoted to supervised mode get health checks, retry logic, and alerting.
Chain agents and D-Flows into multi-step sequences. Trigger by cron, webhook, or manually.
Connect third-party services so agents and flows can act on real data. Each integration exposes tools that agents can call.
Token costs per call, per tier, per client.