Navigating how AI will reshape resilient organisations - Rowan Bunning.pdf

Navigating how AI will reshape
resilient organisations
LeSS Conference Singapore
11 September 2025

© 2025 Adapt WithStyle and Rowan Bunning. All rights reserved.
About Rowan Bunning
• Organisational agility consultant.
• I live in Sydney. I’m married with a 5 y.o. + dog + fish.
• 27 years in IT, product development and adult education.
• First 10 years: technical software dev. roles on government
projects, enterprise software and co-founding a start-up.
• 17 years: training and coaching in industries from video games to
financial services inc. $290B/day critical infrastructure.
• Started Agile journey in 2001 with eXtreme Programming.
• Australia’s first Scrum Master in 2003.
• Agile Coach at Ken Schwaber’s European partner (U.K.).
• LeSS Friendly Scrum Trainer teaching Certified LeSS Basics
• Have delivered 550 Scrum Alliance certification courses.
2
2005
2006
2008
2018
2017
2015
2010
2023
Some through:

and LeSS-like adoptions
• 2008 – Thomson Reuters (160 people, LeSS Huge patterns)
• 2016 – Wargaming, now Riot Games (7 teams)
• 2022 – Findex (5 teams, Executive as Product Owner)
less.works/case-studies/other-less-case-studies
3

My Global EMBA journey
4

Adaptability appearances in my EMBA
5
www.hbrreprints.org
A Leader’s Framework
for Decision Making
by David J. Snowden and Mary E. Boone
Wise executives tailor their
approach to fit the complexity
of the circumstances they face.
Reprint R0711C
“Diligence is ongoing and self-
sustaining. It is not something done
once and for all but is repeated and
reinforced over time.”
162
https://doi.org/10.1177/0008125617707975
California Management Review
2017, Vol. 59(3) 162–190
© The Regents of the
University of California 2017
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0008125617707975
journals.sagepub.com/home/cmr
Special Issue on Behavioral Strategy
Strategy as Diligence:
PUTTING BEHAVIORAL
STRATEGY INTO PRACTICE
Thomas C. Powell1
SUMMARY
Researchers in behavioral strategy are producing new insights on strategic decision
making. At the same time, a few pioneering companies are discovering ways to
put behavioral strategy into practice. This article draws on behavioral research
and strategy practice to present an approach called diligence-based strategy. In
markets comprised of people rather than rational economic agents, the analysis of
competitive advantages matters less than the diligent execution of fundamental
activities. Diligence-based strategy offers an applied method for formulating
and executing strategy in organizations, showing how managers can leverage
technology and management discipline to drive business success in the twenty-first
century.
KEYWORDS: diligence, behavioral strategy, strategy process, strategic management,
strategic planning, decision making
I
n 2014, Concha y Toro UK (CyT)—an importer-distributor of wines made
in Chile, Argentina, and California—faced a crisis in competitive strategy.
Global distributors with established brands were moving aggressively into
the U.K. market, smaller entrants were experimenting with new busi-
ness models, and downstream consolidators were shifting the balance of power
to a few large corporate retailers. Confronted with the threat of eroding mar-
ket share, declining profit margins, and an aging business model, CyT executives
knew something had to change.
But CyT did not follow the conventional path for managing large-scale
strategic change. Executives did not articulate a crisis or launch a strategic audit of
market trends or competitive threats, and the company made no attempt to revo-
lutionize its market strategy or business model. Instead, executives turned their
attention to a small number of ordinary business activities such as procuring
1University of Oxford, Said Business School, Oxford, UK
Copyrighted material. Do not distribute. For more information, please contact cmr@haas.berkeley.edu.

Agenda
1. Resilience and beyond
2. AI Adoption Trends
3. Navigating Processes
4. AI Adoption and Org Design
5. Navigating Structures
6. Navigating People
7. Conclusions
6

Caveats
• “Prediction is very difficult, especially if it's about the future!”
• Mainstream management views are are shared not
necessarily because they are correct, but to be aware of what
managers may be influenced by.
• Studies of low variable work may not translate to high
variability work.
• Parallels shared are broad similarities, not to be mistaken for
equivalence.
7

8
We’ll look at
some emerging
solutions from
here. Warnings:
• Small sample
size
• May not
cross the
chasm

Disruption feels real
10
Overview of most impactful threats
F I G U R E 1
Trends expected to be a major or severe disruption for organizations
Share of respondents indicating that a given threat will create major or severe disruption for their organization
Reference survey question:
Which of the current trends or uncertainties do you think will impact your organization the most?
Technology, e.g. cybersecurity, generative Al 52%
Regulatory changes, e.g. policy shifts, compliance 51%
Changing market dynamics and customer preferences 44%
Macroeconomic factors, e.g. trade policies, inflation 43%
Capital and balance sheets, e.g. rising interest rates 43%
People and workforce, e.g. talent shortages 42%
Supply chain disruptions and resource availability 40%
Geopolitics, e.g. international conflicts 37%
Pessimistic growth outlook, e.g. low consumer confidence 32%
Globalization, e.g. instabilities in manufacturing regions 30%
Volatile and increased energy prices 29%
Climate change and rising importance of ESG 27%
Rising demand in the industry (including due to economic recovery)
Misinformation and disinformation 27%
Societal and political polarization 19%
Others 4%
29%
Source: World Economic Forum.

Robustness
11

Robustness
12
robustness proves inadequate and organisational performance begins to suffer, resilience steps
in to prevent complete collapse and ultimate failure, allowing the organisation to recover when
Source: Hillson (2023)

13

14
performance as lessons are learned and improved practices are implemented (both
represented by the grey areas in Figure 2).
Figure 2.
Resilient response to
different stress
scenarios
Resilience (bounce back)

Resilience is increasingly relevant in business
“Call it the resilience gap. The world is
becoming turbulent faster than
organisations are becoming resilient.”
(Hamel 2003)
15

Resilience in software systems
16

Definitions of resilience have been evolving
• 1998 - Mallak: “the ability of an organization to absorb stress,
recover critical functionality, and thrive in altered
circumstances.”
• 2017 - ISO: Resilient organisations “can absorb and adapt to
the changing business environment while continuing to deliver
on the objectives that enable survival and prosperity.”
• 2022 – World Economic Forum: ”the ability to deal with
adversity, withstand shocks, and continuously adapt and
accelerate as disruptions and crises arise… it is also the ability
to reinvent and innovate in response to disruptions.”
”Resilience” has become increasingly about adaptability and
leveraging turmoil for advantage.
17
Bounce back
Bounce forward

“Strategic resilience”
Strategic resilience is not about responding to a
one-time crisis… It’s about continuously
anticipating and adjusting to deep, secular
trends that can permanently impair the
business. It’s about having the capacity to
change before the case for change becomes
desperately obvious.”
– Gary Hamel
18

Antifragility
19
“An antifragile system
learns, grows and self-
improves when exposed to
error, failure, noise,
uncertainty, randomness or
attack”
(Benam, 2022).

Resilience
20
which benefit from stress, as illustrated in Figure 4.
Figure 4.
Antifragile responses
to different stress
scenarios
of performance impairment is somewhat proportional to the level of stress, and it persists
while the stressor remains, before rising to previous levels as stress falls away. This is true
whether the stress occurs in a single pulse or over a more sustained period. For this reason,
the profile of performance level appears inversely correlated with the stressor profile, albeit
not the perfect inverse, and with some delay before the effect is evident. Although full
performance is often regained, some residual impairment might result; alternatively, the
possibility of “bouncing forward” offers the prospect of recovering to a higher level of
performance as lessons are learned and improved practices are implemented (both
represented by the grey areas in Figure 2).
vs. Antifragility

Optimising Goals Compared
21
Theterm antifragility was coined byNassim Nicholas Talebin hisbook Antifragile: Things that
gain from disorder (Taleb, 2012). Interestingly, and somewhat frustratingly, the book contains
E
robustne
and

Team skills
E F G
Product Backlog
D E F
skills required
Team skills
A B C
A B C
skills required
Team skills
C D E
d
Will slow down as they learn a bit of d.
Principles:
• Whenever we can use specialisation to maximum advantage, we use it.
• Whenever specialisation becomes a constraint, we break the constraint.
• Better to do a high value thing slower than a low value thing fast.
• Expanding team skills aids adaptability and working in value order.
Do we have stressors in ?
Yes. Teams switching beyond current skills / knowledge.

Strategy #1: Sensing shifts
23
Market control
Bureaucratic
control
Clan control
• self-managing teams
• self co-ordination
• decisions at level of richest
information
PO
≪component≫
Publishing
≪component≫
Scheduling
≪component≫
Expenses
≪component≫
KPI Dashboards
☸
AI contributions
• Early weak signal
detection
• Noise reduction
• Contextual sense-making
• Pattern surfacing at
scale
• Scenario generation
Product Owner with high domain engagement
Business Domain

Strategy #2: Decentralisation
24
• Distributed sensing
• Diversity of AI agents
• Decision making at
the edge
• Autonomy with
guardrails
• Knowledge flow
without bottlenecks
AI contributions

Strategy #3: Redundancy
25
AI contributions
• Model ensembling
• Parallel sensing
• Diverse perspectives
• Adaptive failovers
Item 1
Item 2
Item 3
...
Item 8
…
Item 12
Team
Wei
Team
Shu
Team
Wu
Component
A
Component
B
Component
C
With feature teams, teams can always work on the highest-value features, there is less delay for
delivering value, and coordination issues shift toward the shared code rather than coordination
through upfront planning, delayed work, and handoff. In the 1960s and 70s this code coordination
was awkward due to weak tools and practices. Modern open-source tools and practices such as
TDD and continuous integration make this coordination relatively simple.
system
www.craiglarman.com
www.odd-e.com
Copyright © 2010
C.Larman & B. Vodde
All rights reserved.

Strategy #4 started with 648 experiments!
2008 2010
•Systems Thinking
•Lean Thinking
•Queueing Theory
•False Dichotomies
•Be Agile
•Feature Teams
•Teams
•Requirement Area
•Organisation
•Large-Scale Scrum
•Scrum
•Large-Scale Scrum
•Test
•Product Management
•Planning
•Coordination
•Requirements & PBIs
•Design & Architecture
•Legacy Code
•Continuous Integration
•Inspect & Adapt
•Multisite
•Offshore
•Contracts
AI contributions
• Scaled, very low-cost
experimentation
• Improved experiment
quality
• Meta-learning
• Stress-testing and
amplification

Strategy #5: Knowledge retention
27
“In the antifragile organisation, organisational memory is an
important contributor to the persistence of long-term benefits and
permanent performance enhancement after the current stressor is
no longer present. This is best achieved by ensuring the
development of a learning organisation.”
(Hilson 2023)
AI contributions
• AI training and retention
• Stable teams
• Multilearning inc. domain
• Learning organisation

Summary - Strategies for antifragility
28
Antifragility
strategy
LeSS strategies AI Contributions
Sensing shifts Product Owner with high
domain engagement
• Early weak signal detection
• Pattern surfacing at scale
• Noise reduction
• Contextual sense-making
• Scenario generation
Decentralisation De-centralised coordination
patterns
• Distributed sensing
• Diversity of AI agents
• Decision making at the edge
• Autonomy with guardrails
• Knowledge flow without bottlenecks
Redundancy Feature teams • Model ensembling
• Parallel sensing
• Diverse perspectives
• Adaptive failovers
Ecosystem of
experiments
Experiments in books 1 & 2
Sprint Retro experiments
Overall Retro experiments
• Scaled, very low cost experimentation
• Improved experiment quality
• Meta-learning
• Stress-testing and amplification
Knowledge
retention
Learning Organisation • AI training and retention

Types of organisational antifragility
1. Innate: empowers individuals to respond immediately whenever
stress arises, adopting generic responsive postures that build
strength and improve performance.
2. Adaptive: specific and targeted responses that tackle the
particular nature of the current stressor, retaining the capability
to respond immediately if a similar situation recurs.
3. Rheopectic: forms structures that were not previously possible,
including strengthened business processes and novel working
relationships, within a learning context that ensures any
beneficial changes become embedded permanently.”
4. Emergent: arises organically from development within an
environment and culture that supports appropriate risk-taking,
enabling it to take the right risks safely.
(Hilson 2023)
29

30
Types of Anti-fragility

Rheopectic antifragility through Communities
“Rheopectic antifragile organisations
take advantage of increasing pressure to
form structures that were not previously
possible, including strengthened business
processes and novel working
relationships, within a learning context
that ensures any beneficial changes
become embedded permanently.”
(Hillson 2023)
31
Guide: Community Work
A community is a group of
volunteers from the teams who share
an interest or topic and have the
passion to deepen their knowledge
or take action through discussion
and interaction with peers.
Participation in communities is
completely voluntary.
Communities create informal
networks between teams that are
essential for learning, coordination,
and continuous improvement.
(Larman 2016)

32
Google trend on resilience related terms

Use case impact
34
Exhibit 3
Web <2023>
<Vivatech full report>
Exhibit <3> of <16>
Using generative AI in just a few functions could drive most of the technology’s
impact across potential corporate use cases.
McKinsey & Company
Note: Impact is averaged.
¹Excluding software engineering.
Source: Comparative Industry Service (CIS), IHS Markit; Oxford Economics; McKinsey Corporate and Business Functions database; McKinsey Manufacturing
and Supply Chain 360; McKinsey Sales Navigator; Ignite, a McKinsey database; McKinsey analysis
Impact as a percentage of functional spend, %
Impact, $ billion
Marketing
Sales
Pricing
Customer operations
Corporate IT1
Product R&D1
Software engineering
(for corporate IT)
Software engineering
(for product development)
Supply chain
Procurement management
Manufacturing
Legal
Risk and compliance
Strategy
Finance
Talent and organization (incl HR)
0 10 20 30 40
0
100
200
300
400
500
Represent ~75% of total annual impact of generative AI

Poll: AI adoption attitude
Q: Which attitude to AI adoption does your org. have?
• A: Front-runners for AI adoption. Confident and aligned on
strategy, with AI processes currently in place to primarily grow
the company, all with minimal difficulty.
• B: Understand the need to adopt AI, mainly to stay
competitive, but experiencing challenges in integrating it
across the business.
• C: Some individuals have a general understanding of AI, but it
is not their main priority, mainly due to the lack of leadership
buy-in. Currently experimenting with AI or exploring options.
• D: Don’t understand the role AI could play in the business
currently. Intend to use it, but don’t know where to start.
35

37
AI Adoption Archetypes
A B
C
D

38
Study results in Australian SMEs
17% 24%
36%
23%

Agentic AI is at the top of the hype cycle…
39
(Gartner 2025)

Yet 81% of orgs are conservative or not investing
40
(Financial Times 2025)

Poll: Primary AI adoption motivation
Q: What is the primary motivation for AI adoption in
business?
a) Become more flexible
b) Make better decisions
c) New revenue
d) Faster reporting
e) Reduce costs
41

the most popular AI
adoption goal is to
reduce costs.
Hypothesis: We believe that…
43
Test Card
We believe that
step 1: hypothesis
And measure
step 3: metric
To verify that, we will
step 2: test
We are right if
step 4: criteria
The makers of Business Model Generation and Strategyzer
Copyright Strategyzer AG
Test Cost: Data Reliability:
Critical:
Time Required:
Test Name
Assigned to
Deadline
Duration
Test Card: https://www.strategyzer.com/library/validate-your-ideas-with-the-test-card

Answer in Australia:
Reduce costs
44
(Decidr 2025)

Poll: AI adoption payoff curve
46
Augmentation Semi-autonomy Full autonomy
Overall payoff
B) Concave
C) Convex
A) Linear
Extent of adoption
Q: Which shape is the AI
adoption payoff curve?
a) Linear
b) Concave (fast then
slow increase)
c) Convex (slow then fast
increase)

AI adoption payoff curve
48
Augmentation Semi-autonomy Full autonomy
Overall payoff
C) Convex
Extent of adoption
Answer: Convex
Orgs that “prioritize core
function transformation
over diffuse productivity
gains” see far greater
returns (Lee 2022)
74% of orgs had yet to see
meaningful returns(BCG
2024)

Poll: How much is current use of AI in software
dev. improving overall performance*
* Overall performance = overall throughput, DORA metrics and
quality KPIs.
Options:
• A) 1,000%
• B) 100%
• C) 50%
• D) 20%
• F) 0%
• G) -20%
50

AI is improving software
development performance at a
product group level
52

Study: No Measurable Organisational Impact
“When viewed at the company level, these
team-level gains do not scale. Correlations
between AI adoption and organization-wide
delivery metrics are weak or non-existent.
Faros AI analyzed telemetry from 1,255 teams and over 10,000
developers.
53
The AI Productivity Paradox:
AI Coding Assistants Increase
Developer Output, But Not
Company Productivity
What Data from 10,000
Developers Reveals about
Impact, Barriers, and the
Path Forward
Insights from Faros AI’s analysis of 1,255
teams across leading software organizations
July 2025
(Faros 2025)

Poll: Why no improvement
Q: Why no overall improvement?
A) Culture of scepticism about AI
B) Downstream bottlenecks
C) Insufficient leadership support
D) AI tools not yet mature enough
54

Downstream bottlenecks
“When viewed at the company level,
these team-level gains do not scale.
Correlations between AI adoption
and organization-wide delivery
metrics are weak or non-existent.
This is due to uneven adoption,
workflow bottlenecks, fragmented
tooling, and the lack of
coordinated enablement
strategies.”
56
1. Downstream bottlenecks cancel out gains. AI accelerates code generation, but review, testing, and
deployment stages have not kept pace. Increased PR volume overloads reviewers, test frameworks
remain brittle, and release pipelines are rarely optimized for higher velocity. According to Amdahl’s
Law, the system's speed is constrained by its slowest link—without lifecycle-wide modernization,
AI’s beneﬁts are quickly neutralized.
(Faros 2025)

10 units / day
Faster AI steps = bigger downstream queues
57
Before AI adoption
After local AI adoption
Step 1 e.g. code Step 2 e.g. integration test
1 unit / day
😣

A Value Ratio of 5% is typical in development
“We have done many timelines with product development groups and have
not seen a value ratio in a development organisation higher than 7 percent. In
other words, 93 percent or more of the time in development was waste time.”
This is consistent with observation by others, such as (Ward06) who estimate
an average 5% value ratio in product development.”
(Larman 2008)
58

Scenario question
Assume a value ratio of 5%
(i.e. 95% of lead time is idle and non-value-adding process time)
You put a lot of effort into implementing AI. The result is a 50%
reduction to value added time.
Q: What improvement have you made to lead time?
a) 100%
b) 50%
c) 5%
d) 2.5%
59

Bottom-up AI adoption dynamics
60
Goal: bottom-up AI adoption (without whole system goal and design).
perception of overall
improvement from
local improvement
investment in local
optimisation
overall improvement
localised measures
and incentives
# of local
optimisations
Belief: local
improvement
contributes to global
improvement
↑
O
Reinforcing
loop
☹

An overall navigation approach
1. Align on what you’re optimising for.
2. Sense emerging tensions.
3. Generate and evaluate options using thinking tools.
4. Reflect on what we’ve learnt previously.
5. Consider emerging solutions (e.g. from early adoptors)
6. Experiment and iterate, evaluating relative to the whole.
62

"Adaptiveness is not a self-licking icecream
cone!"
In political jargon, a self-licking ice
cream cone is a self-perpetuating
system that has no purpose other than
to sustain itself. - Wikipedia
63
Performance characteristics
What performance characteristics do
we need to succeed?
CAPABILITIES
/ OPTIMISING
GOALS
STRATEGY
Direction

Example Optimising Goals (1 of 3)
64
Output Predictability
Maximising the occurrence of
speciﬁc outputs being delivered
when expected by stakeholders.
Working to Priority
Maximising the occurrence of
everyone working on the highest
priority items for the product or
service at all times.
Releasability
Maximising the ability to release
for use sustainably at target
quality level at the timing
preferred by stakeholders.
Utilisation
Maximising the proportion of
available time that people are
occupied by tasks, thus pre-
emptively minimising idle time.
Throughput
Maximising the output that our
organisation delivers within a
specified period.
Responsiveness
Minimising the time taken to
respond to stakeholder requests.
Customer Value
Maximising the benefit that the
customer can realise from the
product or service relative to
what they give up in order to
receive it.

65
Adaptiveness of the
Product / Service
Maximising the capability to nimbly
adjust the content of the product or
service.
Adaptiveness of the Org.
Maximising the capability to quickly
and efficiently change all org. elements
(structures, processes, rewards, people
practices) to align with the business
strategy.
Innovation
Maximising the probability of
discovering novel opportunities or
solutions to create value beyond what
was previously expected.
Engagement
Maximising employee involvement
in, enthusiasm about and
commitment to their work and
workplace.
Internal Stakeholder
Satisfaction
Maximisation of positivity from people
inside our organisation with significant
interests in our endeavour, relative to
what they give up in order to receive it.
OrganisaEonal Learning
Maximising generaBon and
disseminaBon of new and beCer
knowledge and understanding in order
to improve acBons.
Role & Career Path
Security
Maximising how secure people feel
about the opportunity to conBnue in
their role and career path.

66
Compliance
Maximising the obedience of people
in following policies, procedures and
directives from those in authority.
Confidentiality
Minimising the risk of sensitive
information being disclosed in a way
that can damage our business and/or
reputation.
Quality for the Customer
Minimising customer detected defects
and other quality deviations below
expected.
Opera3onal Reliability
Maximising the consistency of
operaBonal service provision to external
customers at the expected service level,
even when there is variability in demand
and other factors.
Opera3onal Cost
Minimising the cost of ongoing business
operaBons without sacriﬁcing quality of
the product or service.
Specialist Expertise
Maximising the depth of expertise that
individual specialists hold.

What should be stable, what highly adaptable?
• Optimising goal
• Perfection vision
• Feature team
structure
• Core socio-technical
practices
• Product definition
• Requirement areas
• Problem spaces
worked on
• Components worked
on
• Knowledge sharing
• Agreement creation
Relatively stable Gradually adaptable Highly adaptable

An example tension?
“A computer can never be held accountable, therefore a
computer must never make a management decision.”
(IBM 1979)
68

Galbraith’s Star
Model
69
we need to succeed?
STRATEGY
CAPABILITIES
/ OPTIMISING
GOALS
STRUCTURE
PEOPLE
PRACTICES
METRICS /
REWARDS
PROCESSES
Power
What should the formal structure be?
What are the key roles?
How should power be distributed?
Information
How are decisions made?
How does work flow?
What management processes?
Motivation
What incentives will drive the right
behaviour?
What measures should we focus on?
Skills and mindsets
What talent is needed?
What people practices are
key to our capabilities?
Effectiveness requires ALIGNMENT
Culture, Performance, Results
Direction
POLICIES
(Galbraith 1977)

What have we learnt
from failed “Agile”
adoptions?
70
we need to succeed?
STRATEGY
CAPABILITIES
/ OPTIMISING
GOALS
STRUCTURE
PEOPLE
PRACTICES
METRICS /
REWARDS
PROCESSES
Power
What should the formal structure be?
What are the key roles?
How should power be distributed?
Information
How are decisions made?
How does work flow?
What management processes?
Motivation
What incentives will drive the right
behaviour?
What measures should we focus on?
Skills and mindsets
What talent is needed?
What people practices are
key to our capabilities?
Culture, Performance, Results
Direction
Most partial “Agile” changes
have created misalignment
Friction
POLICIES
Effectiveness requires ALIGNMENT (Galbraith 1977)

Poll: AI direct impact on org design?
Q: Which element do current AI adoptions directly alter
most?
a) Strategy (direction)
b) Structure (power)
c) Processes (information)
d) Rewards (motivation)
e) People (skillsets/mindsets)
71

Similarly, AI
adoption can
destabilise!
73
Strategy
Structure
Changed
Processes
Rewards
People
skillsets,
mindsets
direction
power,
decision
authorities
information
motivation,
incentives
Alignment =
Effectiveness
Culture
Policies
AI
Friction
F
r
i
c
t
i
o
n
Friction
Friction

What will local AI assistants into this achieve?
Team 1 Team 2 Team 3 Team 4 Team 5 Team 6 Team 7 Team 8
Manager
End customers/users Other stakeholders
Quoting
Leads and
Opportunities
Premium
calculation
iOS app CRM
Data
warehouse
Android
app
PMO / Tribe

A) Compromising the parts for the whole
or
B) Compromising the whole for the parts?
Is your organisation…
76

AI adoption in pockets doesn’t improve the whole
77

non-systemic AI adoption will
amplify weaknesses in your
Organisational Design!
78

Poll: AI and team size
Q: Deep adoption of AI will lead to the number of humans in
product development teams:
a) Becoming larger
b) Staying about the same size
c) Becoming a little smaller (e.g. 4-5)
d) Becoming very small (e.g. 2-3)
80

Tiny teams
“An AI-native startup run by one or a few people using
automation to deliver results at scale.”
Drivers:
• Accessible AI tech
• Prompt and context engineering
• No/low code with AI
• AI agents
• Founder priorities shifting to flexibility
• Focus on cost-efficiency and remote work
82
Source: Money University – The AI Powered Tiny Team.
Human
headcount
Revenue
per
employee
Relative
value

Example tiny teams leveraging AI
83
Cursor
12 people
$100M ARR
Bolt
15 people
$20M ARR in 60 days
Gumloop
9 people
Goal: $1B ARR
Gamma
30 people
$50M ARR
Oleve
4 people
5M users
$6M ARR
youtube.com/watch?v=xhKgTkzSmuQ

Poll: AI and middle management
Q: AI adoption will lead to the number of middle managers:
a) Increasing
b) Staying about the same
c) Decreasing a little
d) Decreasing a lot
84

What mainstream
management literature is
saying
86
5% on coding, 10% less on project management
“First, using gen AI allowed coders to work more
autonomously, and because they were
collaborating less, it reduced the need for them or
their managers to coordinate tasks or projects with
others.
As gen AI becomes better at scheduling,
coordinating, and checking quality, the remaining
managers should be free to do more valuable
work, including pitching in on the hands-on tasks
typically done by individual contributors (in a
software environment, this might be coding).”
(HBR 2025)

AI descaling middle management
“Gen AI can, in some instances, replace a
manager, mentor, or any other individual
that junior employees may ask for help,"
“We will likely see an increase in agility in
companies that use gen AI,” Hoffmann
says. “This will lead to the flattening of
corporate hierarchies, which will help
streamline productivity and reduce the
need for so many middle management
tasks.” (Hoffman 2025)
87
Working Paper 25-021
Generative AI and the Nature of Work
Manuel Hoffmann
Sam Boysel
Frank Nagle
Sida Peng
Kevin Xu
also

Structural parallels
88
AI native startups
Multi-learning Generalists
Go See Player-coach
Sprint Review Show and tell

Widespread thinking in tech. companies
90
Win with Generalists Win with Specialists
Maturity
Stage
AI as an API
Data Loops
Training Layer
Win on Modelling
(AI Engineer - - Linkov, 2025)
Concerns: “exhaust the knowledge”
”mature orgs need to be the best in the world in something specific”

Which direction can AI help with most?
91
less
less
SCOPE
OF
WORK
MANDATE
SCOPE OF SKILLS MANDATE more
more
increasing
Scope of
Work
Mandate
UNBOUNDED
TASKS
CAPABILTIES
PARTIAL
BUSINESS
WHOLE
BUSINESS
less SCOPE OF SKILLS MANDATE more
FUNCTIONAL END-TO-END EXPANDING UNBOUNDED
MULTI-SKILL
increasing
Scope of
Skills
Mandate
(Flemm 2025)

Poll: AI and team mandates
Q: AI adoption most presents the opportunity to:
a) Increase scope of skills mandate only (move to right)
b) Increase scope of work mandate only (move upward)
c) Increase both scope of skills and work mandates (move
upward to the right)
d) None of these
92

Alignment with… small cross-functional teams
If the design is very small cross-functional teams with higher-
order cognitive skills, what does that mean for….
• skills development?
• mindset?
• the line organisation?
• incentives?
95
Strategy
Structure
Processes
Rewards
People
Policies
Small, Cross-
functional

Poll: Degree of specialisation
Q: With AI adoption, will the degree of specialisation will
become:
a) Narrower and deeper
b) No significant change
c) Broader and less deep
96

GenAI retrieval is not learning
98
Aspect Using AI (quick info) Genuine learning
Cognitive engagement
Quick answers; minimal thinking
needed.
Requires deep thinking and
problem-solving.
Self-reflection & planning
Little self-check; trust AI answers
without questioning.
Plan learning, monitor
understanding, and adjust
approach.
Social interaction
Mostly solo; no discussion or
feedback.
Collaborative: ask questions, discuss,
and get feedback.
Applying knowledge
Info is given; you may skip
practice.
Hands-on practice: apply
knowledge in real tasks.
Depth of understanding
Surface-level facts that may not
stick.
Deep understanding: concepts
connect and stick.

LeSS encourages: Multilearning
“Such learning by doing manifests itself along two dimensions:
across multiple levels (individual, group, and corporate) and
across multiple functions. We refer to these two dimensions of
learning as “multilearning.”
99
137
The new new
product development
game
Stop running
the relay race and
take up rugby
Hirotaka Takeuchi and
Ikujiro Nonaka
In today's fast-paced, fiercely competitive
world of commercial new product development, speed
and flexibility are essential. Companies are increasingly
realizing that the old, sequential approach to developing
new products simply won't get the job done. Instead, com-
panies in Japan and the United States are using a holistic
method—as in rugby, the ball gets passed within the team
as it moves as a unit up the field.
This holistic approach has six characteris-
tics: built-in instability, self-organizing project teams,
overlapping development phases, "multilearning," subtle
control, and organizational transfer of learning. The six
pieces fit together like a jigsaw puzzle, forming a fast and
flexible process for new product development, fust as im-
portant, the new approach can act as a change agent: it is a
vehicle for introducing creative, market-driven ideas and
processes into an old, rigid organization.
Mr. Takeuchi is an associate professor and
Mr. Nonaka, a professor at Hitotsubashi University in fa-
pan. Mr. Takeuchi's research has focused on marketing and
global competition. Mr Nonaka has published widely in
Japan on organizations, strategy, and marketing.
The rules of the game in new product
development are changing. Many companies have dis-
covered that it takes more than the accepted basics
of high quality, lov^ cost, and differentiation to excel in
today's competitive market. It also takes speed and
flexibility.
This change is reflected in the emphasis
companies are placing on new products as a source of
new sales and profits. At 3M, for example, products less
than five years old account for 25%* of sales. A 1981
survey of 700 U.S. companies indicated that new prod-
ucts would account for one-third of all profits in the
1980s, an increase from one-fifth in the 1970s.'
This new emphasis on speed and flexi-
bility calls for a different approach for managing new
product development. The traditional sequential or
"relay race" approach to product development-
exemplified by the National Aeronautics and Space
Administration's phased program planning (PPP)
system-may conflict with the goals of maximum
speed and flexibility. Instead, a holistic or "rugby"
approach-where a team tries to go the distance as a
unit, passing the ball back and forth-may better serve
today's competitive requirements.
Under the old approach, a product de-
velopment process moved like a relay race, with one
group of functional specialists passing the baton to the
next group. The project went sequentially from phase
to phase: concept development, feasibility testing,
product design, development process, pilot produc-
Authors' note: We acknowledge the
contribution of Ken-ichi Imai in the
development of this article. An earlier
version of this article was coauthored by
Ken-ichi Imai, Ikujiro Nonaka, and
Hirotaka Takeuchi. It was entitled
"Managing the New Product Development
Process: How Japanese Companies
Learn and Unlearn" and was presented
at the seventy-fifth anniversary
Colloquium on Productivity and
Tbchnology, Harvard Business School,
March 28 and 29,1984.
1 Booz Allen & Hamilton survey
reported in Susan Fraker,
"High-Speed Management
for the High-Itch Age,"
Fortune,
March 5,1984, p. 38.
(Takeuchi 1986)

Task stack model
100
Apex
Tasks
- human only
Advanced tasks
– human-led with
AI support
Mid-level tasks –
augmented by AI
Foundation tasks
– fully automated
AI
Adoption
in
organisations
e.g. boilerplate code
documentation
e.g. refactoring, API design,
legacy code translation
e.g. system architecture design,
integration of disparate systems
e.g. nuanced business requirements,
creative problem solving,
cross-functional collaboration
Synthesized from (Acemoglu 2011)

Progressive AI job overlap
101
Apex
Tasks
- human only
Advanced tasks
– human-led with
AI support
Mid-level tasks –
augmented by AI
Foundation tasks
– fully automated
AI
Adoption
in
organisations
Job A
Job B
Job C
Synthesized from (Acemoglu 2011)

Consider the way manager roles have changed
102
(Larman 2016)

Policy change through trust building
1. Identify who has authority to change the policy
2. Ask “what would need to be true for this policy to change?”
3. Agree a measure and threshold:
e.g. the escaped defect rate is as low as it is currently,
sampled from >20 releases.
4. Create an incremental roadmap toward that with interim
measures and thresholds.
5. Quantify the compounding payoff from the policy change to
strengthen the investment case for improvement.
6. Use: Observability, Evals, Fuzz testing etc.
104

Use an incremental trust building roadmap
105
Complete
observability
Error rate
<5%
Error rate <
with people
Indicative example:
Sufficient
for policy
change

Thinking tools useful for systemic AI adoption
• Systems thinking
• Optimisation & perfection vision
• Casual Loop Diagramming
• Lean
• Value stream modelling
• Flow
• Organisation Design
• Galbraith’s Star Model
• Org Topologies
• LeSS principles
107

Key messages
1. The term “Resilience” has become increasingly about adaptability and leveraging turmoil for
advantage.
2. With LeSS, we have strategies aligned with antifragility.
3. Take a Systemic approach to AI adoption.
4. Organisational AI results that matter require our skills in org. design, systems thinking and
Lean to re-align:
• Processes
• Structures
• Rewards
• People, and
• Policies.
5. AI benefits require experimentation, validation, tuning and trust-building from a whole-of-
system perspective.
108

References (1 of 3)
• 2017, I. S. O. (2017). [C] Security and Resilience‐Organizational Resilience‐Principles and Attributes.
https://www.iso.org/standard/50053.html
• AI Engineer - Linkov. (2025, July 24). Structuring a modern AI team — Denys Linkov, Wisedocs. Retrieved 2025-09-13 from
https://www.youtube.com/watch?v=SbUxRluVRwk
• AI Engineer (2025). AI Engineer World’s Fair 2025 - Tiny Teams : YouTube. Retrieved 2025-09-13 from
https://youtube.com/watch?v=xhKgTkzSmuQ
• Acemoglu, D., & Autor, D. (2011). Skills, tasks and technologies: Implications for employment and earnings. In O. Ashenfelter &
D. Card (Eds.), Handbook of labor economics (Vol. 4, pp. 1043–1171). Elsevier.
https://doi.org/10.1016/S0169-7218(11)02410-5
• Bland, D. J., & Osterwalder, A. (2019). Testing Business Ideas: A Field Guide for Rapid Experimentation. John Wiley & Sons.
http://books.google.com.sg/books?id=8smxDwAAQBAJ&hl=&source=gbs_api
• Chui, M., Hazan, E., Roberts, R., Singla, A., & Smaje, K. (2023). The economic potential of generative AI.
https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai-the-next-
productivity-frontier
111

References (2 of 3)
• Decidr. (2025). National AI readiness index report.
https://www.decidr.ai/national-ai-readiness-index-report-2025
• Dunlap, N. (2025). The AI Productivity Paradox Report 2025.
https://www.faros.ai/blog/ai-software-engineering
• Times, F. (2025). Agentic AI - how bots came for our workflows and drudgery FT Working It [video]. Retrieved 2025-09-13
from
https://www.youtube.com/watch?v=e85AxYW0Qyk&t=258s
• Mundial, F. E. (2025). Resilience Pulse Check: Harnessing Collaboration to Navigate a Volatile World.
https://es.weforum.org/publications/resilience-pulse-check-harnessing-collaboration-to-navigate-a-volatile-world/
• Galbraith, J. R. (1977). Organization Design. Prentice Hall
http://books.google.com.sg/books?id=Bz-3AAAAIAAJ&hl=&source=gbs_api
• Gartner. (2025). The 2025 Hype Cycle for Artificial Intelligence Goes Beyond GenAI.
https://www.gartner.com/en/articles/hype-cycle-for-artificial-intelligence
• Review, H. B. (2025). How AI Is Redefining Managerial Roles.
https://hbr.org/2025/07/how-ai-is-redefining-managerial-roles
• Hillson, D. (2023). Beyond resilience: towards antifragility? Continuity & Resilience Review.
https://doi.org/10.1108/crr-10-2022-0026/full/html
112

References (3 of 3)
• Hoffmann, M., Boysel, S., Nagle, F., Peng, S., & Xu, K. (2024). Generative AI and the Nature of Work.
https://www.econstor.eu/handle/10419/308375
• Krivitsky, A., Flemm, R., & Larman, C. (2025). The Org Topologies Primer.
https://www.orgtopologies.com/
• Larman, C., & Vodde, B. (2008). Scaling Lean & Agile Development: Thinking and Organizational Tools for Large-Scale Scrum.
Pearson Education
https://play.google.com/store/books/details?id=HbRo4kYnTnMC&source=gbs_api
• Larman, C., & Vodde, B. (2016). Large-Scale Scrum: More with LeSS. Addison-Wesley Professional
https://play.google.com/store/books/details?id=KivKDAAAQBAJ&source=gbs_api
• Mallak, L. (1998). Putting organizational resilience to work. INDUSTRIAL MANAGEMENT-CHICAGO THEN ATLANTA, 8–13.
• University, M. (2025). The AI Powered Tiny Team Money Podcast. Retrieved 2025-09-13 from
https://www.youtube.com/watch?v=4Ed93oCNA7Y
• Takeuchi, H., & Nonaka, I. (1986). The new new product development game. Harvard business review, 64(1), 137–146.
https://hbr.org/1986/01/the-new-new-product-development-game
• Taleb, N. N. (2012). Antifragile: How to Live in a World We Don’t Understand.
http://books.google.com.sg/books?id=hN9puwAACAAJ&hl=&source=gbs_api
• Vlikangas, L., & Hamel, G. (2003). [C] The quest for resilience. Harvard Business Review.
https://hbr.org/2003/09/the-quest-for-resilience
113

Navigating how AI will reshape resilient organisations - Rowan Bunning.pdf

Navigating how AI will reshape resilient organisations - Rowan Bunning.pdf

More Related Content

Similar to Navigating how AI will reshape resilient organisations - Rowan Bunning.pdf

More from Rowan Bunning

Recently uploaded

Navigating how AI will reshape resilient organisations - Rowan Bunning.pdf