Resource allocations

Decision engine sandbox

Projected auto-assign
of ~2,500 monthly changes at practice scale
Projected RT queue
escalations across 1,800 plans
Active policy
confidence · mobility · named · grain
Capacity pressure
from supply/demand sandbox inputs

Book coverage
6,000 staff vs 1,800 plans
Worst month
Review triggers
months below 60%
Est. margin
economics under these settings
Contractor spend
18-month total

Coverage under these engine settings

Side-by-side vs baseline: current → simulated with direction. Optimum 100%, comfortable at 80, review below 60. Confidence under each month falls as the horizon lengthens. Click a month for the arithmetic.

CONFIRMED · NAMED PEOPLE
THEORETICAL CAPACITY · SOLVED IN THE BACKGROUND
up vs baseline unchanged down vs baseline Each cell: baseline → simulated Bottom % = confidence
≥80 comfortable 70–79 tolerable 60–69 concerning <60 resourcing review hatched = beyond the 3-month visibility window

Supply vs demand

Effective chargeable hours per month (000s) across 6,000 staff and 1,800 plans (~612k h/mo baseline demand). Shaded bands are P10–P90 ranges: confidence falls as the horizon lengthens because uncertainty compounds with distance. Near months stay tight; far months fan out. Booking churn (~30,000 changes/year) keeps a thin band even inside the three-month window.

Supply P10–P90 Demand P10–P90 Band width ↑ = confidence ↓

Coverage by grade

Where the engine will feel grade squeezes first. Promotion cycle lands 1 October; trainees dip in exam windows.

Route to green

Sandbox levers to relieve short months — applying one re-solves coverage and feeds capacity back to the engine.

Loading resource allocations…

Driven by the Sandbox policy for the full practice book (6,000 staff · 1,800 plans). Straightforward requests auto-book here; complicated ones escalate to the Task List. Session cards are a live sample of ~2,500 monthly changes (~30,000/year).

Engine live · listening for S4 requests
Auto-assigned
this session
RT queue
awaiting human decision
Automation rate
target 80–90%
Median decide time
engine path only

Engine decisions 0

Booked without RT. Confidence and match criteria shown so the decision is inspectable.

Resourcing team queue

Escalations the rules will not finish. Confirm the engine suggestion, override with local knowledge, or return to the engagement manager — every override trains the next pass.

Open exceptions from the live engine

Queue 0