Agentic Workspaces

Your team and AI.
One place to work.

Bring your people, AI-powered apps and guided processes together—with shared context, defined permissions and AI resources chosen for the work.

What makes it agentic? AI can carry out tasks: handle requests, update information and move work forward within the rules you define.

OperationsML · Maya, purchasing lead

My work

Morning, Maya.

One case needs your attention.

Needs me · 1Waiting · 3Completed · 8
Supplier onboardingCase 014

Review the new supplier

Supplier
Northstar
Your role
Purchasing lead
Next step
Review documents

Your team’s ruleA purchasing lead must confirm before approval.

Needs your confirmation

Amy

The next step, the context and the rule.
Together, where the work happens.

Illustrative workspace · Sample case and activity

A place for everyday work

Open your workspace.
Find what needs you.

Your team enters the workspace to do the work. People find the apps and processes assigned to them, see cases needing their role, and pick up where they left off.

Apps that carry out tasks

Make requests, work with information and take the actions your role allows. Amy runs supported behavior from the rules your team has defined.

How Amy runs your apps

A clear next step

Follow a case through its decisions and handoffs. See what needs you, what is waiting and what has been completed, with recorded history to look back on.

Your colleagues, within reach

Find people in the team directory and talk through the work in private, one-to-one messages. The team has a shared place to return to.

Shared context

The work comes
with its context.

Who is involved? What information matters? Which rules apply?

Your team explains these details to Amy. Amy connects them in an intent: a shared description of how the work should happen, which you can inspect and improve.

Apps and processes run from that understanding. People work with the information and actions their role permits.

See how your knowledge becomes an intent

The same supplier example

One case. Connected understanding.

  1. The information

    Supplier details and documents for review.

  2. The responsibility

    A purchasing lead reviews the request.

  3. The rule

    Approval needs the lead’s confirmation.

The person, the AI and the process work from the same agreed description.

AI does the work. On your terms.

Decide who can act.
Choose the AI behind the work.

Give people a clear place to use AI, with responsibilities and access defined before they begin.

People & permissions

Access follows the role

Workspace owners invite people and assign their app and process roles. Using the work and changing its underlying instructions are separate permissions.

AI resources

Make model choices explicit

Owners manage shared AI model connections. Deployment administrators allow or block models for modelling and app use.

Human decisions

Keep approval in the process

Define the decisions that need a person. Amy checks app permissions and supported data changes; guided processes use assigned roles and explicit confirmations.

Activity & spending

See what happened and what it used

Authorized users can inspect recorded app activity and process history. Deployment administrators review model calls, token usage and estimated costs.

Owners manage people and settings in Amy’s authoring area. Team members enter the workspace to use the apps and processes available to them.

Start with work you know

One process.
A team ready to try it.

Bring the person who knows the work. You can begin with a clear process model and decide together what AI should take on.

  1. Explain and review

    Describe the task, its exceptions and the decisions that need a person. Review the understanding Amy captures.

  2. Choose and try

    Shape an app or guided process. Configure access and AI resources, then try the behavior before making it available.

  3. Bring your people in

    Invite colleagues and assign their roles. They enter the workspace to find and use the work available to them.

A few practical questions

Before your team begins.

How the workspace fits the way you create, share and use AI with Amy.

Do we need to code or deploy each app?

Amy runs supported agentic apps directly from their intent. You configure the work, access and AI resources and try the behavior, without a separate app codebase to build or deploy. See how execution works.

Does everyone need to learn how to create an intent?

People can use the apps and processes assigned to them without authoring access. The people responsible for defining the work review and change its intent in Amy’s authoring area.

What about shared prompts?

Amy helps collaborators define and refine reusable AI instructions. Take the finished prompt to your chosen AI tool, where execution and access follow that tool’s controls. Explore shared prompts.

Does shared context include private messages?

Shared context means the relevant rules and information available to an app or process. One-to-one messages remain between the two people in the conversation; Amy does not read them.

Your team’s Agentic Workspace

Give your people a place
to put AI to work.