Process & workforce migration

Before / After

For each scenario in flight: how the work changes, how the people move. The combined business-process and HR-migration view, side by side.

Scenarios covered
3
Net headcount
-3
across all scenarios
HR cost
£215k
training + transition support
Annual gain
£453k
productivity, fully-loaded
Training days
115
active

Balanced — claims triage + L1 copilot

Cut claims cycle time by 30% · cutover window: May 2026 → Jul 2026
Before

Customer files a claim. It lands in the L1 queue. An adjuster reviews documents, validates evidence, applies policy, decides liability + payout. L2 supervises a sample.

Process steps
  1. 01Customer submits claim via portal or phone
  2. 02L1 adjuster picks claim from queue
  3. 03L1 reviews evidence + policy + rate cards
  4. 04L1 drafts adjudication; L2 reviews 1-in-5 cases
  5. 05Customer notified of outcome
Cycle time
9.5 days mean (with long tail)
Error rate
3.2% adjudications need post-hoc adjustment
Queue depth fluctuates 50-180 claims. Surge events hammer cycle time.
Customer:NPS 38. Wait time is the top complaint.
Workforce (7 headcount)
  • Claims Adjuster (L1)×4
  • Claims Adjuster (L2)×2
  • Head of Claims×1
Tools used
ClaimDeskDocuFlowOutlook 365
After

Triage agent (autonomous) classifies, dedupes, attaches evidence, routes. L1 copilot (supervised-autonomous) drafts the adjudication; an Exception Triager reviews flagged cases. Policy-Drift Analyst monitors agent behaviour.

Process steps
  1. 01Customer submits claim via portal or phone
  2. 02Triage agent classifies + dedupes + routes (autonomous)
  3. 03L1 copilot drafts adjudication (supervised)
  4. 04Exception Triager reviews flagged cases only (~25%)
  5. 05Customer notified of outcome
  6. 06Policy-Drift Analyst reviews drift signals weekly
Cycle time
5.5 days mean — target 6.6 (-30%) hit with margin
Error rate
1.4% — agent drift caught before becoming customer-visible
Agent capacity is essentially elastic. Exception volume is the throughput bottleneck.
Customer:NPS forecast 47-52. Wait time complaints expected to halve.
Workforce (5 headcount)
  • Exception Triager (Claims)×1
  • Policy-Drift Analyst×1
  • Claims Adjuster (L2)×2
  • Head of Claims×1
Tools used
ClaimDeskTriage AgentL1 CopilotDrift MonitorDocuFlow
Net headcount
5from 7-2
Roles before
3
Roles after
4
Annual productivity gain
£285k
HR cost £56k

Workforce migration

4 L1 adjusters → 1 Exception Triager + 1 Policy-Drift Analyst + 2 lateral moves into L2 advisory bench. No external hires needed; one external hire kept in reserve for Policy-Drift Analyst pending Beatrice's decision.

28 training days
Ravi Mehta
May 2026
Claims Adjuster (L1)Exception Triager (Claims)
pathwayin progress
Beatrice Sandoval
Jun 2026
Claims Adjuster (L2)Policy-Drift Analyst
pathwaydrafted
Two more L1 adjusters
Jul 2026
Claims Adjuster (L1)L2 advisory (lateral)
role mergepending decision
External candidate
Aug 2026 if needed
(external)Policy-Drift Analyst (reserve)
new hiredrafted
Critical
  • Beatrice's pathway sign-off — high flight risk; retention conversation booked.
  • Policy-Drift Analyst role must be filled before L1 copilot crosses 75% autonomy threshold.
Risks
  • Marginal assessment on correctness if rate-card freshness slips.
  • Customer-facing language drift in dispute responses (mitigated by Outcome Steward review).
Wins
  • Net £230k/yr productivity gain, fully resourced internally.
  • Two emergent roles created, one on a known template.
active

Exception Coach — dispatch resilience

Reduce dispatch exception MTTR by 50% · cutover window: Apr 2026 → May 2026
Before

Dispatch coordinators receive exception alerts (driver hours, traffic, mech failure). They diagnose, decide a fix, and execute it manually in the TMS. The fix takes longer than the diagnosis.

Process steps
  1. 01Exception alert lands (TMS event)
  2. 02Coordinator diagnoses root cause
  3. 03Coordinator chooses re-route / swap / reschedule
  4. 04Coordinator executes in TMS
  5. 05Coordinator notifies customer if SLA at risk
Cycle time
MTTR 14.2 minutes mean
Error rate
0.6% exceptions resolved sub-optimally (e.g. avoidable customer-facing delay)
Surge events (storms, tunnel closures) overload the team.
Customer:NPS 41. SLA breach rate 1.4% on impacted shipments.
Workforce (9 headcount)
  • Dispatch Coordinator×8
  • Head of Dispatch×1
Tools used
Helios TMSRouteOptimaOutlook 365
After

Coach agent watches the TMS event stream and presents a precomputed diagnosis + fix recommendation when an exception fires. Coordinator confirms with one click; customer notification is auto-drafted by the agent.

Process steps
  1. 01Exception alert lands (TMS event)
  2. 02Coach agent diagnoses + presents top 1-2 fixes (real-time)
  3. 03Coordinator confirms fix (one click)
  4. 04Agent executes in TMS, drafts customer notification
  5. 05Coordinator reviews + sends notification
Cycle time
MTTR 6.8 minutes mean — target 7.1 (-50%) hit
Error rate
0.4%
Surge tolerance dramatically improved — agent is the throttle, not the team.
Customer:NPS 47 forecast. SLA breach rate halved.
Workforce (9 headcount)
  • Dispatch Coordinator×7
  • Agent Steward (Dispatch)×1
  • Head of Dispatch×1
Tools used
Helios TMSRouteOptimaCoach AgentDrift Monitor
Net headcount
9from 9
Roles before
2
Roles after
3
Annual productivity gain
£168k
HR cost £14k

Workforce migration

One coordinator (Jamie) takes on Agent Steward responsibilities — paid uplift, no headcount change, becomes template for further Steward roles in the org.

12 training days
Jamie Okonkwo
Apr 2026
Dispatch CoordinatorAgent Steward (Dispatch)
pathwayin progress
Other 7 coordinators
Apr 2026
Dispatch CoordinatorDispatch Coordinator (training)
no changein progress
Critical
  • Jamie is the only Steward initially — needs back-up identified before Auto-resolver scenario commits.
Risks
  • Coach agent latency under surge (>5s p95) would erode the time saving — drift monitor on it.
Wins
  • First Agent Steward in the org. Template for Claims, Billing, others.
  • No headcount reduction — pure capability uplift.
draft

Continuous reconciliation — billing terraform

Reshape Helios's commercial model · cutover window: Q4 2026 → Q3 2027 (planned)
Before

Monthly invoice generation. End-of-month is a sprint. Customers receive invoices on day 1-3 of the following month. Disputes consume Customer Success time. DSO ~32 days.

Process steps
  1. 01Mid-month: data validation pass
  2. 02Month-end: invoice generation (currently autonomous)
  3. 03Day 1-3: invoices issued
  4. 04Days 1-30: payment + reconciliation
  5. 05Disputes handled inline
Cycle time
End-to-end 33 days (settlement complete)
Error rate
0.4%
Capacity not the constraint; commercial model is.
Customer:NPS 53. Customers asking for real-time visibility.
Workforce (3 headcount)
  • Agent Steward (Billing)×1
  • Customer Outcome Steward×2
Tools used
Stripe BillingSAP S/4Billing Agent v3
After

There is no 'invoice generation' step. Each shipment settles against contract terms as it completes. Customers have a continuously-updated settlement position. Disputes resolve inline at shipment-level. Billing OrgUnit dissolves; settlement merges into Treasury and the customer-relationship side merges into Customer Success.

Process steps
  1. 01Shipment completes
  2. 02Settlement engine reconciles shipment vs contract in real-time
  3. 03Customer position updated continuously
  4. 04Disputes raised + resolved at shipment level
  5. 05Treasury monitors aggregate position; Settlement Steward owns edge cases
Cycle time
Real-time (the question doesn't apply)
Error rate
0.05% projected
N/A — pull-based, not push-based.
Customer:NPS 60+ projected. Real-time visibility is the new baseline.
Workforce (2 headcount)
  • Agent Steward (Settlement)×1
  • Treasury Analyst×1
Tools used
Settlement EngineSAP S/4 (read-only)Treasury Console
Net headcount
2from 3-1
Roles before
2
Roles after
2
Annual productivity gain
£0
HR cost £145k

Workforce migration

Billing OrgUnit fully dissolves. Tom moves Steward role from Billing to Settlement. Outcome Stewards merge into Customer Success. Treasury Analyst hired externally + Ines as internal candidate. 1 person needs careful pathway support (Noah).

75 training days
Tom Whitfield
Q4 2026
Agent Steward (Billing)Agent Steward (Settlement)
pathwaydrafted
Noah Kowalski
Q4 2026
Customer Outcome StewardCustomer Outcome Steward (CS-attached)
role mergepending decision
Ines Marchetti
Q1 2027
Customer Outcome StewardTreasury Analyst (settlement)
pathwaypending decision
External candidate
Q1 2027
(external)Treasury Analyst (settlement)
new hirepending decision
Critical
  • Customer contract renegotiation is the binding constraint — not technology.
  • Decision pending. Marginal pre-decision assessment on regulatory and workforce-intent.
Risks
  • Revenue recognition lines need finance/audit sign-off.
  • External Treasury Analyst hire is in scarce talent market — salary band needs review.
Wins
  • Eliminates the slowest part of cash conversion entirely.
  • Sets up new product capability ('real-time settlement') as a competitive lever.