Blocks

What is this agent allowed to use, and where does a person step in?

A canvas for designing business agents: triggers, agents, routers, guardrails, human approvals and actions as the flow, and under each agent its equipment — model, skills, MCP servers with per-tool switches and ask-first, knowledge and memory — on square ports with dashed links. A searchable palette (drag, or click to equip the selected agent), an inspector form per node kind, design checks that point at the node they are about, a simulated test run that pauses at approvals, undo/redo and auto-layout. Three sample designs: support resolution, invoice processing, lead qualification. Plain data in, no runtime.

Which tables are there, and how do they hang together?

An entity-relationship view of a database on the flow canvas: custom table nodes with a row per column (keys, nullability, personal data), foreign keys as custom edges with crow’s-foot end marks that meet each table at the right row, a table list by schema with search and visibility, an all-columns / keys-only / names-only switch, auto-layout, and an inspector with columns, followable relations, indexes and a readable CREATE TABLE. Renders a plain DataModel object.

Where is everyone, and who is late?

A fleet where it is: each vehicle's planned route as a dashed MapRoute, the part already driven solid, the vehicle's MapMarker at the join and the depot marked; the list shows stops done on a Meter and minutes against plan, and selecting a vehicle fades the others.

Where did the run fail?

An execution waterfall for one multi-agent run: a stat strip where every number carries its comparison (duration vs p95, retries, tokens in/out, cost vs workflow average, failed at), the charts package’s Gantt at sub-second granularity with one row per agent span nested under its caller, and a context-flow chain that shows the hop where a required field was dropped.

How does an agent answer a data question with more than text?

An A2UI answer composed by the agent: a headline, KPI tiles with sparklines, two AutoChart figures titled with their findings, the method and follow-up actions. A click on the chart drills into the next surface.

The sign-in form beside a brand panel painted with the primary pair. Container-query layout: the panel stands down when narrow.

Drag cards between columns or use each card's Move menu (keyboard, screen reader and touch), with a live-region announcement, per-column points and WIP-limit warnings in words.

Where does the process really go, and where does it wait?

A map-first process-mining workspace over an order-to-cash log: the six KPIs as one ribbon, every control in a toolbar so nothing floats over the map, and beside the map a panel of statistical insights — throughput distribution against the SLA, the slowest hand-overs, the weekly trend, box and violin plots by region and channel — plus variants as colour strips, the selected element and deviations. A dock holds the dotted chart, the performance spectrum, the workload heatmap and the case table; a replay animates every order as a token. One useProcessExplorer instance drives all of it, so no two panels can disagree.

Chat surface with transcript, composer, tool calls and a context rail.

How do I tell a customer’s story with their film, the numbers and the screens?

A case study as a reading page: headline and lead, the customer’s film in the ui Video player, results as metric tiles, the numbered narrative with pull quotes beside a sticky fact sheet, a screenshot gallery with a lightbox, related stories and a call to action.

How does a listener move through a show — episodes, chapters, the transcript?

A show’s episodes beside one player composed from the ui MediaPlayer parts — cover art, the waveform as scrubber, a bar with speed — plus clickable chapters and a transcript that follows playback; choosing an episode from the list starts it.