Orbit
Agentic enterprise intelligence workspace
Brings together the fragmented information, conversations, tasks, meetings and project signals enterprise teams work with every day — and turns them into decisions and next steps.
Public demo environment — use sample data only; do not enter confidential, personal or production information.
01Context
Enterprise teams spend a large share of their day reconstructing context — what was decided in a meeting, what is due today, what is blocking a project. The information exists, but it is spread across meetings, email, messages and project tools.
The challenge: Explore how specialised agents, persona-aware reasoning and MCP-based tool connectivity can turn fragmented enterprise information into actionable intelligence.
02Approach
Catch Up — understand
Turns meeting conversations into decisions, action items, risks, priorities and follow-ups.
Daily Plan — prioritise
Brings calendar, messages, tasks, meetings and deadlines into one contextual plan for the day.
Readiness — execute
Coordinates specialised agents to investigate project risks, dependencies and readiness.
Observe everything
Tracks LLM calls, tool calls, latency and cost, so the system can be tuned like any production service.
03Architecture
04Key decisions
Specialised agents over one generalist
- Why
- Summarising a meeting, planning a day and assessing project risk are different jobs.
- Trade-off
- More agents means more orchestration — handled by a dedicated layer, in exchange for focused, testable agents.
Persona-aware reasoning
- Why
- A manager, an engineer and a programme lead need different things from the same information.
- Trade-off
- Personas add configuration, but make the output relevant to the person reading it.
Observability from day one
- Why
- Agentic systems fail quietly — through latency, cost or a bad tool call.
- Trade-off
- Instrumenting LLM calls, tool calls, latency and cost adds work up front and makes the system tunable.
05Outcomes
- A working public demo of three agentic experiences.
- A reference architecture for persona-aware, MCP-connected enterprise agents.