Skip to main content
addly

Turn a fictional page into verifiable actions

Read a structured summary, create two local actions, refresh the surface, and distinguish the interface from real content analysis. This quick start covers an interface and demonstration data: it is not an installation, a compatibility test, or Marketplace evidence.

Prepare the testOpen the lab
Confluence · Content macro7 to 10 minutes

Expected output

An exportable observation, not an automatic score.

Reset restores the fictional preset. Your observation and verdict stay local until you explicitly copy them.

What you will build

Actionable meeting record

Read a structured summary, create two local actions, refresh the surface, and distinguish the interface from real content analysis.

Read product limitations

Prerequisites before you click

  • A real page or meeting record whose expected decisions and actions you already know.
  • Your minimum rule: every action needs an owner and every decision needs a source.
  • A human-validation step before any future summary is shared.
  • No data to paste: this lab uses its own fictional content.

The initial preset is documented and restorable.

You know exactly what is already displayed before the action. ‘Reset all’ restores those values without retaining your trials.

  1. STATE 01

    One approved decision and one FOLLOW UP item are visible.

  2. STATE 02

    No action has been created and no refresh status is shown.

  3. STATE 03

    The text is entirely fictional; no AI service or Confluence page is connected.

One action, one expected result, no decorative step.

Instructions use only controls present in the demo. If a behavior is not interactive, it appears below as blocked evidence.

  1. 01

    Read the hierarchy

    Compare the summary, DECISION card, and FOLLOW UP card.

    Expected result

    Important information is differentiated without suggesting that a real page was analyzed.

  2. 02

    Create the local actions

    Activate ‘Create actions’ below the summary.

    Expected result

    A confirmation names two fictional actions; no remote issue or task is created.

  3. 03

    Refresh the surface

    Activate ‘Refresh’ and read the added status.

    Expected result

    The status confirms a local refresh and explicitly reminds you that the data remains fictional.

  4. 04

    Record the trust condition

    Note the source, human validation, and data processing that a real version would have to prove.

    Expected result

    The local verdict does not confuse a convincing layout with a validated analysis capability.

lab Confluence

Use the preset, then keep the trace.

The guide does not monitor your clicks; it tells you what to observe. Reset, observation, verdict, and Markdown export remain under your control.

Guided demonstration journey

The steps direct observation; they do not automatically detect your actions.

Step 1 of 333% of the guide displayed

Read the structured summary

Compare the approved decision with the FOLLOW UP item in the simulated Confluence macro.

Observe: The surface separates both information types without claiming to analyse a real page.

addly · SMART SUMMARY

What to take away

The committee approved the new journey. Three actions remain open before the June 24 pilot launch.

CURRENT
DECISION

Pilot journey approved

FOLLOW UP

Finalize the adoption measure

Demonstration data · Controls rendered with official Atlaskit components.

Record the local result

These notes stay on this page and are not sent to any service.

0/1,200

Local verdict

Step 1 of 3: Read the structured summary.

Demonstration interface and data. Host-product controls are rendered with Atlaskit; no app is installed.

Verification recipe

‘Blocked’ means this demo cannot produce the evidence. It means neither failure nor compliance of the future product.

Interactive

Decision and follow-up hierarchy

Observable in the simulated Confluence macro.

Interactive

Local creation of two actions

Interactive; the response names the two fictional objects.

Interactive

Refresh and state feedback

Interactive with explicit local status.

Blocked evidence

Summary quality, AI model, sources, and remote creation

Not testable: no service, customer content, or manifest is connected.

Troubleshooting and limitations

The first diagnostic separates a misunderstood local action from a capability that simply does not exist in the showcase.

Why does the summary content not change?

Refresh does not run analysis; it tests state feedback only. The content is deliberately fixed.

Where are the actions created?

Nowhere outside this page. The confirmation is a local demonstration state, not a Confluence or Jira write.

What should be required before a pilot?

The provider, flows, retention, sources, human validation, and exact permissions must be documented from the real implementation.

This quick start tests a concept in the browser. Before a pilot, require a real listing, manifest, compatibility, permissions, data processing, contract, and support channel.

Open the shareable datasheet

The quick start ends with a traceable decision.

Add the exported observation to the packet, assign the blocked evidence, and decide whether to continue, ask, pause, or reject.

Open the evaluation packetReturn to all four quick starts