Executive Summary
AI doesn’t fix broken organizations—it magnifies them. Rich Allen, author of User Needs Mapping, reveals why 95% of AI implementations fail not because of technology limitations, but because organizations deploy powerful tools on top of dysfunctional structures. The pattern is stark: clear team boundaries aligned around user needs enable AI to reduce toil and amplify effectiveness. Confused structures with blurred ownership cause AI to multiply dysfunction.
Core insights:
- AI amplifies whatever organizational design you already have—excellence or dysfunction
- Organizations with “outside in” thinking and clear value delivery thrive with AI; those with unclear structures struggle to know where AI belongs
- Code acceleration solves the wrong problem—code was never the bottleneck in the first place
- Most AI deployments generate more content that doesn’t serve user needs
- Team Topologies provides “infrastructure of agency” for both human and AI effectiveness
- Winners ask: “Where does this help users make effective progress?” before deploying AI
- Cloud adoption followed the same pattern—piling new technology on existing dysfunction made things worse, not better
- Without clarity on team types and interactions, organizations use AI as a solution looking for a problem
- The 80/8 Delivery Gap persists: 80% of companies believe they excel at customer value; only 8% of customers agree
The AI Implementation Paradox
Your organization is racing to adopt AI. Leadership workshops explore use cases. Pilots launch across departments. Everyone fears being left behind.
Yet MIT research reveals a sobering reality: 95% of AI implementations fail to meet their vision.
The common explanation focuses on technology maturity, data quality, or use case selection. But Rich Allen, author of User Needs Mapping, identifies the real issue: “AI is very much an amplifier. It will amplify the dysfunction within an organization.”
This isn’t metaphor—it’s mechanism. Organizations with clear team boundaries, aligned around user needs, and understanding of how they deliver value deploy AI and see reduced toil, faster delivery, better outcomes.
Organizations with unclear structures, blurred ownership, and confused interactions deploy AI and see multiplied dysfunction, higher costs, slower teams.
Same technology. Opposite results.
The difference isn’t the AI. It’s the organizational design.
Why "Outside In" Organizations Thrive Whilst "Inside Out" Organizations Struggle
“The organizations that are already ‘outside in’ thinking and aligning their teams in a way that delivers value effectively will thrive with AI,” Rich explains, “because they already have a clear understanding of how they deliver value and therefore they can recognize the boundaries in which AI can support them.”
The critical question becomes: “Where does it make sense for AI to support us delivering value to this user need?”
This framing—starting with user needs, then identifying where AI adds value—inverts the typical approach. Most organizations start with AI capabilities and search for problems to solve. “I’ve got a tool and I need to find a problem for this solution,” as Rich puts it.
But organizations that lack clear understanding of team structures and how they interact “are potentially going to struggle with where AI can be effective.” Without clarity on users, needs, and value streams, every AI deployment becomes guesswork.
This connects directly to research Mark shares from Bain: 80% of companies believe they excel at customer orientation and delivering customer value. Yet when Bain asked their customers, only 8% agreed their suppliers excelled at delivering value.
This massive “Delivery Gap”—from 80% to 8%—reveals something profound. Organizations believe they understand and serve users effectively. Users experience something quite different.
AI deployed into this gap doesn’t close it. AI amplifies it.
The Code Generation Illusion: Solving the Wrong Problem
Consider what’s happening in software development—the most visible arena for AI adoption.
“AI is accelerating the development of code,” Rich observes. “But code was never really the problem in the first place.”
The bottleneck in software delivery isn’t typing speed or syntax knowledge. It’s understanding what to build, for whom, and why. It’s managing dependencies between teams. It’s clarifying ownership and reducing context switching. It’s aligning around user needs rather than internal functions.
AI that accelerates code generation whilst leaving these structural issues untouched creates a predictable outcome: “We’re just generating more content, which isn’t actually necessarily serving or meeting user needs.”
More code. Faster deployment. Same misalignment.
“The ones that will win,” Rich argues, “are the ones that are really considering where does this really help users make effective progress?”
Team Topologies: The Infrastructure of Agency
Stuart Winter-Tear coined a phrase that captures why organizational design matters so profoundly for AI success: “Team Topologies is the infrastructure of agency.”
“By using these ways to articulate how we’re structured—our team types and how they interact—that will enable agency, whether it’s human agency or AI agency, way more effectively,” Rich explains.
This reframes the AI conversation entirely. Before asking “What can AI do?”, organizations must ask “How do our teams work?” Before deploying AI capabilities, organizations must clarify team boundaries, interaction modes, and value streams.
Team Topologies provides four fundamental team types—stream-aligned, enabling, complicated subsystem, and platform—each with specific expectations about behaviour, interactions, and relationship to flow of value. These aren’t just labels. They’re design patterns that reduce cognitive load, clarify ownership, and enable fast flow.
When this infrastructure exists, AI fits naturally. Stream-aligned teams identify where AI reduces toil in their value stream. Platform teams build AI capabilities that others consume as self-service. Enabling teams help others adopt AI practices effectively.
Without this infrastructure, AI deployment becomes chaotic. Which team owns this AI capability? How does it interact with other systems? Who’s accountable when it produces problematic outputs? Where do handoffs occur?
The infrastructure of agency—clear articulation of team types, boundaries, and interactions—enables both human and AI effectiveness. Its absence creates confusion that AI amplifies rather than resolves.
The Cloud Adoption Parallel: We've Seen This Pattern Before
Organizations rushing towards AI should remember what happened with cloud adoption.
The promise was simple: move to the cloud, gain instant speed and cost savings. What actually happened? For many organizations, bills increased and teams slowed down.
Why? Because they’d piled new technology on top of existing dysfunction. The cloud amplified whatever organizational patterns already existed. Teams with clear ownership and well-designed boundaries moved faster and saved money. Teams with unclear boundaries, constant context switching, and blurred accountability found that cloud technology made everything more expensive and complicated.
AI follows the same pattern, only accelerated.
The technology works. The organizational infrastructure to support it—in most organizations—doesn’t exist.
The Question That Changes Everything
Before deploying your next AI initiative, Rich suggests organizations ask a deceptively simple question: “Where does this really help users make effective progress?”
Not “What can AI do?” Not “How do competitors use AI?” Not “What AI capabilities exist?”
Rather: “Given what we understand about our users and their needs, where would AI genuinely add value to their progress?”
This question only makes sense if you can answer several preceding questions:
- Who are our users, really?
- What needs are they trying to meet?
- How do we currently deliver value to meet those needs?
- Where are the bottlenecks and friction points?
- Which teams own which capabilities in that value delivery?
- How do those teams interact and hand off work?
Without answers to these foundational questions, AI deployment becomes a gamble. With answers, AI deployment becomes targeted and strategic.
The difference determines whether AI amplifies effectiveness or dysfunction.
Why This Matters Now: The Acceleration of Consequences
The rate of technological change continues accelerating. AI capabilities expand monthly. Competitive pressure mounts. Organizations feel compelled to act.
But speed without direction creates waste. AI deployed on unclear foundations doesn’t just fail to deliver value—it actively creates problems. Hallucinations amplified by lack of oversight. Biases embedded in processes at scale. Dependencies multiplied across unclear boundaries.
The organizations that will thrive aren’t those rushing fastest towards AI adoption. They’re those taking time to build the organizational infrastructure that enables AI to amplify their strengths rather than their weaknesses.
This means:
Starting with users and needs, not technology and capabilities. What are users trying to accomplish? Where do they experience friction? What progress are they seeking to make?
Clarifying team boundaries and interactions. Which teams own which capabilities? How do they interact? Where are dependencies necessary versus artefacts of structure?
Building “outside in” thinking. Aligning around user value streams rather than internal functions. Designing teams around outcomes rather than outputs.
Creating infrastructure of agency. Using Team Topologies patterns to articulate how teams work, reducing cognitive load and enabling fast flow.
Deploying AI where it serves user progress. Not everywhere AI could work, but specifically where it helps users make effective progress towards their needs.
This isn’t just better practice. In an environment where AI amplifies whatever you currently have, it’s survival.
The Path Forward: Infrastructure Before Amplification
AI adoption doesn’t start with selecting models or running pilots. It starts with organizational design.
Do your teams understand who their users are and what they need? Can they articulate how they deliver value? Are team boundaries designed around user value streams or internal functions? Do interaction modes reduce or multiply dependencies?
If the answers aren’t clear, AI won’t solve that problem. It will magnify it.
But if you can answer those questions—if you’ve built the infrastructure of agency through clear team design and “outside in” thinking—AI becomes a powerful enabler rather than an expensive distraction.
The choice isn’t whether to adopt AI. The choice is whether to build organizational foundations that let AI amplify your effectiveness rather than your dysfunction.
Your competitors are deploying AI. But deployment alone doesn’t create advantage. Thoughtful deployment on solid organizational foundations does.
The question is: which will you build first?
Listen to the Full Conversation
Rich Allen and Mark Blackwell explore AI amplification, organizational design, and the infrastructure of agency in depth, including practical guidance on Team Topologies, User Needs Mapping, and building “outside in” thinking.
Watch on YouTube: https://youtu.be/L9qFz-WmdGc
Listen on Buzzsprout: https://www.buzzsprout.com/2012667/episodes/16152903
About the Speakers
Rich Allen is author of User Needs Mapping: Aligning Teams Around What Matters and a leading practitioner of Team Topologies. With two decades in software engineering and organizational transformation, Rich helps organizations move from “inside out” to “outside in” thinking, designing team structures that enable fast flow of value.
LinkedIn: https://www.linkedin.com/in/richallen/
Website: https://userneedsmapping.com/
Book: https://www.amazon.com/User-Needs-Mapping-Aligning-Matters/dp/B0FVBBKPKP
Mark Blackwell is founder of Arkaro, specialising in change management, strategy, innovation, and commercial excellence for the agriculture, food, and chemicals industries. Arkaro’s collaborative “do it with you” approach works closely with clients through a four-step process: Understand, Co-create, Enable, Sustain. We don’t just coach—we get on the pitch with you.
Ready to Build AI on Solid Foundations?
Your organization may be rushing towards AI adoption. But without clear organizational infrastructure—teams aligned around user needs, boundaries designed for fast flow, understanding of how value gets delivered—AI will amplify dysfunction rather than reduce it.
At Arkaro, we work with organizations in agriculture, food, and chemicals to build the foundations AI requires: clarity on users and needs, team structures that enable autonomy, processes designed for adaptation rather than optimization.
We build organizational capability to deploy AI thoughtfully.
Our collaborative “do it with you” approach means we don’t just provide frameworks—we work alongside your teams through Understand, Co-create, Enable, Sustain to leave behind sustainable capabilities, not just slide decks.
If you’re seeing the pattern Rich describes—AI initiatives that generate activity without delivering value, unclear boundaries that multiply rather than reduce complexity, teams working hard but not aligned around what users need—let’s discuss whether the Arkaro Approach could help.
Arrange a call with Mark Blackwell
Connect on LinkedIn: Mark Blackwell
Email: mark@arkaro.com
We work with organizations that recognize AI success depends on human factors, not just technical capabilities. We get on the pitch with you.
Related Reading from Arkaro
On AI Implementation & Organizational Design:
AI Implementation Failure: Why 95% Fail & How to Fix It – Why AI projects fail on human factors using the SPACES model and neuroscience research
AI & the Octopus Organization: Adaptive Transformation – Why distributed intelligence and human alignment matter more than technology deployment
People-Centric AI Change Leadership: The PEOPLE Framework – Practical approach putting people at the centre of AI transformation
On Team Alignment & Structure:
User Needs Mapping: Align Teams Around What Matters – Complete exploration of Rich Allen’s framework connecting strategy to structure
Why Traditional Management Fails and How Adaptive Organizations Succeed – Moving from control to flow through trust, value streams, and team boundaries
Neuroscience of Collaboration: The SPACES Model Explained – How clear boundaries, autonomy, and certainty enable team performance
On Strategy Implementation:
Strategy Implementation: Bridge the Corporate Gap – Translating high-level direction into practical decision rules
Strategy is an Adaptive Challenge not a Technical Problem – Why expert-driven approaches fail and collaborative solutions succeed
Make Strategy Work: The Power of Integrated Business Planning – Moving strategy from slides to sustainable execution
On Customer Understanding & Value Delivery:
B2B Win-Loss Analysis for Cross-Functional Growth – Using Jobs-to-Be-Done methodology to understand customer needs beyond features
Exposing Bottlenecks: The Key to Building a Great Strategy – Root cause analysis for identifying what truly constrains growth
On Change Management:
Transforming Change Resistance: 4-Step Collaborative Approach – Why resistance reveals valuable insights and how to transform it
Why ADKAR and the Arkaro Approach Are Essential Partners – Combining structured change management with collaborative methods
External Resources & Further Reading
User Needs Mapping Website – Complete resource including case studies and community
User Needs Mapping Book – Rich Allen’s comprehensive guide
Team Topologies – Organising for fast flow of value
MIT AI Research – Research on AI implementation success and failure rates
Stay connected: Subscribe to Arkaro Insights newsletter for regular perspectives on AI adoption, change management, and building adaptive organisations in complex industries.
