Workstreams Setup
A workstream is a plan: tracks, phases, deliverables and what has to be true for each to count as done. Workstreams is the part of OpenX that will hold those plans for your team, so that you and every AI you work with read the same plan. Today it is an experimental prototype of its front door: a menu, a setup window and a Team window. Nothing is built behind them yet, and the setup is kept in your browser (see Kept in this browser, for now).
Who sees it
The Workstreams menu sits in the top bar, to the right of fAI. It shows only for people who hold the
Workspace.Experimental right. Without the right the menu is not there at all; it is never greyed out. The
right comes with a plan's license, not from being an owner. See
Permissions.
The menu holds two items:
- Workstreams > Setup before setup has run. It reads Workstreams > Settings once it has, and opens the same window either way. The word changes the next time the menu redraws, so a reload always shows it.
- Workstreams > Team, which shows beside it while the prototype is tried out, whether or not setup has run. See Workstreams > Team.
Set it up
Workstreams > Setup opens Workstreams setup. Like the Cloud windows, it opens as a panel you can move and dock. At the top, a pill says where setup stands: NOT SET UP before you have saved, then ON or OFF from what you last saved, never from what you are typing. Under it are three cards:
- Use Workstreams in this workspace - a switch, on to begin with.
- How your workstreams AI behaves - the Description and AI notes boxes, open and first.
- Where workstream data lives - the store choice, below.
Press Complete setup to keep it. Cancel puts back what you last kept. If you chose your own store and no graph Data Store can be picked, the button stays off with a line saying why (see below).
After that, the window is titled Workstreams settings, its button reads Save, and the pill reads ON. Switching Workstreams off and pressing Save turns the pill to OFF but leaves setup done: the menu still reads Settings, and your description and notes are kept for when you switch it back on.
Why the description and AI notes come first
They are the one part of setup that changes how Workstreams behaves, so they come before anything about storage. Every AI reads the description and the notes before it works on one of your workstreams here:
- The Description says what your workstreams are for.
- The AI notes say how you want AIs to work on them, one short thought per note.
They are the same two boxes you find on a node, a Cloud record and the workspace itself. Press Add note for each note. A blank note is dropped when you save. For what to write in them and why they never grant a right, see Describe Your Nodes.
Where workstream data lives
Two choices, drawn as two wide rows. You pick one:
| Choice | What it is |
|---|---|
| Fathym shared store (low cost) | Run by Fathym and kept apart by workspace. Nothing to set up. Chosen to begin with. |
| This workspace's own graph data store | A Data Store of this workspace with the Gremlin (graph) API, in your own Azure. Only those are offered. |
Your workspace does not need a Data Store to use Workstreams. If it has no Data Store with the Gremlin API, the second row is greyed out and says so, and the shared store stays chosen. A SQL Data Store is never offered. Data Stores are added in Cloud > Data Stores, which shows once the workspace has its private landing zone; see Data Stores.
With the second row chosen, a list shows the workspace's graph Data Stores, the first already picked; change it if you want another. Complete setup and Save stay off, with a line saying why, while there is no graph Data Store to pick.
The window also says that the store is connected in a later step. For now, your choice is recorded and nothing is read from or written to either store.
Kept in this browser, for now
The setup is prototype state. Pressing Complete setup or Save keeps these in your browser's own storage, one set per workspace: whether Use Workstreams in this workspace is on, the Description, the AI notes, which store you chose, and when setup first completed.
So, for now:
- It is not shared. Another browser, another device or a teammate sees Workstreams > Setup again.
- It never reaches the OpenX API, and nothing is created in Azure.
- Clearing site data forgets it. Setup will read as not done.
- If your browser blocks site storage, the window says nothing was kept. Allow site storage and try again.
- On a page that has just loaded, the menu may read Setup for a moment before it reads your saved Settings.
The window says this at the top, under its title. A saved Workstreams record in the workspace, shared by the team, is meant to replace this prototype state. Until then, treat what you enter as a way to try the screen and give feedback, not as a record.
Workstreams > Team
Workstreams > Team opens Workstreams team. It is not who is on the team; that stays in Team Members. It is who each member is in the system.
What it shows depends on your rights. With Workspace.Experimental and Workspace.Experimental.UX
together, it opens the Team page: a social page for each person, with their links and a
chat with their AI. With Workspace.Experimental alone, it opens team records, the
window described last below. Workspace.Experimental.UX is a right of its own: having
Workspace.Experimental does not bring it. See Permissions.
The Team page
The Team page is an experimental prototype. The people on it are your workspace's team members and you; their titles, bios, posts, work, skills and chats are invented samples, marked as samples. Nothing on it is sent anywhere, and nothing is kept after you close the window.
The People tab shows a card for each person: their picture with a dot for whether they are around, their name, their title, two lines of their bio, their links as icons, and Ask their AI, which opens their page on the chat. Press anywhere else on a card to open their page.
A person's page opens under All people, which takes you back. At the top is their header band: a large picture with the same dot, their name and title, whether they are around and their local time, their links as icons, and two buttons, Ask Ada's AI (it opens the chat) and Ask Ada directly. Ada stands for whoever's page you are on. Asking them directly is not built yet: the prototype says that nothing was sent. On your own page the band holds the bio interview button instead.
Under the band is their bio, two lines with more to read the rest. Then come the tabs: Posts, Working on, Knows, Issues, the chat, and on your own page My AI and privacy, where you choose what your AI may answer about and who sees your posts. A section with nothing in it is hidden on someone else's page; on your own page it stays, and an empty Posts, Working on or Knows says how it fills. The hub's other tabs, Who knows what, Team room and Velocity, sit beside People.
Their links
Each link is an icon in the page's own colour, named for what it opens: hover for the name, and a screen reader reads the same name. A link that is an invented sample keeps a small sample mark. Only work links are shown, each told apart by its address:
| Link | Told apart by | Named |
|---|---|---|
| GitHub | GitHub: ada or a github.com address | Ada on GitHub |
| a linkedin.com address | Ada on LinkedIn | |
| Microsoft Teams | a teams.microsoft.com address | Ada on Teams |
| Slack | a slack.com address | Ada on Slack |
an email address or mailto:; their email and other emails | Email Ada | |
| Website | any other http or https address, such as a blog or x.com | Ada's website |
The icons are plain icons, not the companies' logos: their brand rules do not allow a redrawn one-colour logo. A web link opens in a new tab. A handle that is not an address, such as a git author name, is listed under the icons as text.
Chat with their AI
The chat tab looks like Thinky's chat. It starts with a notice, You are chatting with Ada's AI, not Ada. Your questions sit on the right, marked You. Each answer sits on the left, marked with their picture and "Ada's AI", and above it, always, the line Answered by Ada's AI, not Ada: the answer comes from their AI, from what they built and discussed, and is not Ada speaking. Under the answer are its sources and Ask Ada directly. Their AI only answers about what Ada allows in My AI and privacy; outside that it says so, shares nothing and points you to Ada.
The Team room uses the same chat parts. Your messages sit on the right and everyone else's on the left. When someone is @mentioned, their AI answers for them, marked Ada's AI, for Ada, and each message says whose AIs read it. On your own AI's answers you can Correct it or Take over.
Team records
With Workspace.Experimental alone, each person is a row, listed by email. A green Team member pill
marks someone who is on the team now, and the pill at the top counts them. A person you kept a record for
who has since left stays listed after the members, marked No longer a team member, with their record
kept. Press a person to open their record inside their row, and press them again to close it. A record
holds:
| Box | What goes there |
|---|---|
| Bio | A few lines about who they are and what they work on. |
| Microsoft Teams | Who they are on Teams. |
| Slack | Who they are on Slack. |
| Other emails | One per line. Each needs an @, or Save stays off with a line saying so. |
| Other handles | One per line, such as GitHub: sam-acme or a git author name. |
Press Save to keep it; Cancel puts back what was kept. There is no resume or employment field. The records are kept in your browser in the same way as Setup, and the bio is typed by hand: writing it from a short interview with Thinky, which the person accepts, is planned and not built. The Team page reads these records too: a bio, Teams name, Slack name, emails and handles kept in this browser are what it shows for that person, in place of the sample.
Related
- Describe Your Nodes - the Description and AI notes boxes.
- Team Members - who is on the workspace's team.
- Permissions - where
Workspace.Experimentalcomes from. - Data Stores - the Cloud > Data Stores window.
On fathym.com: OpenX →