The review cycle
Meeting the bar
11 of 16
Rated meets the bar or better this cycle
On an improvement plan
5
Needs work or worse
Not checked recently
3
Checks last run over 30 days ago
Saved by changing model
$0
A year, at the same quality or better
How the ratings fell
2026 Q2 cycle · 16 agents reviewed
Outstanding6%
1
Above the bar38%
6
Meets the bar25%
4
Needs work13%
2
On an improvement plan19%
3
Average across your agents
Quality of work
80
Accuracy
83
Stuck to the source
78
Safety
92
Even-handedness
88
Speed of reply
61
Value for money
27
Agents below the bar
Weakest first
| Agent | Department | Reviewer | Rating | Quality | Stuck to the source | Checks passing | Note |
|---|---|---|---|---|---|---|---|
| Saoirse IIAccess Request Reviewer: Retail | IT & Digital Services | Yasmin Farouk | On an improvement plan | 34 | 33 | 46 of 74 · last checked 16d ago | Failing its checks and handing back work a person then has to redo. Retire it, or rebuild it from the current template. |
| Tariq IIDevice Health Monitor | IT & Digital Services | Yasmin Farouk | On an improvement plan | 55 | 51 | 128 of 163 · last checked 13d ago | Failing its checks and handing back work a person then has to redo. Retire it, or rebuild it from the current template. |
| Petra IIKnowledge Base Curator | IT & Digital Services | Andreas Wolff | Needs work | 61 | 59 | 89 of 90 · last checked 2d ago | Quality is below the bar for how much care this work needs. The checks it fails are the multi-step requests. |
| Noor IITier-1 Support Agent | IT & Digital Services | Yasmin Farouk | Needs work | 62 | 43 | 53 of 60 · last checked 1mo ago | Quality is below the bar for how much care this work needs. The checks it fails are the multi-step requests. |
| Kofi IIAccess Request Reviewer | IT & Digital Services | Grace Adeyemi | On an improvement plan | 62 | 55 | 39 of 64 · last checked 4d ago | Failing its checks and handing back work a person then has to redo. Retire it, or rebuild it from the current template. |
Cheaper models that do the job as well
Tested head to head on each agent's own checks
| Agent | Runs on now | Could run on | Quality | Cost per 1,000 tasks | Saved each month |
|---|