See what your agents actually accomplish
A new workforce dashboard connects agents, recorded work and human review—so the next conversation is about performance, not how many agents you deployed.
An agent roster tells you what you have deployed. An operating review asks harder questions. Which agents are doing useful work? Where does a person have to step in? What should the team fix next?
Vocion's workforce measurement release brings those questions into Scorecard, with a view that starts at the organization and lets you follow the evidence down to a workspace, a team and an agent.
Start with the workforce. Follow the work.
The overview shows accessible, non-archived workspaces in your current organization. Open a workspace to inspect its teams and agents. Expand an agent to examine recorded activity, review coverage and cost evidence, with links to the places where an operator can investigate.
The rows are objects you can explore, rather than a leaderboard that ranks agents by how often they talk. An enabled agent with no recorded work is a question to investigate. A busy agent with little review coverage is a different question. Neither observation, on its own, tells you whether the business is better off.
Daily and weekly windows help operators decide what needs attention now. Longer windows help account for work that runs less frequently. Missing measurements stay visible.
Give the business outcome its own record
A proposal can involve an analyst, a writer, a reviewer and a retry. That is still one proposal.
Registered work items now connect action proposals and native mission runs to a stable business record. The record carries acceptance criteria, the accountable agent, configuration context and linked evidence. Human reviews can record an outcome, a structured reason and reported effort. A retry stays attached to the original item.
This gives an executive a more useful distinction: work was attempted, execution completed, or a person accepted the result. Those are separate states. A completed run does not automatically become accepted throughput.
Turn the review into a next step
Investigation prompts connect recorded evidence with a suggested action. Missing review coverage may call for sampling outputs. Feedback-processing failures may need an operational fix. Stale evaluations may need refreshed cases before a performance claim is credible.
These are prompts to investigate, not automatic diagnoses. The operator still needs to examine the work, choose a change and evaluate its effect.
What to expect today
The measurement release supports explicitly registered human-assisted, scheduled and multi-agent work. It does not automatically capture every business workflow. Recorded model costs can be partial, and overlapping cost sources should not be added together.
A manual baseline can be recorded as a measured observation or a clearly identified planning assumption. An estimate with no timed samples remains an assumption. It is not realized savings.
For leaders, the immediate benefit is a better operating conversation: which work has evidence, which results need review, and where the next improvement effort belongs. For operators, it is a path from a summary to the underlying work.
Try it: in a deployment containing the measurement release, open Scorecard → Workforce performance with an authorized admin account. Choose a workspace, inspect one workflow and review its evidence before setting a performance target.
Read how one work item spans multiple agents, or how feedback becomes evidence of improvement.
Release evidence
The implementation and acceptance plan, baseline provenance change, and native mission evidence change document the shipped scope and its limits. Availability depends on your deployment version.