Prepared for Paul Davis and the WW2 Headquarters leadership team

Building the World's Leading AI-Enabled History Production Platform

A phased AI Lab designed to help WW2 Headquarters expand its global reach, increase content output and reduce production overhead while preserving historical accuracy, editorial judgement and human control.

Problem-first automation. Human-led creativity. Continuously improving production.

Chris Bradshaw, Founder, Collective Intelligence · July 2026

A message from Chris Bradshaw

Four outcomes that matter

01

Global Reach

Identify new international audiences and adapt distribution around real market intelligence.

02

More Content

Generate more long-form, short-form and platform-specific output from every production.

03

Lower Production Burden

Reduce repetitive editing, publishing, administration and channel-management work.

04

Editorial Trust

Keep historians, producers and editors visibly in control of facts, context, tone and final approval.

Five things the discovery call confirmed

01

Growth must sit alongside efficiency

Reducing editing and publishing costs matters, but cost reduction alone will not solve the business challenge. The programme must also support audience growth, discovery, conversion and sustainable revenue.

Efficiency will not solve the problem on its own. Growth and discovery must be built into the programme.
02

Global discovery is a major opportunity

Second World War history has a genuinely global audience. WW2 Headquarters should use audience and social intelligence to understand which topics, formats and stories are most likely to perform in different territories.

03

Editorial accuracy is non-negotiable

The system cannot rely on generic internet fact-checking or unsupervised AI output. Historical claims, archive usage, editing decisions and contributor meaning must remain grounded in approved sources and expert review.

We cannot compromise historical accuracy, context or tone.
04

Leadership time is limited

The programme must minimise demands on a small, time-poor team through focused workshops, targeted interviews, clear decision requests and asynchronous collaboration.

05

ROI must guide prioritisation

The programme should distinguish between what AI could do and what is commercially worth doing. Growth potential, cost reduction, risk, effort and speed to value must all be considered.

The opportunity is to scale reach without scaling headcount at the same rate.
Vintage broadcast microphone

The intelligence that wins is not the most powerful — it is the most organised.

From initial brief to a fuller programme

The discovery call expanded our understanding of what the programme needs to achieve. The original brief covered production efficiency. The revised direction addresses both growth and efficiency.

Original Brief
  • Publishing automation
  • Long-form editing
  • Short-form clips
  • Social distribution
Revised Scope
  • Global audience intelligence
  • International market discovery
  • Content and distribution strategy
  • Editorial trust architecture
  • Long-form production automation
  • Short-form content at scale
  • Membership and community support
  • Organisational memory
  • Back-office automation
  • Commercial growth measurement

Six priority areas

Each priority is independently scoped and sequenced based on commercial value, technical feasibility and editorial safety. Click any card to expand.

01

Global Audience Discovery and Social Intelligence

Build an intelligence layer that identifies promising audiences, territories, themes, formats, competitors and emerging content opportunities.

Capabilities

  • Territory-level audience analysis
  • Competitor and adjacent-channel research
  • Topic and format trend detection
  • Content-performance analysis
  • International distribution recommendations
  • Localisation and translation testing
  • Evidence-led content planning

Outcomes

  • Stronger international reach
  • Reduced guesswork
  • Better content decisions
  • Faster identification of growth opportunities
  • Territory-specific publishing plans
02

Short-Form Content and Distribution Engine

Turn approved long-form content into platform- and territory-specific short-form assets.

Capabilities

  • Clip identification
  • Virality or engagement scoring
  • Multiple aspect ratios
  • Captions and titles
  • Platform-specific copy
  • Territory variations
  • Approval workflows
  • Scheduling and distribution

Outcomes

  • Increased weekly output
  • More audience entry points
  • Greater value from existing footage
  • Reduced manual channel-management effort
03

AI-Assisted Long-Form Editing

Test how much of the first-pass edit AI can complete safely.

Capabilities

  • Speaker detection
  • Camera-selection logic
  • Transcript editing
  • Silence and repetition removal
  • Rough-cut assembly
  • B-roll recommendations
  • Captions and chapters
  • Human correction capture

Outcomes

  • Reduced editing hours
  • Lower cost per film
  • Faster publication
  • Reusable editing rules
  • A measurable human-review model
04

Podcast and Website Publishing Automation

Automate repetitive work from source ingestion through to draft publication.

Capabilities

  • RSS or source detection
  • Transcript and chapter generation
  • Website-ready copy
  • SEO and structured metadata
  • CMS draft creation
  • Human review
  • Publishing logs
  • Organisational memory

Outcomes

  • Faster turnaround
  • Reduced administration
  • Fewer errors
  • More consistent metadata
05

Membership and Community Support

Prepare for membership growth without requiring linear growth in support headcount.

Capabilities

  • Approved membership FAQs
  • Onboarding
  • Account guidance
  • Content recommendations
  • Community support
  • Human escalation

Outcomes

  • Faster member responses
  • Lower support workload
  • Better retention and engagement
  • Scalable membership operations
06

Back-Office Administration

Explore contained automation around invoices, expenses, financial preparation, reporting and recurring administration.

Useful but should sit behind opportunities that drive growth or remove major production constraints.

Capabilities

  • Invoice and expense processing
  • Financial report preparation
  • Recurring administration
  • Supplier communications

Outcomes

  • Reduced administrative overhead
  • Fewer manual errors
  • More time for production

Note: Useful but should sit behind opportunities that drive growth or remove major production constraints.

Library archive shelves

Every word broadcast to millions began as a decision made by one careful editor.

Human control at every stage

No AI output reaches an audience without expert review and explicit human approval. Historical accuracy, archive rights and contributor meaning are protected at every step.

Approval Workflow

1Approved Sources
2AI-Assisted Research or Production
3Automated Verification
4Confidence and Provenance Display
5Expert Human ReviewHuman
6Approved PublicationHuman
7Corrections Stored in Organisational Memory

Built-In Safeguards

Approved source libraries
Citation and provenance tracking
Confidence thresholds
Human review of factual claims
Human review of sensitive edits
Archive-rights checks
Contributor-meaning protection
Prompt and model versioning
Correction logs
Approval audit trails

How the Lab works

The Lab is not a one-time delivery. It is a continuously improving production system built on five founding principles.

1

Problem first, technology second

Every workflow starts with a real operational constraint or commercial opportunity, not a technology capability.

2

Human creativity, AI execution

Historians, producers and editors make creative and editorial decisions. AI executes, accelerates and scales.

3

Build and validate in levels

Nothing moves to production until it has passed quality review and met confidence thresholds defined by the team.

4

Visible human approval

Every AI output that reaches an audience carries a clear approval record. Nothing is published without human sign-off.

5

Build once and learn continuously

Each workflow generates data, corrections and improvements that feed back into the system and reduce future effort.

The Continuous Improvement Loop

Research01
Test02
Review03
Approve04
Measure05
Learn06
Improve07

Each cycle of the loop generates data, corrections and improvements that reduce future effort and increase reliability.

Newspapers printing press archive

The archive is not the past. It is the foundation of everything that comes next.

Discovery, alignment and AI Operating Model

Phase 1 produces the foundations everything else depends on: a clear operating model, a prioritised portfolio, a measurement framework and the right first pilot.

What happens in Phase 1

  • Leadership alignment workshop
  • Current workflow mapping
  • Current cost and headcount analysis
  • Audience and social-intelligence research
  • Global growth opportunity analysis
  • Editorial-control design
  • Human-review requirements
  • Technology and data assessment
  • Agent opportunity mapping
  • Governance design
  • Client-time model
  • ROI scoring
  • Operating-cost forecasts
  • Selection of the first pilot

Deliverables

  • Signed-off AI Operating Model
  • Prioritised opportunity portfolio
  • Global audience-discovery plan
  • Editorial accuracy and approval framework
  • Current and future-state workflow maps
  • Recommended first pilot
  • Six-to-twelve-month roadmap
  • Refined implementation costs
  • Twelve-month operating-cost model
  • Human-versus-agent cost comparison
  • Measurement framework

ROI Scoring Model

Every opportunity is evaluated across eight dimensions to identify the highest-value, lowest-risk starting point.

01Revenue potential
02Cost reduction
03Time saved
04Editorial risk
05Client effort
06Technical complexity
07Speed to value
08Ongoing cost

Minimal time on your side

The programme is designed around a small, time-poor team. Workshops are focused. Review requests are clear and bounded. Asynchronous input is maximised wherever possible.

Kickoff workshop

Half day

In person or remote

Leadership and key decision-makers

Stakeholder interviews

1–2 hours each

Individual or small group

Targeted and structured — 3 to 5 people

Review and feedback

1–2 hours

Asynchronous or short call

Draft review of operating model and findings

Blueprint sign-off session

2 hours

Leadership team

Final review and approval to proceed

Phase 1 and beyond

Phase 1 — Six Weeks

Weeks 1–2Discovery
  • Stakeholder intake
  • Kickoff workshop
  • Workflow mapping
  • Cost analysis
  • Editorial-risk mapping
  • Audience research begins
Weeks 3–4Design
  • Opportunity scoring
  • Social-intelligence findings
  • Operating-model design
  • Editorial-control framework
  • Technical assessment
  • First-pilot definition
Weeks 4–6Blueprint
  • Blueprint completed
  • Roadmap agreed
  • Budget and operating-cost model
  • Pilot plan signed off
  • Low-risk proof-of-concept work may begin

Twelve-Month Horizon

Phase 1

Discovery and operating model

Weeks 1–6
Phase 2

First proof-of-value pilot

Months 2–4
Phase 3

Production automation

Months 3–6
Phase 4

Content and distribution engine

Months 6–9
Phase 5

Governed end-to-end platform

Months 9–12

A balanced scorecard

The programme measures success across four dimensions. No single metric tells the full story. Growth without editorial quality is not success. Efficiency without sustainability is not success.

Growth

  • International audience growth
  • Views by territory
  • Engagement by territory
  • Membership conversion
  • Audience retention
  • Revenue per content asset
  • Localised-content performance

Efficiency

  • Editing hours saved
  • Publishing time reduced
  • Administrative steps removed
  • Cost per film
  • Cost per episode
  • Cost per short
  • External production cost reduced
  • Additional headcount avoided

Editorial Quality

  • Percentage accepted without correction
  • Number and severity of factual corrections
  • Archive and provenance compliance
  • Contributor-meaning accuracy
  • Human review time
  • Approval turnaround

Operating Health

  • AI and infrastructure cost
  • Reliability
  • Failure rate
  • Leadership time required
  • ROI by workstream
  • Time to measurable value

Phased. Reviewed. Approved at each stage.

Investment is structured in phases. Each phase is independently reviewed and approved. Nothing proceeds without your explicit sign-off.

Phase 1

Discovery, alignment and AI Operating Model

3–5 weeks
£8,000 – £12,000
Phase 2

Administration and publishing automation

4–8 weeks
£12,000 – £20,000
Phase 3

AI-assisted long-form editing

6–12 weeks
£20,000 – £35,000
Phase 4

Short-form content factory

5–10 weeks
£15,000 – £25,000
Phase 5

Automated distribution and continuous learning

6–12 weeks
£20,000 – £40,000

Ongoing AI Lab — Optional

£3,000 / month

Optional retainer covering monitoring, improvement, new workflow development and strategic support.

Technology and Infrastructure

Early stage£1,000 – £3,000 / month
Scaled production£3,000 – £8,000+ / month

Paid directly to providers. Not a margin item.

Programme Totals

Core programme£75,000 – £132,000
Core programme + 12 months Lab support£111,000 – £204,000

Each phase is independently reviewed and approved. Payment is staged to each phase milestone. Ranges reflect scope variables confirmed during Phase 1.

Start lean. Build evidence. Invest further when the returns justify it.

We understand that committing to a programme of this scale requires confidence — in the return, and in the timing. Cash is not abstract. Every pound spent on this programme is a pound that could go elsewhere. That is exactly why the programme is designed the way it is.

Phase 1

£8,000 – £12,000

investment

You get the AI Operating Model, ROI scoring for every opportunity, a twelve-month cost model, and a recommended first pilot with a refined implementation budget. Before committing a single pound to Phase 2, you will know what it costs, what it saves, and what it returns.

Exit with full financial clarity before committing further.

Phase 2

£12,000 – £20,000

investment

Administration and publishing automation. Transcript generation, CMS drafts, metadata, scheduling and publishing logs are handled by the system. Human time is freed for creative and editorial work.

£18,000 – £35,000 / year

Estimated savings from reduced administrative and publishing overhead — typically 8–12 hours per week across the production team.

Phase 3

£20,000 – £35,000

investment

AI-assisted long-form editing. First-pass edits assembled automatically. Editors review, correct and approve rather than building from scratch. The system learns from every correction.

£30,000 – £70,000 / year

Estimated reduction in editing hours across your production schedule. Varies with volume of films produced.

Phase 4

£15,000 – £25,000

investment

Short-form content factory. Each approved long-form asset generates multiple short-form clips at scale — without proportional headcount. More audience entry points from the archive you already own.

£20,000 – £40,000 / year

Equivalent cost of producing comparable short-form output manually or through external agencies.

Phase 5

£20,000 – £40,000

investment

Distribution automation and continuous learning. Savings from earlier phases compound. The system improves itself, reducing the marginal cost of each new workflow.

Compounding

Phase 5 amplifies returns from all previous phases. Savings measured at programme level.

Investment Round

Building the evidence that supports your next round.

Each phase generates measurable data: hours saved, cost per film reduced, content output increased, administrative headcount avoided. By Phase 3, you will hold concrete operational evidence — not projections — showing what the programme costs and what it returns. That evidence is exactly what strengthens an investor conversation. We are not asking you to fund a promise. We are building the proof, phase by phase, that makes the promise credible.

What makes this different

Antique map and compass

Six reasons organisations choose Collective Intelligence.

We start with problems, not technology

We do not arrive with a pre-built AI product and look for a place to put it. We start by understanding what is costing you time, money and missed opportunity. Technology decisions follow.

We understand editorial environments

We have worked inside media organisations, publishing environments and content-led businesses. We know that accuracy, tone, archive rights and contributor meaning are not constraints to work around — they are the product.

We build for human control, not AI autonomy

Our Lab model is built on visible human approval, confidence thresholds, audit trails and correction loops. AI handles scale and repetition. Your team handles judgement.

We think commercially, not just technically

The programme measures success by audience growth, cost reduction and revenue impact — not by how many AI tools are deployed. Every workflow must justify its ROI before it is built.

We deliver working systems, not reports

Phase 1 produces a plan. Every subsequent phase produces working, tested automation that your team uses in production. Nothing is theoretical.

We build the capability into your organisation

The Lab is yours. We design it so that knowledge, prompts, workflows and improvements accumulate inside your team — not in our heads. You are not dependent on us indefinitely.

Begin with Phase 1

Phase 1 answers four critical questions before any implementation begins. It is the foundation every subsequent phase depends on.

01

Where can AI create the strongest commercial growth?

02

Where can it remove the greatest operational cost or constraint?

03

What editorial and human controls are required?

04

Which first pilot gives WW2 Headquarters the fastest credible proof of value?

Phase

Discovery and Operating Model

Investment

£8,000 – £12,000

Duration

3–5 weeks

Book a Follow-Up CallReview from the top