Skip to main content

Workstreams, agents and projects now appear as cards in chat

When the assistant surfaces a workstream, an agent, or a project mid-conversation, it now renders as a clickable card grid instead of a plain text list. One pattern, three places to use it — covered with examples below.

Written by Dan Pillay

The chat is becoming the navigation. When you ask the assistant what's available — workstreams to start, agents that fit a task, projects already in flight — it answers with a grid of clickable cards instead of a wall of text. One click moves you straight into the thing you asked about.

This used to be the Workstreams-only behaviour. As of this week it covers all three: Workstreams, Agents, and Projects.

Animated walkthrough — opening chat, selecting Dunder Mifflin, asking 'what workstreams are set up for this brand?' and seeing the workstream card grid render with Open Workstream buttons

Above: opening chat → selecting Dunder Mifflin → asking "what workstreams are set up for this brand?" → workstream card grid renders with one-click Open buttons. ~30 second walkthrough.

The three card types

1. Workstream cards

When you ask what to start on, the assistant surfaces workstreams — multi-step strategic flows scoped to your brand. Each card shows:

  • The workstream name (e.g. "Weekly Social Ideation", "Monthly Content Calendar", "LinkedIn Organic Growth")

  • The starting agent — "Start with Cultural Zeitgeist" tells you exactly where the flow begins

  • A one-line description of the outcome ("trend-led content ideas", "monthly content direction")

  • Your progress — "0/5 complete", "3/7 complete" — so you can pick up where you left off

  • An Open Workstream button — jumps straight into the flow with your brand pre-selected

2. Agent cards

When you ask what agents could help with a specific task, the assistant surfaces agent cards. Each shows:

  • A coloured icon + agent name (e.g. "Brand Positioning", "Brand Messaging Framework", "Value Proposition")

  • Suite and status tags — "Marketing Suite", "Beta", "+2 more" — so you know what kit it belongs to

  • A short description of what the agent does

  • A Run Agent button — fires the agent on the brand currently selected in chat

  • A Learn more link — opens the agent's full description without committing to a run

Crucially, agent cards can render in the middle of a larger reply. If you ask the assistant to plan a brand-positioning project, it can describe the approach and embed the relevant agents inline, then continue with next steps — all in one coherent answer.

3. Project cards

When you ask about your projects — what's in flight, what to pick up — the assistant surfaces project cards. Each shows the project name, the brand it's for, and a button to open it. Two common uses:

  • Recommendations matched to a goal — "I want to expand into a new market, which projects are relevant?" returns the cards that match.

  • Details on a project you're already working on — "What's the status of [project name]?" returns the card alongside the answer, so you can jump into the project view in one click.

How to trigger each — sample prompts that work

You don't need a special command. The assistant decides when cards are the right answer — but these prompts reliably surface them:

For workstreams:

  • "I want to create social media posts for my brand"

  • "What workstreams are set up for this client?"

  • "Where should I start on this brief?"

  • "Show me the content workstreams"

For agents:

  • "What agents are available to help with brand positioning?"

  • "Which agents do I use for paid media reporting?"

  • "Show me the social agents"

  • "I need to draft a press release — what should I run?"

For projects:

  • "What projects are open for this brand?"

  • "Which projects could I use the new campaign brief in?"

  • "Show me the active projects in the marketing suite"

What the assistant does in flow

The cards aren't a separate "menu mode" — they're rendered when the assistant naturally needs to point you at something. So you'll see them appear in three patterns:

  1. Direct answer — "Here are the workstreams you can access" followed by the grid. Best for when you asked "what's available?" directly.

  2. Mid-reply embed — the assistant walks through an approach, then drops in 3–5 agent cards as the recommended toolkit, then continues. Best when the cards are part of the strategy, not the whole answer.

  3. Recommendation with reasoning — the cards arrive with a "Best starting point: use X first" line, so you don't just see the options, you see which one to pick first and why.

Tips for using it well

  • Ask in plain English. "What's available?" works better than typing slugs or scrolling the Agent Library. The assistant resolves intent to the right catalogue.

  • Stay in the same chat. Clicking a card opens the agent or workstream in a new tab — your conversation context stays intact for follow-up.

  • Use it for triage. "What should I run first for this brief?" returns a curated card grid with a recommended starting point, not a generic list of every agent.

  • Cross-route between types. Ask about a workstream, then ask "which agents inside it should I customise?" — the assistant pivots from workstream cards to agent cards in the same thread.


The shorthand: if the answer to your question is "here, use this" — the assistant now says it with a card. Try it on your next morning chat instead of opening Workstreams, Projects or the Agent Library by hand.

Did this answer your question?