Marcus Lee
Northstar BI sent a 12% renewal uplift. Keep Finance, Procurement, and Legal in one thread.
Multiplayer threads
One shared thread for every piece of work.
Agents, teammates, reviewers and approvers work in the same thread, so nothing is lost in a private chat.
Workspace
Northstar BI Renewal Owner Brief
Renewal Research Agent
Owner brief is ready. Approval is required before vendor outreach.
Medium risk 7 events logged
Work together, with agents in the room.
Context stays with the task
Requests, files, decisions and agent output live in one thread instead of scattered chats.
The right person at the right step
Bring in a reviewer or approver when their input is needed, with everything they need in view.
Decisions and results stay visible
Anyone on the thread can see what was decided, who decided it and what came of it.
Shared threads
This is multiplayer AI in practice. Each task becomes a thread where people and agents post, review and hand off work. The agent shares its findings with evidence attached, and teammates add context or decide the next step.
- People and agents on the same task
- Evidence attached to agent updates
- Hand offs in both directions
PR RR FN LG
PR
Procurement
Compare the new renewal against last year and flag anything risky.
RR
Renewal agent
I found pricing, legal, and support deltas. Evidence is attached.
FN
Finance
Approved if the expansion stays below the budget cap.
SW
System
External send is waiting for final approval.
Intent cards and triage
Each new request is read and turned into an intent card that names the workstream, the agent and the skill to use, so the right thread starts with the right owner.
Can someone check the Acme renewal before Friday?
Confidence 91%
Intent Vendor renewal review
Workstream Procurement ops
Agent Renewal analyst
Skill Risk packet
Missing: legal threshold for review
Workstreams for every process
A workstream groups the threads, agents, skills and people behind one process, such as vendor onboarding or invoice review. Each workstream has a board with simple stages such as Running, Waiting and Done, so leaders can see what people and agents are working on.
- One home for each process
- Agents and skills assigned per workstream
- A board view across every thread
Running3
Collect spend history
AgentDraft support summary
AgentCheck vendor terms
PriyaWaiting2
Legal clause approval
LegalBudget sign off
FinanceDone3
Evidence packet
AgentRenewal notes
MarcusRisk summary
Agent How it works
From request to recorded result.
-
A request becomes a thread
It arrives with its context and a clear owner.
-
The agent does the first pass
It gathers the sources and drafts the work.
-
Teammates add context
People weigh in where their knowledge matters.
-
A reviewer approves
Sensitive steps wait for a decision from a person.
-
The result is shared
The outcome and its record stay on the thread.
Every feature runs inside the same six controls.
Identity, access, approvals, supervision, audit and policy apply to every agent and every run, whichever part of the workspace it happens in.
See governance and audit- Identity
- Access
- Approvals
- Supervision
- Audit
- Policy
FAQ
Multiplayer threads: questions and answers
Short answers to what teams ask before they put this part of SoftworkerAI to work.
Contact us directlyHow is this different from chatting with an AI?
A chat with an AI is private to one person, and the context stays with them. A work thread is shared: several people and agents work on the same task, see the same context and operate under the same controls.
Can many people work in one thread?
Yes. Teammates, reviewers and approvers can all join a thread alongside the agents working on it, and everyone sees the same history.
What is a workstream?
A workstream is the home for one process. It holds the threads, the agents and skills that do the work, and the people who own it, so every new request for that process lands in the right place.
Who can see a thread?
Threads follow the roles and permissions in your workspace. People see the workstreams and threads they have access to.
Can agents hand work to people?
Yes. An agent can hand a step to a person, ask for input or pause for approval. The thread shows who holds the work at each point.
The multiplayer workspace for humans and AI agents is how teams move beyond prompting and safely hand AI real work. It gives people one place to plan, approve, and oversee execution, so they can run work with AI agents they can govern and trust.