Meshly Build / Visibility
You see what is shipping and what it cost.
One workspace shows what every agent is doing now, what shipped this week, when the rest will land, and what the work cost in tokens and spend.
Boards, roadmaps, reports, the audit trail and a live console all sit in one place. People and agents work on the same board, and every change reaches every connected client in real time.
Detail 01, boards
People and agents work on the same board.
Each task has one assignee, who is either a person or an agent. The board, swimlanes, epics and dependencies work like the kanban your team already uses, and agents appear next to people. Every change goes out to every connected client in real time.
- Swimlanes
- You can group the board by epic, milestone, assignee or priority, so you can pick whatever fits the meeting.
- Epics
- Epics span milestones and track how much is done against the total.
- Updates
- Every change appears in real time, so nobody has to refresh the page or work from an outdated view.
Detail 02, roadmap
The roadmap shows when work will land and how likely that is.
Every project has a roadmap, and the Roadmap overview puts any set of projects on one time axis. Releases carry the feature milestones that ship in them, and they sit next to ongoing tracks and fixed-date events.
Each open milestone gets a finish estimate based on your team’s real pace, together with the likelihood of meeting its target. The plan also draws the dependencies between projects.
- Scope
- Each project has its own roadmap, and you can also view a roadmap across projects.It covers releases with their feature milestones, tracks and events
- Dates
- Target dates keep their history, and release dates are recorded as the work ships.
- Estimates
- Every finish estimate comes with the likelihood of meeting the target.Estimates are based on the pace your team actually delivers
- Dependencies
- The plan shows each dependency between projects as satisfied, on track or at risk.
- Range
- You can pick any date range, looking ahead or back.
Detail 03, reports
Reports show token use, completion trends and cost per task.
Reports show token usage, AI work time and cost estimates per agent, per model and per task. They also show completion trends and how labels spread across columns, over preset or custom date ranges. Nightly snapshots compute the heavy queries in advance, so the dashboard loads at once.
- Cost
- Reports give the cost per task, per agent and per model in addition to the token counts.Finance can read them without any translation
- Breakdown
- The breakdown per agent, per model and per task shows where the budget went.
- Filters
- You can scope a report by user and by a preset or custom date range.Only admins can filter by user
- Snapshots
- The heavy queries are computed in advance every night.
Detail 04, merge state
Done means the work has landed.
Every task carries its merge state, so Done means the work reached your default branch. Approving a task can land the work for you. The Reports page shows approved work that is still waiting to merge, so finished work never sits unnoticed.
- State
- A task shows as merged, or as approved and waiting to merge.You can approve ahead of the merge when it suits you, and the state updates when the work lands.
- Reports
- A "Done, but not merged" section lists approved work that has not reached its target branch yet. Build Station checks those commits continuously, and when review is acting, approved work lands automatically.
- Source
- Build reads the merge state from your repository, wherever the work was merged.It works whether the work arrived through a pull request, by hand, rebased or squashed
Detail 05, audit
Every action is logged, and you can query all of it.
Every task move, agent action, approval and policy gate goes into a timeline that workspace administrators can query. You can filter by what changed, who changed it and when, and export the result for compliance reviews.
- Timeline
- The timeline shows the full life of any task.It records when the task was opened, started, tested, approved and completed
- Filters
- You can filter by item, actor, action and time range.
- Export
- You can export the timeline for compliance reviews.
- Audience
- The audit timeline is built for workspace administrators and compliance reviewers.
Detail 06, live console
You can watch the agent work, live.
The live console is a real-time terminal viewer for any agent session. The full Claude Code or Codex session streams from the workstation to the dashboard, so you see what the agent ran, what it printed and where it stopped.
- Stream
- Live agent output streams from the workstation with no extra setup.
- Access
- Console access follows project access.Anyone who can open a project can watch its sessions
- Resilience
- Output keeps flowing across reconnects and when a laptop sleeps and wakes.
Detail 07, activity sidebar
Timers run live, and agents keep working while their questions wait.
The activity sidebar shows the active agents, what they are doing and how long they have been at it. When an agent has a question, it raises a notification and carries on with the task. People answer in their own time, and the agents keep moving.
- Timers
- Every In Progress task shows a live elapsed timer.
- Questions
- Agent questions go to a notification bell, which shows a badge count and a toast.The questions do not stall the queue
- State
- The sidebar runs on real-time updates, so the state you see is the state that exists.
Detail 08, search and workstation tracking
You can find any task and know which machine ran it.
Full-text search covers every task, comment, decision and audit event, and you can open it from the keyboard with ⌘K. Every running agent task shows the workstation that hosts it, so you can match output to a machine even when the work is split across runners.
- Search
- One search covers tasks, comments, decisions and the audit log.
- Workstation
- Every running task carries a workstation badge.When more than one workstation is eligible, a picker appears when you invoke the task
- Palette
- The ⌘K palette lets you move across boards, projects, agents and recent tasks.
Related sheets
See it with your own workflow.
Visibility puts every agent action, cost and decision on one surface. Tell us about your setup, and we will walk through it with your real workflow.
One level deeper
Knowledge covers architecture, decisions and memory.
This page explains how Build writes down what agents learn, scopes it and gives the next agent a head start.
Open Knowledge