How to Implement AI in Your Organisation: A Practical Guide

Last updated: March 2026

Who this guide is for

*Last updated: February 2026*

If you're leading an organisation through AI adoption and it feels harder than it should be, you're in good company. The organisations we work with have usually tried something already. A pilot project here, a team training session there, maybe a policy draft sitting in someone's inbox. And yet the whole thing hasn't quite clicked into something that feels like real momentum.

The good news is that getting AI adoption right is very doable when you know the sequence. The Five Strategic Building Blocks give you that sequence. They're based on what we've seen work across dozens of organisations, and they're designed to build on each other so that each step reinforces the next. This guide walks you through all five, in order, with the reasoning behind each one.

Contents: what you'll find in this guide

Chapter 1: What separates successful AI implementations from the rest
Chapter 2: The Five Strategic Building Blocks
Chapter 3: Block 1: AI Literacy and Skills Training
Chapter 4: Block 2: Cultural Change
Chapter 5: Block 3: AI Usage Policies
Chapter 6: Block 4: AI Impact Assessments
Chapter 7: Block 5: Your 12-24 Month AI Roadmap
Chapter 8: The common patterns to watch for
Chapter 9: Where to start
Further reading

WHAT SEPARATES SUCCESSFUL AI IMPLEMENTATIONS FROM THE REST

The research paints a striking picture. According to MIT and McKinsey, only about 5% of AI pilots turn into sustained organisational capability. 84% of organisations haven't redesigned a single job around AI (Deloitte, 2026). And 85% are stuck at task-level AI use (BCG).

So what are the 5% doing differently? From everything we've seen, it comes down to sequencing. The organisations that get real traction follow a deliberate order of operations. They build literacy before they write policies. They invest in culture alongside skills. And they wait until they have genuine understanding before trying to build a roadmap.

Here's what the typical journey looks like when the sequencing is off. An organisation decides to "do something about AI." They buy some licences. Maybe they run a workshop. Someone in legal drafts a policy. A few enthusiasts start using the tools. And then... not much changes. The enthusiasts keep going, everyone else drifts back to what they were doing before.

The opportunity is in flipping that sequence. Tool adoption paired with real literacy gives people confidence. Policy built on genuine understanding becomes an enabler rather than a blocker. And training backed by cultural change creates lasting momentum rather than a one-off event that fades within weeks.

There's a piece on the AI dismissal reflex that captures something important here. People try an AI tool once, get a mediocre result, and conclude it's overhyped. They don't realise that using AI well is a skill that develops with practice, just like any other professional competency. The organisations that recognise this and invest in building that skill systematically are the ones pulling ahead.

So how do you actually build that kind of momentum?

THE FIVE STRATEGIC BUILDING BLOCKS

At Ten Past Tomorrow, we use a framework called the Five Strategic Building Blocks. The ordering is deliberate and the sequencing matters. Get the order right and each block reinforces the next. Get it wrong and you spend a lot of energy without much to show for it.

Blocks 1, 2, and 3 can run in parallel over the first three to six months. That's the sweet spot for building real foundations. Blocks 4 and 5 come later, typically around twelve months into the journey, once the organisation has genuine traction and enough experience to make informed decisions about where to go next.

We've refined this framework through working with senior leadership teams across a range of industries in New Zealand, from professional services to government to manufacturing. The sequence holds regardless of industry or organisation size.

BLOCK 1: AI LITERACY AND SKILLS TRAINING

This is where everything begins. And we'd go so far as to say it's the single highest-value investment you can make in your AI journey.

AI tools are fundamentally different from software. They're more like working with a brilliant but literal-minded colleague. You can't just install them and expect people to figure it out. The training has to be intentional, hands-on, and built around real business use cases that people recognise from their actual work. When training feels relevant and practical, adoption happens naturally.

And it needs to happen at every level of the organisation, though what each level needs looks quite different.

Executive leaders need to understand the strategic implications: how AI fits into your business model, your competitive position, your industry dynamics. But they also need the same practical, hands-on training as everyone else. You can't lead what you don't understand, and your teams will take their cues from how you engage with the tools. This is the core argument in our piece on why AI adoption stalls without AI-literate senior leadership teams. When the leadership team has genuine literacy, they make better decisions about investment, risk, and pace.

Mid-level managers need to understand how AI affects team-level efficiency, workflows, and processes. They're the bridge between strategy and execution, and they're often the ones who can spot the highest-value opportunities for AI in day-to-day operations. When managers are equipped and confident, they become multipliers for adoption across their teams.

Frontline employees need direct practical training on how to use AI in their daily tasks. Real workflows, real tools, real use cases from their actual jobs. Abstract theory doesn't stick. But show someone how to cut a two-hour task down to twenty minutes using a tool they can access right now, and you've got their attention.

The goal is bigger than tool proficiency. The goal is an "AI first mindset" where employees identify AI opportunities themselves, in collaboration with their teams and managers, rather than waiting for someone at the top to tell them what to do. That's where the real organisational value starts to compound.

BCG found that employee-centric organisations are seven times more likely to reach AI maturity. The San Antonio Spurs went from 14% to 85% AI fluency by embedding training into daily work rather than running separate workshops (OpenAI white paper). The evidence is consistent: when you make literacy the foundation, everything else becomes easier.

BLOCK 2: CULTURAL CHANGE

This is the broadest building block and, I'd argue, the one that has the biggest long-term impact on whether AI actually sticks in an organisation.

Making AI a reflexive part of how the organisation operates is quite different from giving people access to tools. It's about embedding AI into the fabric of how the company thinks, meets, hires, reviews performance, and makes decisions. And it's where the organisations that are genuinely transforming separate themselves from the ones that are just tinkering.

Cultural change covers six dimensions, and every one of them matters:

Embedding AI into organisational rhythms. AI should show up in one-on-ones, small team meetings, department meetings, and all-company meetings. It should be infused into job descriptions, hiring policies, interview processes, onboarding, KPIs, and performance reviews. When AI is woven into how the organisation already works, it stops being a side project and starts being part of the culture. The organisations doing this well treat AI like any other core business capability: it's expected, supported, and visible everywhere.

AI councils and champions. A practical, hands-on group made up of early adopters and AI-curious people from all levels, including juniors who are often closest to being AI natives. Democratic structure. Fortnightly meetings. Competitions and challenges with real rewards: cash, time off, meaningful incentives. The point is to drive adoption from the bottom up and the inside out, to bring shadow AI into the light so it becomes celebrated rather than hidden. The grassroots AI uprising is already happening in most organisations, whether leadership knows about it or not. 75% of knowledge workers are already using AI tools, often without employer approval. AI councils channel that energy productively and turn individual experimentation into shared organisational learning.

Leadership modelling. Senior leaders need to show strong, consistent, and public use of AI tools. CEO memos about what AI means to the company. Statements about the future of work. Active demonstration that leadership is walking the same path they're asking the organisation to walk. Shopify is a great case study here. When Tobi Lutke made AI usage part of performance reviews, non-engineers became the fastest-growing AI user group. That's the kind of shift that leadership modelling creates. People follow what leaders do, not just what they say.

Psychological and emotional management. People will have strong and varied reactions to AI. Excitement, curiosity, scepticism, anxiety about job security, frustration with the learning curve. All of those reactions are valid, and cultural change means acknowledging them and working with them rather than dismissing or ignoring them. Research has found employees hiding time savings for fear of layoffs. If the culture punishes efficiency gains, no one will admit to them. The organisations getting this right create a clear deal: the organisation gets the productivity gains AND the individual gets rewarded for finding them. That alignment is essential.

Organisational change management. How the organisation manages the transition itself: communication cadence, managing resistance, sustaining momentum, and ensuring the transformation doesn't stall after the initial enthusiasm fades. The best change management we've seen treats AI adoption as a multi-quarter journey with regular check-ins, visible progress markers, and honest conversations about what's working and what's not.

Culture and mindset shift. The shift from "AI as an optional tool" to "AI as a reflexive organisational capability." Every employee is now a potential AI innovator. The electricity analogy is useful here: companies that used electricity just to power the same steam-era machines missed the point entirely. The organisations that redesigned their operations around what electricity actually made possible were the ones that created enormous value. AI is at the same stage right now, and the opportunity for organisations willing to rethink how they work is significant.

BLOCK 3: AI USAGE POLICIES

This is where many organisations instinctively start, and we understand why. Governance feels like the responsible first step. But policies written before anyone has hands-on experience with the tools tend to be generic, restrictive, and driven more by anxiety than understanding.

The best AI policies we've seen come from organisations that invested in literacy first. When the people writing the policy actually understand the tools, the result is a document that enables rather than blocks. It answers the questions employees actually ask: How can I integrate AI into my daily workflows? What tools are approved and encouraged? What level of human oversight is required? What ethical considerations should I be aware of?

And those policies feel quite different from the ones written in isolation by a legal or IT team. They're practical, specific, and grounded in real use cases rather than hypothetical risks.

Key principles:

Policies must be developed in collaboration with AI-literate employees who can bring ground-level reality.
They should balance compliance and innovation, protecting the organisation while giving people room to experiment.
They must be iterative and evolve as the technology and people evolve. A static policy in a fast-moving field becomes outdated quickly.
Safe, but permissive. Enabling, not blocking.

The NZ Corrections case is a useful cautionary tale. An explicit AI policy existed on paper but was violated in practice because the training was insufficient. Policy without literacy is policy without teeth. And the cost of that gap can be significant.

Right now, 74% of companies plan agentic AI within two years but only 21% have governance ready (Deloitte). That's a real gap, but the solution isn't to rush out a restrictive policy. It's to build literacy first, then write policies collaboratively with people who understand the tools from experience. The organisations doing this well end up with policies their teams actually follow because those teams helped shape them.

BLOCK 4: AI IMPACT ASSESSMENTS

Once the organisation has genuine traction with Blocks 1-3, it's time to step back and assess the ripple effects of AI across all stakeholders. This typically makes sense at around twelve months into the journey, when you have enough real-world data to make the assessment meaningful rather than speculative.

What it examines:

Which teams, tasks, processes, and stakeholders are benefiting most from AI?
Which face disruption? Are there unintended consequences?
Ethical concerns: job satisfaction, meaningful work, bias, misinformation.
Financial implications: costs, ROI, new revenue streams.
Risk mitigation: data security, compliance, customer experience.
Could customers start doing what you do for them using AI themselves?

There's a useful framework here: hard ROI versus squishy ROI. Hard ROI means picking one workflow, measuring before and after, and connecting the improvement to revenue. Squishy ROI means the broad experimentation that builds organisational fluency and capability, even when you can't tie it to a specific dollar figure. Both matter. Use the hard wins to justify continued investment in the squishy experimentation. The organisations that only chase hard ROI miss the compounding value of broad capability building. And the ones that only do squishy experimentation struggle to justify the investment to their boards.

A word of caution worth noting: a study by METR found developers were 19% slower with AI tools while believing they were 20% faster. Perceived impact and actual impact can diverge significantly. So rigorous measurement matters, and it's worth building that measurement into your process from early on rather than retrofitting it later.

BLOCK 5: YOUR 12-24 MONTH AI ROADMAP

This only becomes viable once Blocks 1-4 are in place. And that's a feature, not a limitation. You can't build a meaningful roadmap before understanding the technology and how it relates to real workflows in your organisation.

Andrew Ng put it well at Davos: "Letting a thousand flowers bloom has failed." Letting individual teams experiment without strategic direction leads to scattered pilots that never scale. The organisations that create real transformation combine top-down strategic direction with bottom-up learning from the people closest to the work. That combination is powerful.

What the roadmap involves:

Identifying high-value use cases with clear ROI (based on real experience, not hypothetical potential).
Scaling proven use cases across the organisation, turning individual wins into organisational capability.
Allocating proper resources: people, training, technology, infrastructure.
Embedding AI into workflows as a mindset, not an add-on. This is the difference between incremental improvement and genuine transformation.

The critical methodology is workflow redesign. Making one step in a process faster with AI is incremental improvement, and it's a good start. But redesigning the entire workflow around what AI makes possible is where the real value sits. The lesson from McDonald's failed AI drive-through is instructive: they bolted AI onto an existing process without rethinking the workflow. It didn't work. The organisations getting the best results are the ones willing to ask "if we were designing this process from scratch today, knowing what AI can do, what would it look like?"

THE FIVE PATTERNS TO WATCH FOR

We see these patterns repeatedly across organisations at different stages of their AI journey:

Starting with governance before education. Cart before horse. Policies written by people who haven't used AI tools tend to be overly restrictive and disconnected from how the tools actually work. Build literacy first, then write governance with informed people in the room. The result is a much better policy that people will actually follow.

Deploying tools without foundational training. Giving people access to Copilot without teaching them how to use it properly is like giving someone a violin and expecting music. AI tools reward skill and practice, and the organisations investing in that training upfront see dramatically better adoption rates and outcomes.

Expecting instant ROI. AI adoption is a capability-building exercise, not a light switch. The organisations pulling ahead invested in learning before they demanded returns. That patience pays off significantly once the capability is in place, because the gains compound over time.

Letting one department own the conversation. AI touches every part of the organisation. If legal or IT drives it alone, you get a narrow view that often skews toward risk avoidance. Cross-functional ownership brings in the perspectives of the people who'll actually use the tools day to day, and it produces better decisions about where to invest and how to move.

Siloing AI in IT. AI is not a technology project. It's an organisational capability. The generative AI revolution isn't about the tools. It's about what the tools make possible when the whole organisation understands them. The organisations treating AI as a technology initiative are leaving most of the value on the table.

WHERE TO START

The right starting point depends on where you are. Here's how we think about it:

If you haven't started yet: Begin with your senior leadership team. Get shared literacy, a shared language, and a shared framework. That alignment at the top changes everything, because it means the leadership team can make informed decisions together about pace, investment, and priorities. Rapid AI Traction does this in four weeks. Then use that foundation to build out the cultural conditions and the team-level training.

If you've started but stalled: The opportunity is almost always in the cultural layer. You've got pockets of adoption but no organisational momentum. Focus on Building Block 2: leadership modelling, embedding AI into rhythms, and creating the conditions for bottom-up adoption. Often a few targeted moves can restart momentum quickly.

If you've got traction and want to go further: The Path to AI Emergence is a structured ten-month programme that takes organisations through all five building blocks systematically, with live sessions, async content, and always-on access. It's designed for organisations that have done the initial work and are ready to build AI into the fabric of how they operate.

Wherever you start, the principle is the same: literacy first, culture alongside, policies that enable, then assessment and roadmap once you have genuine traction. That sequence works.

ABOUT THE AUTHOR

Mark Laurence

Mark is the founder of Ten Past Tomorrow, an AI consultancy and education business based in New Zealand. A trained futurist (Institute for the Future) and practical AI specialist, he works with senior leadership teams to move organisations from AI curiosity to AI capability.

He has worked with 100+ NZ organisations and leads Rapid AI Traction, a four-week programme for senior leadership teams, and The Path to AI Emergence, a ten-month transformation programme.

HOW TO IMPLEMENT AI IN YOUR ORGANISATION