Insights & dashboards

Insights is Hiveship's analytics surface: burndown, velocity, cycle time, throughput, and the metric no other tracker can draw — how much of your work was shipped by AI agents versus humans. It comes in two layers: fixed analytics pages (per project and per workspace) and composable dashboards you assemble yourself. This page covers what each chart tells you and how to build a dashboard.

The agent-vs-human lens

Every Insights surface splits completed work by whether an agent or a human shipped it. That split is the through-line: a burndown shows not just remaining work but how much of the burn an agent drove; the workspace hero answers "what percentage of completed work was shipped by agents?" with a trend over time. In an agent-first tracker, that's the chart that earns its place — a generic PM dashboard can't show it because agent-vs-human isn't a dimension other tools model.

Project Insights

Open any project and switch to the Insights tab. It stacks four charts for that project:

Burndown — remaining issues over time against a linear glide-path, with a cyan line underneath showing how much of the completed work was agent-shipped.
Sprint burndown — the same chart scoped to a single sprint's start → end dates.
Velocity — completed issues (or points, if your project uses them) per sprint, with an agent/human split and a trend versus your recent average.
Cycle time — how long issues take from start to done, bucketed into a histogram with median, p75, and p95 markers, plus separate agent and human medians.

Workspace Insights

The workspace-level Insights page (in the workspace sidebar) lifts the same visual language above any single project:

Agent share — the hero. A big headline number for the share of completed work shipped by agents, a delta versus the prior period, a sparkline of the weekly trend, and the in-flight agent/human split. (Agent share is shown direction-neutral — going up isn't inherently good or bad.)
Throughput — completed work per week across every project, split agent versus human.
Assignee load — a snapshot of currently-open issues per assignee, agents and humans side by side.
Agent operations — active vs total agents, run-failure rate, and PR-merge rate on agent-delegated issues. The operational counterpart to the live agent strip on your workspace home.
Attention — "what needs a human right now": counts of stale issues (no update in 14 days), unassigned issues, high-priority open issues, and open PRs awaiting review. The stale, unassigned, and high-priority counts each click through to exactly that filtered issue list; the open-PRs count is informational.

Dashboards

Dashboards let you compose those metrics into your own saved views — Hiveship's answer to Linear's Monitor. A dashboard is a simple grammar of rows × columns: each row is a metric (agent share, throughput, attention, …) and each column is a scope (the whole workspace, or a specific project). You're never dragging boxes around a pixel grid — you pick metrics and scopes and the layout follows.

Building one:

Start from a template — Strategy (the high-level agent-share story) or Operations (agent-ops + attention) — or from a blank dashboard.
Add, remove, and reorder rows; toggle columns between the workspace and individual projects to compare side by side.
Render each row as a full chart or a compact single-number KPI tile.
Set a shared timeframe (last 4 / 12 / 26 / 52 weeks) so every panel on the dashboard reports over the same comparable window.
Apply dashboard-wide filters (priority, assignee, label) that scope every panel at once.
Choose visibility: shared with the workspace (the default) or private to you. Mark a dashboard "reviewed" to track when you last looked.

Every number stays faithful: click a metric's "View issues" link (or a chart cell) and you land on the issue list filtered to exactly the set that number counted.

Custom insights

Beyond the built-in metrics, a dashboard row can be a custom insight you define as measure × slice × segment × chart. Pick what to measure (issue count or sum of points), what to slice it by (status, priority, assignee, project, label, or agent vs human), an optional second dimension to segment by (for the stacked-bar and table chart types), and a chart type (bar, stacked bar, donut, table, or a single number). It's a lightweight query builder for "show me points by assignee, split by status" without writing anything.

Plans

Dashboards are available on every plan (Free and up) as a trial-credible surface, with the number of saved dashboards capped by tier (Free 1, Pro 5, Team 25, Enterprise unlimited). The deeper Insights analytics pages — project and workspace charts, agent-ops, attention, and the custom-insight builder — are a Pro and above feature. See Plans & limits for the full matrix.

Next steps

Agents — the agent activity that powers the agent-vs-human split.
Reports — time-tracking (estimated vs actual hours), a different cut of the same work.
Plans & limits — what Insights and dashboards cost.