The Question Workspace Cost Dashboards Rarely Answer
Every AI tool in 2026 surfaces a cost dashboard. Notion Custom Agents shows credit consumption. HubSpot Breeze shows resolved-conversation spend. The pattern is the same: a workspace number, a 7-day trend, maybe a 30-day projection, and a budget alarm.
What none of them answer at a glance is the question owners actually ask once they have more than one operator: whose queue is the spend concentrated on? A noisy task that has been retried 30 times costs the same whether the original signal came from your sales lead or your ops manager — but knowing which queue it landed on is the difference between "let's look at this together" and a full team review.
Loop Desk shipped per-task cost attribution last week (rev 51). This week (rev 52) we are closing the per-recipient axis: cost-by-assignee aggregation, surfaced as a sidebar panel, scoped into the cost-spike alarm, and exposed as a programmatic endpoint your MCP host or downstream FinOps tool can read directly.
Why "Per-Teammate" Is the Missing Axis
The cost story we have been building since rev 7 has been almost entirely vertical — that is, it answers questions across time:
- Today vs yesterday vs the 7-day trailing baseline (rev 33/34/35).
- Per cycle drill-down on every desk run (rev 8).
- 30-day projection (rev 9).
- Hard daily cap with auto-pause at 100% and a Slack warning at 80% (rev 20/21).
- Cost-spike alarm that fires when today is 2× the trailing average (rev 32).
Every one of those works at the workspace level. A solo founder sees the right number; a multi-operator team sees it too, but they cannot see the distribution.
That distribution is what FinOps reviewers are starting to ask for. ISO 42001 control families on AI cost transparency call out per-team cost attribution as a maturity step. The recurring pattern in B2B procurement questionnaires is some variant of "show us how AI cost is allocated across your operators." Until rev 52, the only way to answer that on Loop Desk was to scroll the rev-51 top-cost-tasks panel and mentally aggregate by the assignee pill on each card.
What We Shipped
Three surfaces, one helper:
The dashboard panel. A new "Cost by assignee" panel sits alongside the rev-51 top-cost-tasks panel. Each row shows the teammate's total cumulative spend, what percent of the workspace total that represents, a proportional bar, and the underlying token count + task count. Unassigned work groups separately so the breakdown reconciles with the workspace total. The panel hides on solo desks where every row would name the same operator.
The cost-spike alarm gets per-assignee context. The rev-32 cost-spike Slack push now includes a "open-task spend by teammate" line when 2+ named assignees show up. Operators reading the alarm see "today's spend is 3.2× the trailing average — and Maria has $4.20 (40%), Leo has $2.10 (25%)" without leaving Slack. The same data flows into the workspace.cost_spike outbound webhook payload so downstream FinOps tools, project trackers, or board-status dashboards can route the alarm by teammate.
The v1 endpoint. A new GET /api/v1/cost/by-assignee mirrors the dashboard panel for MCP hosts and bearer-auth integrations. Same shape as the panel, accessible programmatically — pairs with the rev-51 /api/v1/tasks/top-cost and rev-13 /api/v1/runs to form a four-axis cost-observability surface (per-cycle, per-task, per-teammate, workspace 7-day) that the upcoming MCP server can wrap as four discrete cost tools without any custom bridge code.
Why This Matters for Governance-First Positioning
Loop Desk's pricing posture has been "flat fee, no per-cycle credits, you cannot overspend." The rev-20 daily cost cap is the structural enforcement. But cost transparency that is only descriptive ("here's what you spent") is half the answer for a B2B procurement reviewer asking "how do you control AI spend at the team level?"
Per-teammate cost visibility is the other half. It says: not only can you cap workspace spend, you can route the spike the moment it happens, you can attribute the cost back to a specific operator's queue, and you can pull that same answer programmatically from any tool that speaks MCP or HTTP. The answer is the same on every channel — dashboard, Slack, email digest, outbound webhook, v1 — because the underlying primitive is shared.
That symmetry is the operating principle of governance-first AI. Whatever the operator can see, the auditor can see, the integration can see, and the procurement reviewer can see.
What's Next
Per-recipient cost-spike scoping in the daily digest email — so a multi-operator team's owners each get their own "your tasks contributed $X.XX to today's spend" line — is the natural rev-53 step now that the per-recipient primitive exists. Watch the changelog for the next iteration.
If you've been waiting for the right moment to evaluate Loop Desk on the cost-transparency axis, this is it. Sign up, connect a source, and you'll see per-teammate cost attribution from the first time two operators have accrued AI spend.