Bring a teammate up to speed on a project
Give a new colleague a starting point in the shared project record, with links back to the evidence.
Interactive sample, with fictional data and scripted agent output. No agent runs and no workspace data is read or changed.
Reduce the time it takes to train clinic staff.
Start with a small pilot, then review evidence before expanding.
We need a training pilot before we can choose the rollout.
The brief, company playbook and finalized transcript supply the context.This is a simulation with sample data.
Inspect the sample record · Northfield Health
Sources
- Project brief
- Reduce the time it takes to train clinic staff.
- Company playbook
- Start with a small pilot, then review evidence before expanding.
- Finalized interview transcript
- We need a training pilot before we can choose the rollout.
Project onboarding note
- Goal
- Reduce the time it takes to train clinic staff. · Project brief
- Approach
- Start with a small pilot, then review evidence before expanding. · Company playbook
- Proposed task
- Draft the training pilot plan · Sam Lee · Finalized interview transcript
- Needs an answer
- Which audience should the draft prioritize?
Try it in your workspace
In your workspace, connect a member’s MCP client, make the source records available to that member, and ask for a draft. Shared-chat agents need their own workspace setup. Starlings provides the context and tools; your chosen agent produces the result.
Connect Claude over MCP · Other agents and connection paths
A request to adapt
Prepare an onboarding note for a teammate joining this project. Explain the goal, current approach and next action using the brief, playbook and latest transcript. Cite sources and flag anything that needs an answer.
Check that the teammate has access to the linked records. A summary or a link does not grant access.
Read the product instructions · See this in your team’s work
What is available today
These are sample agent workflows built from existing capabilities, not built-in one-click automations. Agent output depends on the client, model, instructions and records it can access.
| Capability | Mac | iOS | Web |
|---|---|---|---|
| Docs: the corpus, read and written from Starlings | Works | Works | Works |
| Projects and leads, with their tasks and issues | Works | Works | Works |
| Finalized transcript and AI summary delivery | Works | Works | Works |
Explore another job
Prepare a client-call brief
Walk into the next client call with the goal, the last conversation and open questions in view.
Turn a client call into follow-up work
Use the finalized transcript to propose tasks with an owner and a project, then review the hand-off.
Draft a proposal from your company playbook
Bring the client brief and your established way of working into one proposal draft.
Synthesize interviews with source quotes
Start a research memo from the interview record, keeping the evidence beside the interpretation.
Give a blocked agent a human answer
See how one assigned question appears in chat and Inbox, and how one answer resolves both views.