Skip to content
SAMPLE REPORTPrepared for the leadership teamPennine Wholesale Solutions Ltd
AI Foundation Audit — Sample

Your top 3 AI opportunities, ranked by AI Foundation Audit

A worked example of the report you receive — real structure, real numbers, fully scored. This is what every audit delivers.
SAMPLE REPORT
Company
Pennine Wholesale Solutions Ltd
Industry
Wholesale distribution of electrical and industrial supplies
Company size
150-500 employees
Audit date
11 June 2026
How scoring works

Every process scored on one number: the AOS

Visualization of evaluated processes by ROI (Return on Investment) and AI agent suitability. Bubble size corresponds to the overall Agent Opportunity Score. Color indicates risk level — green (low), yellow (medium), red (high).

Executive summary

The headline numbers

£120,282
Annual Cost Reduction
12
Processes evaluated
66/100
Average AOS

Here is what the audit found, in one paragraph.

Recommended pilot
Proposal Generation
Annual Cost Reduction
£50,098
Portfolio payback
4 months
Year-one investment
£15,238
Portfolio savings
Payback tier: Excellent
Portfolio savings
£120,282
Year-one investment
£45,220
Three-year net
£315,626
Portfolio payback
4–8 months
FTE capacity returned
1.8
The shortlist

Your top 3 AI opportunities

Your top three AI opportunities, ordered by payback speed — the fastest to pay back listed first.

Proposal Generation

Strong Candidate
68/100
Strong Candidate
£50,098
Annual savings
4 mo.
Payback
RiskHigh risk

Accounts Receivable

Strong Candidate
55/100
Strong Candidate
£39,257
Annual savings
5 mo.
Payback
RiskHigh risk

Email & Ticket Triage

Ideal First Agent
76/100
Ideal First Agent
£30,927
Annual savings
6 mo.
Payback
RiskMedium risk
Opportunity map

Where every process lands

All evaluated processes plotted by ROI and suitability. Your top 3 are highlighted.
Top 3 opportunity
Other evaluated process
Low riskMedium riskHigh risk
Suitability
3
1
2
ROI
1Proposal Generation2Accounts Receivable3Email & Ticket Triage

Each dot is one of your evaluated processes: the horizontal axis shows its ROI potential and the vertical axis how suitable it is for an AI agent today, so a process moves right as the financial upside grows and up as it becomes easier to automate reliably. The top-right quadrant holds your strongest, safest places to start — high return and high suitability — which is why your top 3 cluster there, while the bottom-left holds low-value, harder-to-automate work to leave for later. Each dot's colour adds the third dimension, implementation risk (green low, amber medium, red high), so a top-right dot that is also green is the ideal first agent.

How to read this Top-right is best — high ROI, high suitability, low risk.

Worked business case

One opportunity, fully built out — Email & Ticket Triage

£30,927
Annual savings
6 mo.
Payback
76
AOS score
As-is today

Email & Ticket Triage Automation currently absorbs 110 hours per month across a team of six staff at £22 per hour, processing approximately 280 customer emails daily across five query categories: order status, product technical questions, complaints, account and pricing queries, and general enquiries. Manual categorisation consumes 30% of representative time, and an 18% initial misrouting rate forces secondary forwarding on roughly 50 emails per day. Twelve percent of emails require multiple touches before full resolution, and 5% of complaints escalate specifically due to perceived slow response. SLA compliance — initial acknowledgement within 1 hour and first substantive response within 4 hours — depends on individual availability rather than a systematic process. Handoffs between representatives lack structured context, requiring the receiving agent to re-read full email threads and query HubSpot and SAP Business One independently.

To-be with an AI agent

A Decision Agent monitors the Outlook shared mailbox continuously during operating hours, ingests each inbound email as a structured event, and classifies it into one of five defined categories using intent analysis, sentiment scoring, and customer VIP status retrieved from HubSpot CRM. The agent routes each email to the correct internal recipient, writes a structured triage record to HubSpot, and dispatches a personalised acknowledgement within the 1-hour SLA window. For standard query categories, the agent generates a draft first-response pre-populated with SAP Business One order and account data, ready for representative review and send. VIP accounts, high-value complaints above £1,000, and high-negative-sentiment emails trigger immediate Slack escalation alerts to the team lead. All substantive outbound responses require explicit human approval before dispatch.

The agent does
  • Monitors the Outlook shared mailbox continuously during operating hours (8am to 6pm, Monday to Friday) and ingests each new inbound customer email as a structured event for downstream processing.
  • Reads and interprets the full email body, extracting intent, sentiment, urgency signals, and any referenced order numbers, account references, or delivery dates — flagging records where key identifiers are absent so that a clarifying reply can be dispatched without human reading time.
The agent does not
  • The agent does not send any final substantive response to a customer without a human representative reviewing and approving the draft — all outbound replies beyond the automated acknowledgement require explicit human sign-off, in line with the autonomy level of supervised operation.
  • The agent does not approve or communicate any goodwill gesture, credit, or refund decision; all such outcomes above £200 remain exclusively with the head of customer experience as defined by the existing authorisation policy.
  • The agent does not access, modify, or process any financial account data in SAP Business One beyond read-only order and account status lookup, in compliance with GDPR, FCA guidance for account-related queries, and internal data protection policy.
  • The agent does not handle out-of-hours emergency escalations autonomously; these follow the documented on-call rota and are passed to the designated human contact without agent intervention.
The agent escalates
  • Any email identified as originating from a VIP account is immediately flagged and routed to the customer service team lead, bypassing standard queue prioritisation, in line with the defined VIP escalation rule.
  • Complaints referencing missing deliveries where the inferred order value exceeds £1,000 go directly to the head of customer experience, consistent with the existing exception flow for high-value delivery failures.
  • Emails where sentiment analysis detects high negative emotion — anger, distress, or explicit threat of legal action — are escalated to the team lead for immediate human review before any response is drafted.
  • Multi-issue emails where the agent identifies two or more distinct query categories within a single message, representing approximately 15 percent of non-standard emails, are flagged for human triage rather than split-routed autonomously.
  • Technical product questions requiring engineering input — where the agent cannot match the query to an existing knowledge base in SharePoint article — are escalated to the appropriate technical contact with the agent's extracted query summary attached for context.

Full per-phase deliverables, integration spec and agent task list are included in the paid report.

Process flowchart — today vs. with an AI agent
As-is today
110 HRS / MONTH 280 VOLUME 18% ERROR RATE Email & Ticket Triage 1 System Email arrives in shared customer service inbox 2 Human Customer service representative opens and reads email 3 Human Manual categorisation by topic and priority Exception (VIP / complaint)? Yes Escalated No 4 Human Forward to appropriate department or respond directly 5 Human Update HubSpot CRM record with interaction 6 Human Send acknowledgement to customer if not immediate response 7 Human Track resolution in service log spreadsheet 8 Human Close ticket and send satisfaction survey Resolved PROCESS STEPS IN DETAIL 1 Email arrives in shared customer service inbox 2 Customer service representative opens and reads email 3 Manual categorisation by topic and priority 4 Forward to appropriate department or respond directly 5 Update HubSpot CRM record with interaction 6 Send acknowledgement to customer if not immediate response 7 Track resolution in service log spreadsheet 8 Close ticket and send satisfaction survey for complex cases E Any email identified as originating from a VIP account is immediately flagged and routed to the customer service team lead, bypassing standard queue prioritisation, in line with the defined VIP escalation rule. · Complaints referencing missing deliveries where the inferred order value exceeds £1,000 go directly to the head of customer experience, consistent with the existing exception flow for high-value delivery failures. · Emails where sentiment analysis detects high negative emotion — anger, distress, or explicit threat of legal action — are escalated to the team lead for immediate human review before any response is drafted. · Multi-issue emails where the agent identifies two or more distinct query categories within a single message, representing approximately 15 percent of non-standard emails, are flagged for human triage rather than split-routed autonomously. · Technical product questions requiring engineering input — where the agent cannot match the query to an existing knowledge base in SharePoint article — are escalated to the appropriate technical contact with the agent's extracted query summary attached for context.
To-be with an AI agent
AI Agent Human Email & Ticket Triage 1 Monitors the Outlook shared mailbox continuously 2 Reads and interprets the full email body Review needed? No Process completion Yes Responsible person: Reviews & approves escalated cases E Escalates flagged cases End PROCESS STEPS IN DETAIL 1 Monitors the Outlook shared mailbox continuously during operating hours (8am to 6pm, Monday to Friday) and ingests each new inbound customer email as a structured event for downstream processing. 2 Reads and interprets the full email body, extracting intent, sentiment, urgency signals, and any referenced order numbers, account references, or delivery dates — flagging records where key identifiers are absent so that a clarifying reply can be dispatched without human reading time. E Any email identified as originating from a VIP account is immediately flagged and routed to the customer service team lead, bypassing standard queue prioritisation, in line with the defined VIP escalation rule. · Complaints referencing missing deliveries where the inferred order value exceeds £1,000 go directly to the head of customer experience, consistent with the existing exception flow for high-value delivery failures. · Emails where sentiment analysis detects high negative emotion — anger, distress, or explicit threat of legal action — are escalated to the team lead for immediate human review before any response is drafted. · Multi-issue emails where the agent identifies two or more distinct query categories within a single message, representing approximately 15 percent of non-standard emails, are flagged for human triage rather than split-routed autonomously. · Technical product questions requiring engineering input — where the agent cannot match the query to an existing knowledge base in SharePoint article — are escalated to the appropriate technical contact with the agent's extracted query summary attached for context.
Autonomy level
Level 1

An autonomy level of 1 (supervised operation) is appropriate for this process because the email content spans a wide variety of writing styles — from concise B2B buyers to lengthy complaint emails with informal language and abbreviations — and the consequences of misclassification or an incorrectly routed complaint are directly customer-facing. The mix of rules and judgment required (Q18) means the agent can reliably handle classification and draft generation, but a human representative must review and approve all substantive outbound communications. The 15 percent non-standard email rate and the GDPR, FCA, and consumer rights compliance obligations further justify maintaining human oversight at the response stage.

Human-in-the-loop model

The human-in-the-loop model operates at the response approval stage. The agent classifies, routes, and drafts; the assigned customer service representative reviews the suggested draft in HubSpot, edits if necessary, and sends. For escalated items — VIP accounts, high-value complaints, sentiment-flagged emails — the team lead or head of customer experience receives a structured alert with the agent's classification rationale and relevant CRM and SAP data attached, enabling an informed decision without re-reading the raw email. The team lead retains daily inbox monitoring responsibility as a quality control layer over agent routing decisions.

The numbers behind it
Hours saved / month
110 hours/month
Effective hourly rate
£31.24
Monthly savings
£2,577
Annual savings
£30,927
Implementation cost
£8,337–£20,409 · £14,373
Three-year net
£78,408
Payback
6 months
Implementation roadmap
12–3 weeks
Phase 1: Discovery & Integration Mapping
22–4 weeks
Phase 2: Build & Classification Engine
31–2 weeks
Phase 3: Pilot, Validation & Full Deployment
Full per-phase deliverables, integration spec and agent task list are included in the paid report.
Also in your report

Two more business cases, fully built

Each of your top 3 opportunities gets the same depth of analysis. Here is a preview of the other two.
Worked business case

Proposal Generation

£50,098
Annual savings
68
AOS score
Full business case
Worked business case

Accounts Receivable

£39,257
Annual savings
55
AOS score
Full business case
Audit inputs

The profile this audit was scored against

ERP / CRM
SAP Business One, HubSpot, Microsoft 365, Slack, Outlook, Excel, internal warehouse system
API readiness
partial
AI maturity
exploring
Country
UK
Company size
150-500
What's inside

The full report at a glance

Eleven more sections are unlocked the moment you purchase your audit.
  1. Scoring Methodology
    Detailed in the full report
    Locked
  2. Risk Matrix
    Detailed in the full report
    Locked
  3. Technology Architecture
    Detailed in the full report
    Locked
  4. Implementation Roadmap
    Detailed in the full report
    Locked
  5. Financial Framework
    Detailed in the full report
    Locked
  6. Sensitivity Analysis
    Detailed in the full report
    Locked
  7. Monitoring KPIs
    Detailed in the full report
    Locked
  8. Strategic Conclusion
    Detailed in the full report
    Locked
  9. Assumptions & Disclaimers
    Detailed in the full report
    Locked
  10. Appendix
    Detailed in the full report
    Locked

Plus a full implementation brief

Locked

A technical implementation roadmap your IT team can act on immediately. Includes API specifications, integration architecture, and phased deployment plan.

For: IT team & project manager
  • Process analysis
  • Agent specifications
  • Integration architecture
  • Data requirements
  • Security & compliance
  • Implementation phases
  • Testing plan
  • Go-live checklist
What you receive

See your own numbers

Run the AI Foundation Audit on your business and get this exact report — built on your processes, your data, your ROI.

  • Executive Report
    40+ pagesCEO & management
  • Implementation Brief
    30+ pagesIT & project manager
One fixed price, money-back guaranteed.

Delivered to your inbox within 24 hours.

Pricing adapts to your region — see pricing →

easyAI · AI Foundation Audit — sampleRepresentative example — not a real customer