Canary Co

A fictional diversified multinational group used as a creative and research sandbox. Canary Co is where ideas about work, AI, tradition, and change get tested before they hit the real world.

Founded 1923 · Fictional Active sandbox 6 divisions · 8,980 staff

CAN · LSE · Diversified Group · Calderbridge, England

Canary Co

A diversified multinational group operating through six divisions — Ports & Logistics, Health & Beauty, Infrastructure, Connect (telecoms) and Energy, run from a corporate centre founded in Calderbridge in 1923. Listed 1986. AI arrives unevenly across the group: customer operations and marketing move first, terminals and field services more cautiously, engineering and R&D somewhere in between.

£4.2B
Revenue
8,980
Employees
6
Divisions
17
Business units
1923
Founded

New · Worked engagement

From org understanding to AI adoption, in four moves — grounded in a synthetic ~9,000-person build of Canary Co and O*NET-coded occupations.

Understand the organisation · map every task to a three-tier AI model · generate personal learning plans · drive adoption with prompts, activities and a 90-day roadmap. Plus a searchable explorer over all 8,980 synthetic staff.

Open the worked engagement → Interactive Org Explorer →

Story engine

"Head office wants a clean transformation story. The divisions live the real one — patchy tools, retrained roles, quiet workarounds, and a few people protecting the parts of the work that shouldn't be automated. That gap is where every Canary Co story lives."


CC-01 /

Strategic Landscape

Current state · 2026

Business challenges

  • Retailer squeezeMargin pressure from the major multiples. Every penny saved in production gets absorbed by listing fees.
  • DTC growing but hardDirect-to-consumer arm built at speed. Customer data is rich; the logistics to serve it profitably aren't.
  • Brand agingCore household lines command loyalty from over-50s. Younger buyers don't notice the parent brand.
  • Three SAP instancesLegacy of acquisitions. Finance sees one version of the truth; operations see another.
  • Skills gap in analyticsData team hired fast. Churn is high. Tribal knowledge leaves with every leaver.

Opportunities

  • Personalisation at scaleDTC platform and customer data are genuine assets. AI could make them pay.
  • Supply chain intelligenceLogistics arm owns the data. Demand forecasting with AI could shave 3–4% off inventory cost.
  • R&D accelerationFormulation cycles take 18 months. AI-assisted ingredient matching could halve this.
  • Contact centre transformationHigh-volume, high-repetition. Clear AI use case. Early pilots show 40% handle-time reduction.
  • Workforce capabilityStructured learning investment now, before attrition forces reactive retraining.

AI risks

  • Uneven adoptionMarketing runs ahead. Factory floor won't touch it. The gap creates resentment, not transformation.
  • Hallucination in regulated useLegal, compliance, and product labelling are already using AI tools informally. Nobody owns the risk.
  • Data governance vacuumThree SAP instances, no unified taxonomy. AI trained on bad data produces confident nonsense.
  • Job displacement narrativeThe union is watching. Every automation story gets heard as a headcount story.
  • Vendor lock-inEarly choices with one cloud provider hardened into dependency faster than anyone expected.

CC-02 /

AI Risk Register

Risk heat by division · Q1 2026

  • Contact Centre — AI Agents
    High · 8.5
  • Legal & Compliance — GenAI drafting
    High · 7.8
  • R&D — Formulation AI
    Med · 7.2
  • HR — Recruitment Screening
    Med · 6.8
  • Manufacturing — Predictive Maintenance
    Med · 5.5
  • Marketing — Personalisation Engine
    Low · 4.5
  • Finance — Automated Reporting
    Low · 4.0

CC-03 /

Creative Projects

Art · Music · Video · Animation · In development

8 planned

Music · Ambient Score

Factory Noise / Office Silence

A composed ambient album in two parts — the factory floor sounds of Calderbridge in 1978, and the open-plan silence of a modern office in 2026. What manufacturing sounds like when machines replace hands. What knowledge work sounds like when AI replaces typing. Each track is a room.

Score drafted · Recording planned

Video · Documentary Series

Eight People at Canary Co

Short documentary portraits — eight fictional employees, each at a different point in their AI transition. The contact centre manager who runs the pilot. The factory supervisor who doesn't trust it. The data analyst who is its biggest advocate and most honest critic. No narration. Just the work.

Scripted · Not yet filmed

Animation · Motion Series

The Org Chart Moves

Animated org charts showing how Canary Co's structure shifted across three periods — the 1986 listing, the 2015 DTC pivot, and the 2026 AI reorganisation. Each chart starts still. Then roles change, departments merge, new titles appear, old ones vanish. Typography in motion as organisational argument.

Concept · Storyboard needed

Art · Poster Series

Internal Communications 1975–2025

50 years of fictional internal communications reimagined as posters. The 1975 safety notice. The 1986 stock listing announcement. The 2001 Y2K memo. The 2020 remote working policy. The 2026 AI acceptable use guidelines. Typography, tone, and authority of the institutional voice across five decades.

Series planned · First poster Q4 2026

Music · Generative

Shift Patterns

Generative music that responds to Canary Co's fictional production data — output volumes, shift times, incident reports, downtime logs. As the factory runs, the music runs. Idling machines: silence and drone. Full production: polyrhythmic and dense. The data of work translated into the sound of work.

Concept · Technical build needed

Video · Essay

The Balance of Tradition and Innovation

A 20-minute video essay using Canary Co as the frame. What do companies actually lose when they "modernise"? The craft knowledge, the informal systems, the institutional memory encoded in manual processes. Argues that the real cost of AI adoption isn't jobs: it's wisdom.

Script in progress

TBD

Project 08 — Not yet named

A slot held open for the idea that hasn't arrived yet.

Empty · Watching
The failure is the record. The gap is where the story lives.

Full backstory content is maintained in the Zines section. View Canary Co Backstory →

Full org chart content is maintained in the Zines section. View Canary Co Org Chart →

Skills & Prompts Catalogue

Every role at Canary Co, paired with the AI skills they need and the prompts that make them work. Searchable by role and business unit.

Skill Role Prompt Template Use Case Time Saving
Board Report Drafting
CEO/Executive
CEO / Executive
Summarise the following Q[X] performance data into a 3-page board report with exec summary, KPI table, and risks section: [data]
Quarterly board packs, investor updates, leadership briefings 3–4 hrs
per report
Strategic Scenario Planning
CEO/Executive
CEO / Executive
Generate three strategic scenarios for [business challenge] — optimistic, base, and pessimistic. For each: key assumptions, risks, and recommended actions.
Leadership offsites, strategic reviews, risk planning 5–6 hrs
per session
Campaign Brief Generator
Marketing
Marketing Manager
Write a marketing campaign brief for [product/initiative]. Audience: [audience]. Objective: [objective]. Budget: [budget]. Include: positioning, key messages, channel mix, and success metrics.
New product launches, seasonal campaigns, brand activation 2–3 hrs
per brief
Social Copy Variations
Marketing
Marketing Manager
Write 5 social media post variations for [platform] promoting [offer/product]. Tone: [tone]. Keep each under [character limit]. Include a call to action.
Always-on social content, A/B testing, campaign support 1–2 hrs
per batch
Job Description Writer
HR
HR Manager
Write a job description for a [role title] at a [company type]. Include: role summary, key responsibilities (8–10 bullet points), required skills, and what good looks like in 90 days.
Recruitment campaigns, role redesign, succession planning 45–60 min
per JD
Interview Question Builder
HR
HR Manager
Generate 10 structured interview questions for a [role] role, covering: technical competency, cultural fit, problem-solving, and leadership. Include scoring guidance for each.
Competency-based interviewing, panel prep, consistency across hiring managers 1–2 hrs
per role
Variance Analysis Narrator
Finance
Finance Manager
Analyse this budget vs. actuals table and write a 2-paragraph narrative explanation of the key variances. Flag any items requiring management attention: [table data]
Monthly management accounts, board packs, stakeholder reporting 1–2 hrs
per cycle
Finance Policy Summariser
Finance
Finance Manager
Summarise the key points of this financial policy document into plain English. Create a one-page briefing suitable for non-finance managers: [policy text]
Policy rollouts, manager briefings, audit preparation 2–3 hrs
per document
Process Documentation Writer
Operations
Operations Manager
Convert the following process notes into a formal SOPt with: purpose, scope, step-by-step instructions, roles and responsibilities, and exception handling: [notes]
ISO compliance, onboarding, handover documentation, quality management 3–4 hrs
per SOP
Incident Report Drafter
Operations
Operations Manager
Draft an incident report based on these notes. Include: incident summary, timeline, root cause analysis, immediate actions taken, and recommended preventive measures: [notes]
Health and safety, quality incidents, operational failures 1–2 hrs
per report
Change Request Documentation
IT
IT Manager
Write an IT change request document for [change description]. Include: change summary, business justification, technical impact, rollback plan, and testing approach.
IT change control, system upgrades, infrastructure changes 1–2 hrs
per request
User Comms Translator
IT
IT Manager
Translate the following technical IT notice into plain English for non-technical staff. Keep it under 150 words and include a clear action they need to take: [technical text]
System outages, upgrades, security alerts, policy changes 30–45 min
per notice
Response Template Builder
Customer Service
Customer Service Rep
Write a professional customer service response to this complaint: [complaint]. Acknowledge the issue, explain what happened, and offer a resolution. Tone: empathetic but clear.
Complaint handling, refund requests, escalation responses 15–20 min
per response
FAQ Generator
Customer Service
Customer Service Rep
Based on these common customer queries, generate a FAQ document with 15 questions and clear, friendly answers: [query list]
Self-service portals, chatbot training, agent reference guides 3–4 hrs
per FAQ set
Proposal Writer
Sales
Sales Rep
Write a sales proposal for [client name] for [product/service]. Include: executive summary, understanding of their challenge, our solution, commercial terms, and next steps. Tone: confident and specific.
New business pitches, contract renewals, account expansion 2–3 hrs
per proposal
Objection Handler
Sales
Sales Rep
I'm a sales rep selling [product] to [buyer type]. The prospect said: "[objection]". Write 3 responses that address the objection directly, build confidence, and keep the conversation moving forward.
Call prep, coaching, deal review, negotiation support 30–45 min
per scenario
Learning Needs Analysis
L&D
L&D Manager
Based on this job role and performance data, identify the top 5 learning needs. For each: describe the gap, recommend a learning format, and estimate time to competency: [role/data]
Annual LNA, onboarding planning, competency frameworks 3–5 hrs
per role
Course Outline Generator
L&D
L&D Manager
Create a course outline for a [duration] training on [topic] for [audience]. Include: learning objectives, module structure, assessment approach, and facilitator notes.
New course development, compliance training, leadership programmes 4–6 hrs
per course
Meeting Summary Writer
Admin/PA
Admin / PA
Convert these meeting notes into a formal meeting summary. Include: attendees, key decisions, action items with owners and deadlines, and next steps: [notes]
Board meetings, project reviews, all-hands calls 30–45 min
per meeting
Email Drafter
Admin/PA
Admin / PA
Draft a professional email on behalf of [name/role] to [recipient] about [topic]. Purpose: [purpose]. Key points to include: [points]. Tone: [formal/friendly].
Correspondence, stakeholder comms, appointment management 15–30 min
per email
Every prompt is a starting point. Canary Co staff are trained to edit before they send.

Patterns & Automations Library

Reusable AI automation patterns deployed or in testing across Canary Co's business units. Each pattern is a repeatable solution to a common workflow problem.


PAT-01 /

Pattern Library

Reusable automation patterns · 2026

PAT-001 · Reporting

Report Generation

Automated drafting of structured reports from raw data sources. Applies consistent formatting, extracts key metrics, and writes executive summaries. Human review gates before distribution.

TriggerData export / scheduled run
InputsSpreadsheet, database export
OutputsFormatted Word / PDF report
ToolsCopilot + Power Automate
Saving2–4 hrs/week

PAT-002 · Meetings

Meeting Summarisation

Transcription and summarisation of meeting recordings. Extracts action items, decisions, and next steps. Sends summary to attendees and logs in Teams channel within 5 minutes of meeting end.

TriggerMeeting recording complete
InputsTeams / Zoom recording
OutputsSummary email + action log
ToolsCopilot + Teams + Planner
Saving30–45 min/meeting

PAT-003 · Documents

Document Analysis

Extraction and classification of key information from large document sets — contracts, policies, reports. Builds a searchable summary index. Flags anomalies or items requiring human review.

TriggerDocument upload / batch job
InputsPDF / Word documents
OutputsSummary table + flagged items
ToolsCopilot + SharePoint + AI Builder
Saving4–6 hrs/batch

PAT-004 · Intelligence

Competitor Monitoring

Scheduled monitoring of competitor web content, press releases, and job postings. Synthesises weekly briefing with categorised changes: product, pricing, hiring, messaging. Delivered to leadership inbox Monday morning.

TriggerWeekly schedule
InputsWeb sources + RSS + job boards
OutputsIntelligence briefing email
ToolsPower Automate + Copilot + Outlook
Saving3–5 hrs/week

PAT-005 · Customer

Customer Feedback Analysis

Aggregation and sentiment analysis of customer feedback across channels — NPS surveys, reviews, contact centre transcripts. Weekly themes report with volume, sentiment scores, and top verbatims by product line.

TriggerWeekly data pull
InputsSurvey responses, reviews, transcripts
OutputsThemes report + sentiment dashboard
ToolsAI Builder + Power BI + Copilot
Saving5–8 hrs/week

PAT-006 · HR

CV Screening & Shortlisting

First-pass screening of applications against job criteria. Scores candidates on stated requirements and produces ranked shortlist with justification notes. Human hiring manager reviews all output before contact with candidates.

TriggerApplication received
InputsCV + job description + criteria
OutputsScored shortlist + notes
ToolsAI Builder + SharePoint + Teams
Saving2–3 hrs/role

PAT-007 · Compliance

Policy Compliance Checker

Automated review of documents against current policy requirements. Flags deviations, missing clauses, and out-of-date references. Used in contract review, tender responses, and product documentation.

TriggerDocument submitted for review
InputsDocument + policy library
OutputsCompliance report + flagged items
ToolsCopilot + SharePoint + AI Builder
Saving1–3 hrs/doc

PAT-008 · Operations

Supply Chain Demand Forecasting

Predictive demand signals from sales history, seasonality, promotions, and external signals. Weekly forecast update to procurement and production planning. Highlights confidence levels and anomalies for human review.

TriggerWeekly data refresh
InputsSales data, promotions calendar, signals
OutputsForecast update + exception report
ToolsAzure ML + Power BI + Power Automate
Saving6–8 hrs/week

PAT-009 · Learning

Personalised Learning Paths

Automated matching of employees to learning content based on role, performance data, and skill gaps. Generates individual learning plans and sends weekly nudge prompts. Tracks completion and updates line manager dashboards.

TriggerRole change, annual review, onboarding
InputsHR data + skills framework + LMS
OutputsLearning plan + progress dashboard
ToolsViva Learning + Power Automate + Copilot
Saving2–3 hrs/employee/year

PAT-010 · Finance

Invoice Processing & Reconciliation

Automated extraction of invoice data, matching against purchase orders, and routing for approval. Exceptions flagged to finance team. Reduces manual data entry and catches discrepancies before payment run.

TriggerInvoice received (email / portal)
InputsInvoice PDF + PO data
OutputsMatched record + approval routing
ToolsAI Builder + Dynamics 365 + Power Automate
Saving4–6 hrs/week

PAT-02 /

Power Automate + Copilot Integrations

By business unit · Priority automations

Marketing

  • Brief-to-copy pipelineCampaign brief in SharePoint triggers Copilot draft copy generation across channels.
  • Social schedulingApproved content auto-scheduled to Buffer/Hootsuite via Power Automate.
  • Performance alertsCampaign metrics below threshold trigger Copilot analysis and manager alert.

HR & L&D

  • Onboarding workflowNew hire triggers multi-step onboarding: IT access, induction booking, buddy assignment, learning plan.
  • Performance review promptsReview cycle opens, managers receive AI-generated talking points based on employee data.
  • Policy acknowledgement trackingNew policy published triggers acknowledgement flow with automated chasing.

Finance

  • Month-end close checklistAutomated task assignment, progress tracking, and escalation for close process.
  • Expense anomaly detectionExpense submissions above threshold or outside policy flagged automatically to finance manager.
  • Budget v actuals narrativeMonthly data refresh triggers Copilot narrative generation for management pack.
Patterns are not solutions. They are starting points that need a human at either end.

Tensions & Conflicts

The organised friction at Canary Co — the places where AI adoption collides with human expectation. These are the story engines. Each tension is real, recurring, and unresolved.


TEN-01 /

Active Tensions

Unresolved · Live · Q2 2026

TEN-001 Trust & Transparency High heat

Who owns AI outputs? When Copilot drafts a board report, who signed it? When an AI-generated analysis informs a redundancy decision, who is accountable? At Canary Co the attribution question is unresolved — and staff know it. Some managers have started quietly deleting AI draft history before sharing documents. The official answer is "human oversight at all stages". The actual answer is: nobody knows yet.

Head of Legal"If we can't show a document was reviewed by a qualified human, we have no defence. I keep saying this. Nobody writes it down."
Marketing Director"We have 40% more output since we adopted Copilot. If I have to timestamp every human edit, the efficiency gain disappears."
TEN-002 Skills Anxiety High heat

Fear of AI replacement sits underneath every training programme and every transformation roadmap. At Canary Co, it manifests differently by division: in the contact centre it's overt (they've seen the handle-time data). In the finance team it's quiet but present (they know what automated reporting means). In manufacturing it's defiant. The L&D team is trying to reframe "augmentation not replacement" but the message isn't landing when the headcount projections sit in the same board deck as the AI investment case.

Contact Centre Team Lead"We did the training. We got better with the tools. And then they told us the headcount would reduce by 20% by 2027. So what was the training for?"
Chief People Officer"We genuinely don't know which roles will change. Saying that feels weak. But the alternative — pretending we know — would be worse."
TEN-003 Quality vs Speed Medium heat

AI output arrives faster than human review can keep pace with. In marketing, the brief-to-copy pipeline can generate 50 social variants in the time it used to take to write three. The question is whether any of the 50 are as good as the three. The answer is: sometimes yes, mostly no, but faster and cheaper. This calculus is playing out differently across every team — and the "good enough" threshold is shifting in ways nobody has formally decided.

Creative Lead"Speed isn't the brief. The brief is work that's worth making. If I'm editing AI copy all day I'm an editor, not a creative. That's a different job."
CMO"Our output has doubled. Engagement is flat. Make of that what you will."
TEN-004 Data Privacy Medium heat

What can be shared with AI tools? The official policy says: no personal data, no commercially sensitive data, no customer data. The actual behaviour: customer transcripts pasted into public ChatGPT, financial models uploaded to consumer AI tools, HR case notes summarised via apps with unclear data residency. The gap between policy and behaviour is growing. Enforcement is impossible at scale. The data team is building a monitoring tool nobody has approved yet.

CISO"We have a policy. We have approximately zero enforcement capability. The policy exists to limit our liability, not to change behaviour."
Customer Service Manager"I told them to stop using ChatGPT for customer complaints. They stopped telling me about it."
TEN-005 Accountability Medium heat

When AI makes a mistake — a wrong prediction, a biased screening decision, a hallucinated fact in a regulatory document — who is responsible? At Canary Co this question has already arisen once, in a recruitment screening incident where a well-qualified candidate was rejected by an AI model that had been trained on historic data with demonstrable bias. The HR team handled it quietly. Nobody changed the model. The legal exposure is still being assessed.

HR Director"The model gave a recommendation. A human approved it. We acted on it. I'm not sure where responsibility sits and I'm not sure our legal team does either."
Head of AI"The model was doing what it was trained to do. That's not a model failure. That's a governance failure."
TEN-006 Job Definition Simmering

What stays human? Every job description at Canary Co is written for a world that doesn't include AI tools in the task list. As AI takes on drafting, summarising, scheduling, and analysing, the remaining human tasks are: judgement, relationship, craft, escalation, accountability. Nobody has formally redesigned a job description to reflect this. People are doing two jobs — their old job and the new AI-mediated version — without acknowledgement or additional pay.

Finance Business Partner"I used to spend 40% of my time building the model. Now I spend 40% of my time checking what the model built. The job looks the same from the outside."
CEO"The value has shifted, not disappeared. We need people who can direct AI, not compete with it."
TEN-007 Manager vs Managed Simmering

Different access levels to AI tools are creating visible inequality. Some managers have Copilot licences, their teams don't. Some divisions have AI tooling, others are waiting for budget approval. The people with access move faster, produce more, and are more visible to leadership. The people without access are being compared against AI-augmented output without the same tools. The resentment is quiet but building.

Regional Sales Manager"My peer in London has the AI assistant. I don't. We're both on the same targets. Explain that to me."
CTO"We're rolling out by risk tier and use-case maturity. Some divisions get it later. That's not inequality, that's governance."
Tension isn't failure. It's the evidence that something real is at stake.