← eudai.

Implementation guide

Resilient operations and continuous adaptation

One model for bringing new ideas to market, managing the risks that come with them, and organizing work between people and AI: discover, experiment and learn, then operate, adapt and scale.

DiscoverExperiment and learnOperate, adapt and scale
November 2026eudai.ai
Why this model

Innovation and resilience are the same discipline

Most organizations run innovation and risk as separate tracks. One team looks for new products, markets and ways of working. Another looks for what could go wrong. They use different language, different evidence and different timelines, and often meet only when a launch is already late or a risk has already landed.

That separation is getting harder to afford. AI, new business models and converging markets are creating opportunities faster, and every one of them changes the organization's exposure: new dependencies, new partners, and decisions made by systems rather than people. AI is also changing who does the work, so the way work is organized has to keep changing too.

This model brings both into one cycle. Discover how work creates value and where it is exposed. Experiment to test new offerings and new ways of operating before committing to them. Then operate, adapt and scale, with evidence of both what is working commercially and what the organization can withstand.

The pages that follow show each phase through both lenses, innovation and commercialization alongside risk and resilience, and close with a practical first cycle and the roles that make it work.

The work that finds the next opportunity should also show what it takes to run it with confidence.

  • Separate innovation and risk tracks→One shared cycle
  • Launch first, assess risk later→Test opportunity and exposure together
  • Big-bang rollouts→Small, tested experiments
  • Plans as evidence→Measured results
  • Fixed operating model→Continuous adaptation
  • People checking AI at the end→Governance designed into the work

Eudai implementation guide. Related reading on the final page.

eudai.ai2 / 12

The model

Three phases, used for innovation and resilience alike. Discover how work creates value and where it is exposed. Experiment with new offerings and ways of working before committing to them. Then operate, adapt to what you learn, and scale what works.

The to-be model. Discover rises from concrete to abstract through Understand, Map, Identify and Ideate. Experiment and learn is a loop of Test, Measure and Learn. Operate, adapt and scale is a loop of Operate, Adapt and Scale.

Eudai implementation guide. Related reading on the final page.

eudai.ai3 / 12

The three phases, two lenses

Each phase serves growth and resilience at once

PhaseThe question it answersInnovation & commercializationRisk & resilience
DiscoverHow does our work create value, and where is it exposed?Where customer needs, technology or markets are opening new opportunitiesWhich operations, dependencies and partners the business cannot afford to lose
Experiment and learnDoes the new approach work the way we expect?Test propositions, pricing, channels and new ways of working with real customersTest how people and systems respond to disruption before it happens
Operate, adapt and scaleHow do we run it, grow it and keep it current?Scale what customers adopt and build it into the commercial engineEmbed the ownership, controls and visibility that let it run with confidence

Three principles that run through all of it

  1. Start concreteGround ideas in how work and value are created today before redesigning either.
  2. Test before you scaleSmall experiments show how customers, people and systems actually respond.
  3. Keep the loop runningMarkets and risks keep moving. The cycle feeds back into discovery.

Eudai implementation guide. Related reading on the final page.

eudai.ai4 / 12

People and AI in the same cycle

Each phase asks a new question

PhaseThe question it addsWhat changes in practice
DiscoverWhat can we accomplish when human and AI capabilities are considered together?Map today’s work, but don’t assume today’s tasks are the units future operations should be built around
Experiment and learnHow well does the combined system perform, decide and hold up?Test different arrangements of people and AI, not just whether a tool can do one task
Operate, adapt and scaleHow should the division of work, decision rights and controls change as capabilities change?Revisit who or what does the work, decides, verifies and is accountable

Three design principles

  1. Design around outcomesAsk which constraints have disappeared and what has become possible, rather than which existing tasks AI can take.
  2. Experiment with the operating modelMeasure quality, effort and new value from each arrangement of people and AI, then refine it.
  3. Keep reassigning capabilityResponsibilities are not fixed. Keep the expertise to step in when automation fails.

The human and AI lens builds on Sangeet Paul Choudary, The “Human-in-the-Loop” Fallacy, Harvard Business Review, November 2, 2026.

eudai.ai5 / 12

People and AI · who contributes what

Each step draws on both

StepAI contributionHuman contribution
UnderstandSynthesize signals, analyze information, detect patternsSet strategic intent, interpret context, challenge assumptions
MapMap dependencies, workflows, capabilities and constraintsDecide which relationships, decisions and capabilities matter
IdentifySurface newly feasible opportunitiesChoose the problems worth solving and define value
IdeateGenerate options, prototypes and alternativesWeigh trade-offs and apply expertise
TestRun repeatable evaluations and simulationsSet success criteria and examine exceptions
MeasureTrack performance and detect driftJudge consequences and what the evidence means
LearnRecommend changes based on resultsDecide what to change and why
OperateExecute, monitor and improve within set limitsExercise judgment, handle material exceptions, stay accountable
AdaptDetect changing conditions and suggest modificationsRevisit assumptions, redesign capabilities, redirect investment
ScaleReproduce validated capabilities and orchestrate workflowsSet decision rights, boundaries and resource commitments

This is not a permanent assignment of tasks. The split changes as human and AI capabilities develop.

eudai.ai6 / 12

People and AI · governance

Design governance into the work

  1. Design timePeople set objectives, standards, acceptable risks and decision rights.
  2. Run timeAI works within those limits, with automated controls, evidence capture and escalation.
  3. ContinuouslyPeople and AI review performance, spot changing risks and revise the system.

Accountability doesn’t require a person to touch every output. Some decisions need individual approval; others are better assured by monitoring and agreed controls.

An example: moving a content library

ApproachHow it worksWhat it leads to
Review everything by handExperts check every item before it moves to the new libraryReview capacity caps growth, and the old process carries over
Standards firstExperts set standards. AI migrates, checks and flags exceptions. Experts handle what matters and create new contentA library that keeps improving, with expert time spent on new value

Capability orchestration, across every phase

  1. Who leadsWhich decisions should be human-led, AI-led or shared?
  2. Build or retireWhich capabilities to build, buy, automate or retire?
  3. AttentionWhere is human expertise most valuable?
  4. EvidenceWhat shows the arrangement works?
  5. RedesignWhen does a new AI capability justify changing the process itself?

Eudai implementation guide. Related reading on the final page.

eudai.ai7 / 12

Phase 1 · Discover

Move from concrete to abstract

StepWhat you doInnovation & commercializationRisk & resilience
UnderstandBuild a clear view of the work, the value it creates and the outcomes customers depend onWhich outcomes customers value most and will pay forWhich outcomes must keep working through a disruption
MapTrace how the work happens across people, systems, data, partners and customersWhere the value chain could be redesigned or extendedWhere one supplier, person or connection creates an outsized exposure
IdentifyFind opportunities to strengthen the operation and reimagine how work gets doneNew offerings, markets or ways of working worth pursuingWeaknesses the business has learned to tolerate
IdeateDevelop options, including ones that change the operating modelHypotheses about demand, price and channelHypotheses about how the organization would respond

Questions for the leadership team

  1. DemandWhat are customers starting to need that we don’t offer yet?
  2. CombinationWhich of our capabilities would be worth more combined with someone else’s?
  3. ExposureWhere would a single failure hurt customers most?
  4. DriftWhich dependencies have we added without deciding to?
  5. AssumptionsWhat are we assuming will stay the same?

Eudai implementation guide. Related reading on the final page.

eudai.ai8 / 12

Phase 2 · Experiment and learn

Test, measure, learn, and run it again

StepWhat you doInnovation & commercializationRisk & resilience
TestUse pilots, scenarios and small experiments with real people and realistic conditionsPilot the proposition with a small set of customersRun scenarios that put the response under realistic pressure
MeasureAssess performance and impact against what you expectedAdoption, willingness to pay, sales cycle and marginTime to decide, time to recover and repeat friction
LearnApply the evidence to refine the design and decide what to changeRefine the offer, pricing and go-to-marketFix handoffs, ownership and dependencies, then re-test

What makes a useful experiment

  1. 1 SpecificIt tests one assumption the idea depends on.
  2. 2 RealisticIt uses real customers, real people and realistic conditions.
  3. 3 FrequentShort, regular tests teach more than one large pilot.
  4. 4 ObservedSomeone records decisions, objections and handoffs as they happen.
  5. 5 RepeatedThe test returns later to show whether changes held.

Eudai implementation guide. Related reading on the final page.

eudai.ai9 / 12

Phase 3 · Operate, adapt and scale

Run it with confidence, then widen it

StepWhat you doInnovation & commercializationRisk & resilience
OperateBuild what proved itself into routines, ownership and reportingHand the offer to sales, marketing and delivery with clear targetsClear owners, visible status and decisions that don’t wait for a crisis
AdaptAdjust to what operation shows before you scale, and keep reviewing afterwardPricing, packaging and positioning as the market movesNew dependencies, regulation and threats
ScaleExtend what works across teams, markets, sites and partnersNew segments, regions and channels in order of evidenceControls and practice that grow with the operation

Signals that it is time to adapt

  1. New capabilityA new technology or AI capability makes a different offer, way of working or division of work possible.
  2. Market shiftCustomer expectations, competitors or regulation have moved.
  3. New dependencyA new supplier, platform or AI tool has become part of a critical operation.
  4. Changed operationA reorganization, acquisition or new service has moved where value is created.
  5. Repeated frictionThe same objection, handoff or delay keeps appearing.

Eudai implementation guide. Related reading on the final page.

eudai.ai10 / 12

Implementing the model

A first cycle in seven steps

  1. 1Name an executive sponsor and agree the outcomes that matter most
  2. 2Pick one opportunity and the operations it depends on
  3. 3Map value and dependencies with the people who run them
  4. 4Choose the commercial and operational assumptions most likely to be wrong
  5. 5Run a small experiment on each, including how people and AI share the work, and record what happens
  6. 6Make two or three owned changes
  7. 7Re-test, then decide what to scale

Who owns what

RoleOwnsIn practice
BoardOversight of growth and risk togetherAsks for evidence that new initiatives are working and that the exposure they create is understood
Executive sponsorPriorities, funding and removing obstacles across functionsMakes sure innovation and risk teams work from the same evidence
Commercial leadersPropositions, pricing, channels and customer adoptionRun market experiments and own the commercial results
Operations leadersThe operations, their dependencies and changes to themRun operational experiments and own the resulting changes
Process ownersHow work is split between people and AIRevisit the split as capabilities change
Risk and resilience partnersThe method for testing exposure, and the evidenceJoin experiments early and flag risks before they scale

Eudai implementation guide. Related reading on the final page.

eudai.ai11 / 12

Implementing the model · measures and pitfalls

Measure results, not activity

MeasureWhat it tells youHow to read it
Commercial tractionAdoption, revenue and margin from new offersJudge against the hypothesis you tested, not the original forecast
Time to changeHow quickly a tested idea becomes standard practiceShould shorten with each cycle
Repeat findingsWhether the same friction or objection keeps appearingA finding that returns means the change did not hold
Dependency coverageHow much of each critical operation has been mapped and testedGaps usually sit with third parties and informal workarounds
Capability evolutionWhether the organization can now do things it couldn’t beforeHours saved alone won’t show it

Common pitfalls

  1. Separate tracksInnovation and risk working from different evidence and meeting too late.
  2. Launch as proofTreating a rollout as evidence that customers or teams have adopted it.
  3. One big betA single large pilot that is too slow and too visible to learn from.
  4. Skipping discoveryTesting ideas before agreeing where value and exposure sit.
  5. Project mindsetDeclaring it finished when the first cycle ends.
  6. People in the gapsAdding human review to the old process instead of redesigning it.