The dashboard

The dashboard at /dashboard is the first screen you land on after signing in, and it was designed around a deliberately narrow question. Most dashboards exist to answer "look at all these numbers", and they end up overwhelming people who are still trying to figure out what to do next. Our dashboard is built around the much more useful question of "what should I work on right now", and every panel on it earns its place by helping you answer that.

The page is organized into a small number of panels arranged in a grid that adapts to your screen size. On a typical desktop the top row holds the learning and approvals panels, the middle row shows your work in progress, and the lower rows hold operational information about runs and budgets. On mobile the same panels collapse into a single scroll column without losing any functionality. Below we walk through every panel and the reason it exists.

+------------------------------------------------------------+
|  Continue learning   |  Curriculum progress  |  Approvals   |
+----------------------+-----------------------+--------------+
|  Your agents         |  Your swarms          |  Budgets     |
+----------------------+-----------------------+--------------+
|  Recent runs (each row deep-links into the trace viewer)    |
+------------------------------------------------------------+
|  Try a template  (curated, never repeats what you already   |
|                   ran)                                      |
+------------------------------------------------------------+

Continue learning

This panel surfaces the exact lesson and step you were last on in the curriculum, plus a one-click resume button. It is intentionally the largest panel on the page because the most common reason a learner stops making progress is not lack of motivation but the small friction of remembering where they were. We solve that for you by treating the curriculum as something the platform tracks on your behalf — close the tab on Monday, come back on Wednesday, and the lesson opens exactly where you left it.

Curriculum progress

The progress ring next to "continue learning" shows your completion across all curriculum modules, how many quizzes you have passed, and how close you are to being eligible for the certification exam. Clicking through opens the full curriculum overview, where each module expands to show the lessons inside it, your status per lesson, and the time-to-complete estimates. The ring is split into coloured segments so a quick glance tells you both your overall progress and which specific module is still missing — a much better signal than a single percentage.

Approvals inbox

Any time a swarm or agent runs a tool that is wrapped in a human- in-the-loop approval — a refund issued, an email sent, a database write performed — the run pauses and a card appears in your approvals inbox. Each card shows the agent or swarm that requested the approval, the exact tool call it wants to make, the arguments in full, and an estimated cost or impact. You can approve, reject, or edit-and-approve directly from the card; the run resumes within a second of your decision. Cards are persistent until acted on, which means it is safe to walk away from a long-running swarm and know that anything requiring your judgement will be waiting for you when you come back.

Your agents

This panel lists your most recently edited agents along with their provider, current model, and status badge — including the distinctive "Guarded" badge for any agent with guardrails enabled. Each card has the four actions you most commonly want from an agent: open it in the playground to chat with it, open it in theAgent Builder to edit it, export it as portable JSON or as LangChain / LangGraph code, and publish it to the community marketplace. The "View all" link takes you to the full /agents library with search and filters.

Your swarms

Identical in spirit to the agents panel, but for swarms. Each card shows a thumbnail of the swarm graph, the number of agent nodes it contains, when it was last run, and the latest run's status. The thumbnail is generated server-side from your canvas layout so it looks like a tiny version of the real swarm, which makes it much easier to find the one you are looking for than a generic list of names.

Recent runs

A reverse-chronological list of the last ten executions across all playground sessions, swarms, and notebooks. Each row shows the source (which agent or swarm or notebook), the run status, the wall-clock latency, the total tokens used, and the total dollar cost computed from the live model registry. Clicking a row deep- links straight into the trace viewer for that run, which is by far the most common path from "something looks wrong" to "I see what happened" — the dashboard exists in part to make that path as short as possible.

Budgets and burn-down

The budgets panel shows your current month-to-date spend broken down by provider, your configured monthly cap per provider, and a projected month-end total computed from your run rate over the last seven days. The projection is the bit that makes this panel useful as opposed to merely informative — it tells you on day eight of the month whether the rate at which you have been running experiments is going to put you over your budget by day twenty, long before that becomes a problem you cannot react to.

Try a template

The last rail on the dashboard is a small curated set of real-world templates chosen to be relevant to where you are in the curriculum and explicitly excluding anything you have already provisioned into your workspace. The intent is that whenever you land on the dashboard with a few free minutes you have a one-click path to a new interesting thing to look at, which is one of the most reliable ways to keep momentum while learning.