Executive Summary

Most AI transformations fail—not because the technology doesn’t work, but because the approach is wrong.

Instead of betting on big, centralised projects that collapse under their own weight, forward-thinking organisations are lighting “1000 little fires.”

This new model empowers every employee to use AI in their own workflow—reducing risk, delivering faster results, and building a culture of everyday innovation.

In this article, you’ll learn why this bottom-up approach works, how to make it safe and scalable, and what it takes to turn small sparks into a lasting competitive advantage.

When Big AI Projects Go Wrong

When McDonald's announced in 2021 they were partnering with IBM to revolutionise their drive-thru experience with AI, it seemed like the perfect example of digital transformation done right. A tech giant partnering with a global restaurant leader to solve a genuine business problem. The investment was substantial, the vision was clear, and the technology promised to be cutting-edge.

Fast forward to mid-2024, and McDonald's quietly ended the partnership after a series of high-profile failures. TikTok videos went viral showing the AI repeatedly adding dozens of Chicken McNuggets to orders while customers pleaded with it to stop.

Other customers abandoned their orders entirely after the system added random items like butter and ketchup packets to simple ice cream orders. The ambitious project that was meant to streamline operations and enhance customer experience had instead created frustration and mockery.

The Problem with "One Big Fire"

KEY STATISTIC: Over 80% of AI projects fail to deliver on their promises, a rate nearly double that of traditional IT projects. [^1]

The statistics are sobering: over 80% of AI projects fail to deliver on their promises, a rate nearly double that of traditional IT projects. Behind these failures lies a pattern of systemic issues that plague large-scale AI initiatives.

First is the fundamental disconnect between IT and business teams. AI projects often originate from IT departments fascinated by the technology rather than from business units with a clear problem to solve. This technology-first approach leads to solutions hunting for problems rather than addressing genuine business needs. When business users finally enter the picture, they discover a system that doesn't match their actual workflows or requirements.

Second is the overwhelming focus on technology at the expense of the humans who must use it. Technical teams become enamored with AI capabilities while underestimating the importance of user experience, change management, and training. Without early and sustained end-user involvement, the resulting systems face resistance or outright rejection when deployed.

Third is the data reality gap. Large AI initiatives assume high-quality, consistent data across the organisation—an assumption that rarely holds true. Companies discover too late that their internal data is siloed, inconsistent, or simply not at the quality level required for production AI systems. By the time these issues surface, the project is already significantly over budget and behind schedule.

Fourth is the inevitable scope creep that transforms manageable projects into unwieldy behemoths. As more stakeholders become involved, the list of requirements and integrations grows exponentially. What began as a focused solution expands into an attempt to solve every problem at once, creating enormous technical and organisational complexity.

When organisations bet everything on one transformational AI fire, they face tremendous risk. The McDonald's-IBM drive-thru project demonstrated this perfectly—impressive in controlled tests but failing catastrophically in real-world conditions. These ambitious projects leave companies with wasted investment, damaged credibility, and workforces increasingly resistant to future digital initiatives.

The "1000 Little Fires" Alternative

The alternative to betting everything on one high-stakes AI project is a fundamentally different approach—one that recognises a simple truth: AI is an intensely personal experience. Your needs and use cases for AI are inherently different from your colleagues', even if you work in the same team on similar tasks.

What you type into an AI prompt box, the way you interact with AI tools, and the specific workflows you're looking to enhance are unique to you and the work you do every day.

This is the core insight behind the "1000 little fires" approach to AI transformation.

Instead of focusing exclusively on one or two large, centralised AI initiatives, organisations empower every employee to incorporate AI into their daily workflows in ways that make sense for their specific roles and responsibilities.

Every Employee, Every Workflow

Think about the typical workday of any employee in your organisation. They generally engage in three types of workflow activities:

1. Receiving work - This includes reading emails, attending meetings, reviewing documents, and absorbing information from managers, colleagues, customers, and other departments.
2. Doing work - The core tasks that make up their job: writing, analysing, designing, developing, researching, decision-making, or problem-solving.
3. Passing on work - Communicating results through reporting, presenting, coaching, training, or delegating to others.

Each of these touchpoints represents an opportunity for AI enhancement.

Multiply these opportunities by the number of employees in your organisation, and you see the potential scale of transformation: not one big fire, but thousands of smaller flames, each tailored to individual needs and all contributing to a broader transformation.

The Benefits of Bottom-Up Transformation

Research shows this distributed approach yields several advantages over traditional top-down AI projects:

Higher adoption rates: When employees discover and implement AI solutions that address their specific pain points, adoption happens naturally. A 2024 McKinsey study found that millennials in management positions, who report the highest levels of AI expertise, serve as natural champions for transformation when empowered to use AI in their own workflows.[^2]

Faster time-to-value: Small-scale implementations show immediate benefits without the lengthy development cycles of enterprise-wide projects. According to Microsoft's recent findings, 75% of global knowledge workers are already using generative AI, with many bringing their own AI tools to work to handle their increasing workloads.[^3]

Lower risk profile: Instead of betting everything on a single AI initiative, organisations spread their risk across multiple smaller implementations. If one approach doesn't work, it can be adjusted or abandoned without jeopardising the entire AI strategy.

Continuous innovation: When employees throughout the organisation are empowered to experiment with AI, they surface innovative use cases that central IT teams never discover. Organisations like Telstra have reported that 90% of their employees have adopted AI summarisation tools, leading to measurable improvements in customer interactions.[^4]

Greater resilience: A distributed approach creates a culture where adaptation to technology becomes part of everyday work, building organisational muscles for ongoing change rather than episodic, disruptive transformation efforts.

The Mindset Shift

This approach requires a shift in how leaders think about AI transformation. Rather than seeing AI as primarily a technology to be implemented by IT, it becomes a capability to be developed by everyone - from the intern to the CEO.

Gallup's recent AI Readiness and Adoption research found that successful AI adoption requires "shared ownership across the business and employees who feel prepared and comfortable with experimentation."[^5]

The goal isn't to replace the expertise of data scientists or IT professionals—it's to complement their specialised skills with the domain expertise and practical insights of employees throughout the organisation. Technical teams provide the architecture, guardrails, and support, while employees bring the practical understanding of where and how AI can improve their specific workflows.

In essence, the "1000 little fires" approach recognises that meaningful AI transformation isn't about deploying technology—it's about changing how work gets done. And no one understands that better than the people doing the work every day.

Implementation Framework

Implementing a "1000 little fires" approach requires a different playbook than traditional top-down IT projects.

Instead of focusing primarily on technology deployment, leaders need to create the conditions for widespread, employee-led adoption while maintaining appropriate governance and support.

Here's a framework for making this happen in your organisation:

1. Create a Permissive Environment with Clear Guardrails

Start by establishing clear guidelines that encourage experimentation while addressing legitimate security, privacy, and ethical concerns. This means:

• Developing an AI usage policy that communicates what's allowed and what's not
• Setting up secure access to approved AI tools that meet security and compliance requirements
• Providing guidance on data handling, prompt engineering, and output verification
• Establishing ethical principles for AI use that align with your organisation's values

The goal is to make it safe and easy for employees to incorporate AI into their workflows without creating unacceptable risks. Clear guardrails enable confident exploration rather than prohibiting it.

2. Equip and Empower Your People

For AI adoption to spread organically, employees need three things:

Access to the right tools: Ensure everyone has access to appropriate AI tools for their role. This includes generative AI assistants, specialised AI applications for specific functions, and integration of AI capabilities into existing software.

Skills and capability building: Develop a tiered training approach that includes:

• Basic AI literacy for all employees
• Starter prompts including AI Use Case Discovery and Prompt Writer
• Pre-built department and role-specific prompts and GPTs to “click and go”
• Role-specific training for applying AI to particular functions
• Advanced skills for champions and power users

Time and space for experimentation: Explicitly give people permission to spend time exploring AI applications. This includes AI learning hours, hackathons, or specific goals around process improvement through AI.

3. Build a Network of AI Ambassadors and Communities

Identify and support enthusiastic early adopters who will:

• Share successes and best practices with their teams
• Help troubleshoot issues when colleagues encounter challenges
• Provide feedback to central AI teams about needs and opportunities
• Test new capabilities before wider rollout

Create communities where employees share use cases, prompt templates, and workflows. These communities cross functional boundaries to encourage the spread of ideas throughout the organisation.

4. Start With High-Impact, Low-Complexity Use Cases

While the long-term goal is widespread adoption, starting with targeted use cases builds momentum. Look for workflows that:

• Affect many employees (to maximise visibility and impact)
• Involve repetitive, time-consuming tasks that AI can help automate
• Don't require complex integrations or customisations to implement
• Deliver clear, measurable benefits

Common starting points include internal policy navigation (like HR policies and compliance requirements), standard operating procedures that everyone needs to follow, company-wide templates for common documents, frequently asked questions about administrative processes, and department-specific knowledge bases that other teams need to reference regularly.

5. Implement Feedback Loops and Continuous Improvement

To sustain momentum and ensure AI adoption delivers real value:

• Collect regular feedback on AI usage and impact
• Identify and address barriers to adoption
• Recognise and celebrate successful AI implementations
• Continuously refine guidance and support based on real-world experience

This ongoing learning process enables the organisation to adapt as AI capabilities evolve and employee needs change.

6. Connect Bottom-Up Innovation With Top-Down Direction

While this approach emphasises employee-led adoption, executive leadership remains crucial. Leaders will:

• Clearly communicate how AI adoption supports strategic priorities
• Use AI tools visibly in their own work to model desired behaviors
• Remove organisational barriers that slow adoption
• Allocate resources to support the AI transformation journey
• Connect individual improvements to broader organisational impact

The most successful organisations balance grassroots innovation with strategic direction, creating what consulting firm Nagarro calls a "sideways integration" where top-down vision meets bottom-up ingenuity. [^6]

7. Measure What Matters

Establish metrics that track both adoption and impact:

• Usage metrics like active users, frequency of use, and feature utilisation
• Productivity metrics like time saved and work output
• Quality metrics like error reduction and consistency improvements
• Employee experience metrics like satisfaction and reduced burnout
• Business outcome metrics that connect AI use to strategic objectives

By focusing on these metrics rather than just technical implementation milestones, you keep the initiative centered on real business value rather than technology deployment for its own sake.

Demonstrating Success

Traditional IT projects measure success through project-oriented metrics: on-time delivery, within-budget implementation, and technical performance.

A distributed, employee-led AI transformation requires a different measurement framework—one that captures both the breadth of adoption and the depth of impact across the organisation.

Adoption Metrics: Lighting and Spreading the Fires

First, you need to track how widely AI is being adopted throughout your organisation:

Participation rate: What percentage of employees are actively using AI tools? Look for trends across departments, roles, and seniority levels to identify both pockets of success and areas needing additional support.

Usage frequency: How often are employees incorporating AI into their workflows? Regular, daily use indicates AI is becoming part of normal work patterns rather than an occasional novelty.

Use case diversity: Are employees applying AI across different types of tasks and workflows? A healthy transformation shows AI being used in multiple ways, not just for a single application.

Tool expansion: Are employees progressing from basic to more advanced AI applications as their skills and confidence grow? This progression shows increasing comfort with the technology.

Community engagement: How active are your AI communities of practice? Metrics like questions asked, solutions shared, and attendance at AI learning events can indicate growing cultural acceptance.

Impact Metrics: Measuring the Heat

Beyond adoption, you need to understand the tangible benefits AI is delivering:

Time savings: How much time are employees saving by using AI for routine tasks? Recent studies have found that employees using generative AI save an average of one hour per day on administrative tasks, with some reporting up to two hours of daily time savings.[^9]

Work output: Are employees able to produce more or higher-quality work with AI assistance? This includes metrics like number of deliverables produced, response times to inquiries, or throughput on key processes.

Error reduction: Is AI helping to reduce errors or improve consistency? Look for decreases in rework, corrections, or quality issues in AI-assisted workflows.

Employee satisfaction: Do employees report less burnout and more satisfaction when using AI? Regular pulse surveys can track changes in how people experience their work.

Business outcomes: Are these individual improvements translating to meaningful business outcomes? Depending on your organisation, this includes customer satisfaction scores, sales conversion rates, innovation metrics, or financial performance indicators. Metrics like questions asked, solutions shared, and attendance at AI learning events can indicate growing cultural acceptance.

Leading Indicators of Success

Several early signals indicate whether your "1000 little fires" approach is taking hold:

Organic spread of use cases: Are employees spontaneously discovering and sharing new ways to use AI? This indicates the approach is becoming self-sustaining.

Decreasing support needs: As employees become more skilled and confident, they require less hand-holding from central teams.

Rising sophistication: Look for evolution from simple use cases (like drafting emails) to more complex applications (like customer-facing GPTs or decision-making support).

Cross-functional collaboration: Are teams that previously worked in silos now sharing AI approaches and insights? This indicates AI is becoming a bridge across organisational boundaries.

Employee-led innovation: The strongest indicator of success is when employees create their own AI-powered solutions tailored to their specific needs.

From Metrics to Learning

The most value from your measurement framework comes not just from tracking numbers, but from using those insights to continuously refine your approach:

1. Identify both successful patterns (to amplify) and barriers (to address)
2. Share success stories and learnings across the organisation
3. Adjust support, training, and resources based on adoption patterns
4. Connect individual improvements to broader strategic goals

Remember that unlike a single "big fire" project with a clear endpoint, the "1000 little fires" approach is an ongoing journey of continuous improvement.

Your measurement framework will evolve as your organisation's AI maturity grows, focusing first on adoption, then on impact, and ultimately on transformative changes to how work gets done.

Conclusion: From Transformation to Competitive Advantage

The traditional "one big fire" approach to AI transformation creates unnecessary risk and often fails to deliver on its promise. By focusing exclusively on large, centralised initiatives, organisations risk missing out on the distributed intelligence of their workforce.

The "1000 little fires" approach offers a more resilient alternative. By empowering every employee to incorporate AI into their daily workflows, organisations create a grassroots movement that drives sustainable transformation. This distributed approach reduces risk, accelerates time-to-value, and creates a culture of continuous innovation.

This strategy presents a particular opportunity for smaller, more nimble organisations. While large enterprises often get bogged down in complex, years-long AI initiatives, smaller companies can rapidly deploy targeted AI enhancements across their workforce. This creates an unexpected competitive advantage – the ability to outmaneuver larger rivals through widespread, practical AI adoption rather than betting everything on a single transformative project.

The organisations that thrive in the AI era won't be those with the biggest budgets or most advanced technical implementations. They'll be the ones that effectively combine technology with human ingenuity, that balance top-down strategy with bottom-up innovation, and that recognise AI transformation is about changing how work gets done, not just deploying new tools.

The time to start lighting those fires is now.

Author’s note: While the "1000 little fires" metaphor helps illustrate our approach to AI transformation, I recognise that for readers in regions recently affected by actual wildfires, this analogy may carry different weight. The intention is to highlight how small, controlled initiatives can collectively create positive organisational change.

[^1]: RAND Corporation Research Report, "Why AI Projects Fail and How They Can Succeed," 2023.

[^2]: McKinsey Digital, "AI in the workplace: A report for 2025," January 2025.

[^3]: Microsoft, "AI at Work Is Here. Now Comes the Hard Part," 2025.

[^4]: Microsoft Official Blog, "How real-world businesses are transforming with AI — with 261 new stories," April 2025.

[^5]: Gallup, "A People-First Approach to AI Adoption," October 2024.

[^6]: Nagarro, "AI Transformation - Top Down, Bottom Up, or Sideways"

How can Kowalah help?

CIOs and IT leaders trust Kowalah's AI-powered platform to navigate complex AI procurement and deployment decisions with confidence, turning the fear of making costly mistakes into strategic advantage.

Chat with Kowalah to think through your AI strategy, develop your business case and pick the right vendors.

Create best practice documents, processes and policies to put your AI strategy on track.

Sign up for free at kowalah.com/sign-up