OIA · Vital Signs - operational AI maturity scorecard · Worked example Hartwell M&E Group · £45m turnover · 180 staff · Maturity 2.4 / 5 · 4-week diagnostic · Bleed Rate £1.87m (4.2% of turnover)
ILLUSTRATIVE · Operational Intelligence Audit™ · Deliverable 01 of 07

Vital SignsOPERATIONAL AI MATURITY SCORECARD

Assessing current operational maturity across bid, compliance, commercial, reporting, systems and governance domains to establish the baseline for AI workflow implementation.

Prepared for
Hartwell M&E Group
Prepared by
Built AI Ltd
Date
May 2026
Vital Signs™ — AI Maturity Scorecard
← Audit PackDeliverable 01 of 07BAI-HME-AUD-001
Section 01

Overall Maturity Position

Hartwell M&E Group's aggregate AI maturity score across all six operational domains, benchmarked against a five-tier maturity scale.

2.4

Aggregate maturity: 2.4 / 5.0

Band: Reactive — AI usage is informal, unstructured and inconsistent across the business. Individual tools are used but without governance, workflow integration or measurable outcomes.
1.0–1.9
Ad Hoc
2.0–2.9
Reactive
3.0–3.4
Structured
3.5–4.4
Managed
4.5–5.0
Optimised
What this means. Hartwell is typical of a mid-market M&E contractor at this stage. Individual team members use AI tools informally (ChatGPT for drafting, email summaries, ad hoc document help) but there is no consistent workflow, no governance framework, no data classification and no measurable improvement programme. The opportunity is significant because the business has the operational complexity, document volumes and commercial processes where structured AI workflows can deliver measurable value.
Section 02

Domain Scores

Maturity scores across six operational domains, from the weakest (Commercial Control, 1.8) to the strongest (Systems & Governance, 3.0).

Bid & Tender
2.3
Reactive
Compliance & RAMS
1.9
Ad Hoc
Commercial Control
1.8
Ad Hoc
Reporting & Visibility
2.5
Reactive
Service Delivery
2.4
Reactive
Systems & Governance
3.0
Structured
Section 03

Domain Assessments

Detailed five-factor assessment for each domain, scoring process consistency, documentation quality, governance, data readiness and AI opportunity.

Bid & Tender

Score: 2.3 / 5.0 — Reactive
2.3
FactorFindingScore
Process consistencyTender production follows an informal process led by the Estimating Lead. No standardised workflow or checklist. Approach varies by individual and tender complexity.2
Documentation qualityScope extraction is manual and thorough but inconsistent. Assumptions are drafted from scratch each time. Clarifications are captured ad hoc. No bid risk register produced.2
GovernanceCommercial Director reviews high-value bids but there is no formal approval gate, QA checklist or submission compliance check process.2
Data readinessTender packs are self-contained and well-organised. Company data for PQQs exists but is scattered across documents. No centralised assumptions or compliance library.3
AI opportunityHigh. 40% of tender time is admin. 11 tenders declined for capacity. Scope extraction, assumptions, clarifications and QA are all structurally suitable for AI assistance.5

Compliance & RAMS

Score: 1.9 / 5.0 — Ad Hoc
1.9
FactorFindingScore
Process consistencyRAMS are produced by copying and adapting previous documents. Template quality varies significantly. Task sequences are often generic rather than job-specific.1
Documentation qualityHazard identification is reasonable but control measures are often boilerplate. COSHH and permit information is disconnected from the main RAMS document. Briefing records are incomplete.2
GovernanceHSQE Manager reviews RAMS but the review is often compressed by mobilisation timelines. No structured approval gate or version control system.2
Data readinessJob information exists in various forms but is not structured for automated processing. Task libraries and hazard registers are maintained in spreadsheets with inconsistent formatting.2
AI opportunityVery high. RAMS production consumes 1,730 hours per year. Task-based generation, hazard/control mapping and COSHH indexing are all addressable through structured AI workflows.5

Commercial Control

Score: 1.8 / 5.0 — Ad Hoc
1.8
FactorFindingScore
Process consistencyChange events are captured verbally and logged retrospectively. No structured capture process. Notice discipline is inconsistent — average 18 working days vs 10–14 day contractual windows.1
Documentation qualityVariation narratives are written manually with variable quality. Evidence is fragmented across WhatsApp, email and personal folders. 38% of submissions returned for further justification.2
GovernanceCommercial Director oversees the variation process but pipeline visibility is limited to a monthly spreadsheet. No real-time view of uncommercialised events.2
Data readinessVariation log exists with 142 events across 23 contracts. Evidence exists but is not indexed. Data is sufficient for AI workflow deployment with minor structuring.3
AI opportunityCritical. £620k in annual leakage from missed recovery. Change capture, notice automation, evidence indexing and narrative generation are the highest-value AI opportunities identified.5

Reporting & Visibility

Score: 2.5 / 5.0 — Reactive
2.5
FactorFindingScore
Process consistencyMonthly reports and QBRs are produced on a regular cadence. The process is established but heavily manual — 70–80% of effort is data gathering and formatting rather than analysis.3
Documentation qualityReport structure is reasonable but narrative is often copied from previous months with minor updates. Actionable insights are limited. Board reports lack forward-looking indicators.2
GovernanceReporting cadence is defined but there is no structured action-tracking or ownership mechanism. Actions from QBRs are not systematically followed up.3
Data readinessSage 200 CVR exports are available but require manual reconciliation. CAFM data is exportable. Multiple data sources need joining manually each month.2
AI opportunityHigh. 480 hours per month on reporting. Automated margin snapshots, WIP tracking, CNI dashboards and board narrative generation would release significant capacity for analysis and decision-making.4

Service Delivery

Score: 2.4 / 5.0 — Reactive
2.4
FactorFindingScore
Process consistencyHelpdesk operates during business hours with OOH call handling. Ticket creation is manual. Priority classification varies between operators and does not always align with contractual SLA tiers.2
Documentation qualityTicket records exist in Concept CAFM but quality is inconsistent. Response evidence relies on engineer self-reporting. SLA evidence packs are assembled retrospectively for client reporting.2
GovernanceSLA compliance is tracked monthly but deduction challenges lack structured evidence. No automated escalation or countdown tracking.3
Data readinessConcept CAFM contains 4,280 tickets with response times, SLA flags and deduction records. Data is exportable but requires cleaning and mapping to contractual SLA definitions.2
AI opportunityModerate. £100k in SLA deductions. Priority classification, dispatch ticketing and SLA evidence are suitable for AI assistance but require CAFM integration for maximum effect.3

Systems & Governance

Score: 3.0 / 5.0 — Structured
3.0
FactorFindingScore
Process consistencyCore systems (Sage 200, Concept CAFM) are established and used consistently. Document control is reasonable. IT infrastructure is stable and cloud-accessible.3
Documentation qualitySystem documentation is adequate. User guides exist for core processes. Data exports are available but not automated. No API integrations currently in use.3
GovernanceNo formal AI governance framework exists. No data classification model. No approval gates for AI-assisted outputs. No procurement disclosure position defined. Cyber Essentials certification in progress.3
Data readinessCore financial and operational data is centrally held. Export formats are standard (CSV, Excel). Data quality is reasonable but joins between systems are manual.3
AI opportunityModerate. Governance framework is needed as a foundation for all other workflows. Data classification, approval gates and audit trail infrastructure are prerequisites, not standalone AI use cases.3
Section 04

Priority Matrix

The gap between current maturity and AI opportunity for each domain. The largest gaps indicate the highest-priority areas for AI workflow implementation.

#DomainCurrent scoreAI opportunityGapPriority rationale
1Commercial Control1.85.03.2Largest financial leakage (£620k). Highest gap score. Change capture and variation recovery are immediately addressable.
2Compliance & RAMS1.95.03.1Second-largest gap. 1,730 hours on RAMS production. Requires governance framework before deployment (higher H&S approval burden).
3Bid & Tender2.35.02.7Highest data readiness. Fastest speed to benefit. Best prototype candidate. £340k admin burden + £836k declined pipeline.
4Reporting & Visibility2.54.01.5Strong efficiency case (480 hrs/month). Benefits from Phase 1 data improvements. Best deployed in Phase 2.
5Service Delivery2.43.00.6Smallest financial impact (£100k). Requires CAFM integration. Best deployed in Phase 3 after foundational workflows established.
6Systems & Governance3.03.00.0No gap — but governance is a prerequisite. The framework (Deliverable 05) must be established before any workflow deployment begins.
Key insight. Commercial Control and Compliance & RAMS have the largest maturity gaps, but Bid & Tender has the highest implementation readiness. The Opportunity Roadmap (Deliverable 03) resolves this by recommending Bid & Tender and Commercial Control for Phase 1 (highest combined value + readiness), with RAMS following in Phase 2 once governance controls are established.
Section 05

Methodology & Sources

How the scorecard was compiled and what evidence it draws on.

SourceDetailDate
Stakeholder interviews6 interviews: MD, Commercial Director, Estimating Lead, HSQE Manager, Finance Director, FM Operations ManagerW/C 5 May 2026
Document review12 tender packs, 8 RAMS, 6 variation submissions, 4 monthly reports, 3 QBR packs, CAFM export, Sage CVR exportW/C 5 May 2026
Data analysisSage 200 financial extract (18 months), Concept CAFM ticket export (12 months), variation log (all time), tender tracker (12 months)W/C 12 May 2026
Site observationHalf-day observation of estimating team (tender production cycle), HSQE review process, and helpdesk operation8 May 2026
Up next
BR · Bleed Rate
← Or back to the Operational Intelligence Audit page