Building an AI-Ready Operating Model: Why AI, Agile, and Lean Must Work Together

The Wrong Question Everyone's Asking

When working with organizational transformation leads, we've been asked countless times: "Should we be investing in Agile or AI? What about Lean?" Organizations are looking to prioritize their transformation efforts to better prepare themselves for today's challenges while positioning for future growth.

While we understand why the question is being asked, we think it's the wrong question, and here's why.

If you're looking for organizational effectiveness, Agile, AI, and Lean come together to enable organizations to create value in ways they've never been able to before. You can't just add AI to a process that's broken and expect better results. You have to take a step back and think critically about what to do. The real question is: How do we create an operating model that's fit for purpose, digitally enabled, and efficient?

Your competitors aren't choosing between these approaches. They're already using them together. And the gap between organizations that get this integration right and those that don't is widening every single day.

What We Mean by "AI" (And Why It Matters Now)

Before we dive deeper, let's get clear on what we're talking about when we say "AI." We're not talking about some distant future scenario. We're focusing on the AI-solutions that exist today; Large Language Models (LLMs) powering tools for summarization, categorization, and generating text, images, audio, and video. We're talking about the tools your teams are already experimenting with:

• Developers using Cursor, GitHub Copilot, or Claude Code to write code at least 2x faster
• Support teams resolving tickets in minutes instead of hours with AI-powered chatbots
• Marketing teams turning one piece of content into multiple formats across channels
• Finance teams processing routine transactions without human intervention
• HR teams screening resumes and scheduling interviews automatically

We see software companies where developers were already using AI coding tools, but each team was using different tools with no organizational learning or best practices. The result? Inconsistent quality, security concerns, and missed opportunities for systematic improvement.

On the other hand we see companies investing millions in fancy new IT tools, some as broad as copilot to more narrowly-focused tools like Granola, but they're being used in isolation. The result? When talking with executives we are hearing the same refrain. We have made the investment in tooling but we are not seeing the behavior or system change required to get the benefits at the scale required to support the investment.

What comes from a lack of strategy? Money spent without a clear return on the investment, without a clear adoption and change program, an operating model that does not support the intended objectives, a loss of confidence in AI, and organizational stagnation. The question isn't whether AI is coming to your organization. It's whether you're harnessing it strategically or letting it happen haphazardly.

Quick assessment: Ask five people in your organization what AI tools they're using. If you're surprised by the answers, you're already behind on creating an intentional AI strategy.

The Integration Framework: How Agile, Lean, and AI Create Competitive Advantage

Let's explore how organizational agility, Agile development practices, and Lean process engineering create the foundation for AI success, and how AI can accelerate all three in return.

Organizational Agility: Building the Human Network for AI Success

For organizations to move fast and leverage the speed AI offers, they need to be aligned, dynamic, and empowered to make decisions at the edges instead of waiting for central approval.

The Challenge: If you have an old hierarchical, bureaucratic, and misaligned organization, AI will have little impact. You'll just have faster access to insights that still get stuck in approval chains, or you'll go faster and further in the wrong direction. You need a strategy to get velocity.

The Solution: An Agile Operating Model built on three core tenets:

• End-to-end customer-driven prioritization
• Organizational adoption of continuous improvement
• Systems designed for incremental delivery

What This Looks Like in Practice:

Speed and Responsiveness: Agile systems focus on delivering quality and impact quickly. They get solutions into users' hands to gather feedback and create safe tests for real-world impact.

Real breakthroughs happen when you change how fast you can respond to what you're learning. Instead of weekly quality meetings where you talk about last week's problems, set up daily feedback loops that let you adjust processes in real-time based on what your AI is telling you.

The key insight? Weekly meetings mean you're always operating on week-old information. You need an organizational structure that can move as fast as your AI tools can generate insights.

Alignment and Prioritization: Every organization struggles with alignment. If you're not clear end-to-end on what's important, AI will just add more noise to an already chaotic system.

Organizations need to maintain what we call a "golden thread," where each employee can trace how their work ties directly to organizational vision. You need to be as clear about what you're not doing as what you are doing.

Continuous Improvement Mindset: The biggest barrier to AI adoption isn't technical competency. Generative AI has a low barrier to entry. The biggest barrier is helping people see how to deploy it differently and think differently about their work.

Here's what we consistently see when organizations roll out AI tools: the teams that succeed aren't necessarily the most technically savvy. They're the ones already comfortable with experimentation and changing their processes based on what they learn.

Technical skills can be taught. Willingness to experiment and adapt? That's a mindset that needs to be built into how your organization works.

Edge Empowerment: To create speed and continuous improvement, empower the teams closest to the work and loosen control from the center. Here's why: if every AI decision needs central approval, you'll get change implemented by people who know the least about the area where they're deploying the technology. If every change comes from the center will it have the right input from the people who are supposed to use it day to day? And if it doesn't, will anyone actually adopt it?

The teams doing the work know what works and what doesn't. Trust them to experiment and iterate.

Multi-Disciplinary Delivery: Here's what we've learned from the most successful AI implementations: don't separate your software engineers from your data engineers from your domain experts. Instead, bring that diverse knowledge and skills together in integrated teams that can move from problem identification to solution deployment without navigating a maze of dependencies. When you silo these roles, you're creating handoffs and increasing the potential for bottlenecks.

The magic happens when these disciplines collaborate from the start. They catch problems earlier, iterate faster, and build solutions that actually work together (just like they do).

Try this approach: Identify your next AI opportunity. Instead of assigning it to your "AI team," create a small multi-disciplinary team that includes domain experts, technical capabilities, and customer insight. Give them two weeks to deliver something testable.

Agile Development: The Scientific Method for AI Implementation

The foundational AI models and platforms weren't developed using waterfall planning, so you shouldn't expect to leverage these tools effectively with waterfall approaches either.

Scientific Method Approach: Data scientists and researchers who created these tools used hypothesis-driven development. They tested ideas, gathered insights, formed new hypotheses, and tested again. When they hit dead ends, they changed course or abandoned directions entirely. This is Agile development applied to AI.

What This Means for Your Organization:

Iterative Development: As teams look to improve their delivery by leveraging AI tools, they'll need to embrace the same experimental approach. The path to creating new ways of working won't be straight. Teams will try things that don't work and need to pivot or abandon approaches that aren't delivering value.

Here's what we see time and again: marketing teams start experimenting with AI for content generation. Their first few experiments usually fail to deliver meaningful results. But here's the key: if you're working in one or two-week iterations with clear success criteria, you can kill those approaches quickly and redirect effort to what actually works.

Most breakthrough AI applications we see aren't the first thing teams tried. They're what teams discovered after they stopped banging their heads against approaches that weren't working.

It doesn't matter if you're working with generative AI, traditional machine learning, or any AI approach. The magic isn't in getting it right the first time, it's in failing fast enough that you still have energy and budget to find what does work.

Release-Based Discipline: We've seen this pattern repeatedly. Teams sit on AI solutions that could already create customer value, but they're too nervous to release them because they aren't "academically perfect" yet. One of us experienced this firsthand while working in an AI lab, where the discipline of regular sprints and releases forced teams to pull up and assess progress. Huge amounts of time were saved when the decision was made to just kill the work stream and move capacity to something of higher value when it was determined that the value was just not there. The concept of incremental delivery creates discipline and forces teams to work incrementally rather than trying to "boil the ocean." Regular release cycles serve three critical functions:

• Clear Achievement Criteria: Teams work toward specific, achievable goals where success can be measured objectively
• User Value Assessment: Regular check-ins with end users determine whether you've already achieved their goals or need to try different approaches
• Portfolio Management: When experiments aren't delivering value or technical feasibility becomes questionable, you can redirect capacity to higher-value opportunities

The opportunities for AI are limitless, but your organizational capacity isn't. There's no need to continue bouncing against dead ends when you could be creating customer value with approaches that work.

Lean Process Engineering: Designing Workflows for Human-AI Collaboration

Once Agile operating models and development practices are established, Lean principles play a vital role in enabling work to flow efficiently across the organization.

The Critical Insight: AI introduces opportunities for speed and automation, but without thoughtful system design, injecting AI into current workflows may simply magnify existing inefficiencies or overwhelm legacy processes with volumes and speeds they weren't designed to handle. Remember the approval process that took days? The backup of approvals grows exponentially when the things needing approval can be generated in minutes.

Value Stream Focus Before Automation: Before deploying AI, map how work, information, and decisions currently flow from customer request to value delivery. This reveals:

• Delays caused by excessive handoffs or approvals
• Steps that don't add meaningful value
• Gaps in coordination, communication, or data access

For example: organizations want to use AI for scheduling. When they map their current process, they discover that the scheduling bottleneck isn't capacity calculation. It's getting preferences and verification information. Fix the information flow first, then add AI to optimize scheduling based on complete data.

Don't automate broken processes. If your current system is slow because people can't find the information they need to make decisions, AI won't fix that. It'll just make bad decisions faster. Map the process, find the real bottleneck, fix that first. Then AI can actually help instead of just speeding up dysfunction (and increasing frustration).

Designing for Even Flow: Lean systems prioritize consistent, sustainable rhythm. When AI enters the picture, that rhythm can become highly variable unless processes are adapted accordingly.

Organizations need to design for:

• Smooth handoffs between AI systems and human decision-makers
• Reduced wait times and batching
• Clear role definition: what AI handles automatically and where human oversight remains essential
• Feedback mechanisms that allow continuous improvement of AI performance

Human-AI Workflow Design: Historically, processes were designed around how humans work: meetings, approvals, email chains, and task lists. AI changes that equation fundamentally.

To benefit from AI's potential, organizations must design systems that assume hybrid human-AI collaboration:

Decision Automation Readiness: Where AI can be trusted to make or recommend decisions, workflows need to accommodate faster execution without manual gating that creates bottlenecks.

Real-Time Data Availability: AI cannot operate effectively on inaccessible or fragmented information. Systems must surface structured, relevant data as part of normal workflow.

Continuous Feedback Integration: AI systems improve through feedback loops. Processes need mechanisms for capturing and applying insights from both successful and unsuccessful AI recommendations.

Load Management with Human Oversight: AI may generate steady streams of tasks that legacy systems built for batch processing cannot handle effectively. Design workflows where AI handles routine tasks while humans focus on complex judgment calls and quality assurance.

Think about how transportation companies handled route optimization. Most did daily planning sessions where they figured out routes for the day. AI and real-time information changed that. Instead of static daily plans, they create continuous optimization that adjusts routes based on real-time conditions: traffic, weather, delivery updates. UPS has been doing this for years.

The trick isn't just adding AI to your existing planning process. It's redesigning how your drivers get information and how dispatch communicates changes. You're moving from "here's your route for today" to "here's your next stop based on what's happening right now."

The biggest barrier to AI adoption isn't technical. It's trust.

Start with Human Review: Begin AI implementations with human review of all outputs. As confidence builds and patterns emerge, gradually automate routine decisions while maintaining human oversight for exceptions.

Clear Escalation Paths: Design systems where AI knows when to ask for help. Define clear criteria for when decisions should be escalated to humans: high dollar amounts, unusual patterns, customer complaints, or safety concerns.

Transparency and Explainability: People need to understand why AI made specific recommendations. Build systems that can explain their reasoning in terms humans can evaluate and challenge.

Easy Override Capability: Humans should always be able to override AI decisions quickly and easily. This isn't a sign of system failure. It's a feature that builds confidence and improves the system over time.

Feedback Integration: Create simple ways for humans to provide feedback on AI decisions. This improves the AI and gives people a sense of agency in the process.

Start with high-frequency, low-risk decisions where humans can easily verify AI performance. Build trust and expertise before moving to higher-stakes applications.

Creating the Virtuous Cycle: How AI Accelerates Agile and Lean

The relationship between Agile, Lean, and AI isn't just one-directional. AI can make your Agile and Lean practices more effective, creating a virtuous cycle of improvement.

AI-Enhanced Testing and Feedback: Both Agile and Lean emphasize rapid testing and continuous iteration. AI can accelerate these feedback loops significantly.

Examples we've seen in practice:

• Digital twins that allow engineers to test process changes in simulated environments before implementation
• AI-generated synthetic customer personas that provide product feedback during development
• Automated A/B testing that runs multiple experiments simultaneously and surfaces insights faster than manual analysis

AI-Accelerated Workflows: Many Agile and Lean activities involve repetitive analysis and documentation. AI can handle the routine work while humans focus on insight and decision-making.

Practical applications:

• AI-generated user stories based on customer research data
• Automated value stream mapping that identifies bottlenecks from workflow data
• Dynamic roadmap updates based on changing priorities and capacity

Systematic Improvement: Instead of just applying AI randomly, use your continuous improvement mindset to enhance your own Agile and Lean practices. Ask: How can we remove manual processes that slow down our improvement cycles?

Quick experiment: Run a retrospective with your teams. As you uncover what's working well, and what isn't, identify the most time-consuming manual aspect and test whether AI can handle it, freeing your team to focus on higher-value analysis and decision-making. Do this each and every Sprint.

Implementation Roadmap: Getting Started

Based on our work with clients across industries, here's a practical approach to building your AI-ready operating model:

Phase 1: Assessment and Foundation

• Audit current AI usage across your organization
• Map your three most critical value streams
• Identify decision-making bottlenecks that slow response to change
• Select your first cross-functional AI experiment team

Phase 2: Pilot Implementation

• Run three small AI experiments using Agile development practices
• Apply Lean analysis to one core process before adding AI
• Establish feedback loops and measurement criteria
• Document lessons learned and best practices

Phase 3: Systematic Expansion

• Scale successful approaches to additional teams
• Build organizational capabilities for continuous AI experimentation
• Integrate AI considerations into your standard Agile and Lean practices
• Establish governance that enables edge empowerment while maintaining alignment

Remember, everything's an experiment. Give these approaches a try. Figure out what works well for your organization, what doesn't, and how to improve. The organizations that develop this capability systematically will create competitive advantages that are difficult for others to replicate.