Last month I sat in a conference room with a VP of Operations who was genuinely confused. They'd deployed AI across nine different use cases. Nine. They had the dashboards to prove it. Time saved: 84 hours per month. Money saved: $9K per month.
"So why," she asked, "does it feel like nothing has actually changed?"
Good question.
Because saving time and saving money - while nice - isn't the same as transformation. She'd hit the ceiling of Horizon 1, and she didn't even know there was a ceiling.
Here's what we've been telling companies about AI: there are three different horizons to break through, and most people don't realise they're stuck looking at the first one.
The Three Horizons (Or: Why Your AI Implementation Feels Like Treading Water)
Horizon 1 is Efficiency. You're doing the same work, just faster or cheaper. This is where everyone starts because it's easy to measure and easy to sell to your CFO.
Horizon 2 is Amplification. You're giving people superpowers they didn't have before. The work itself changes because humans can now do things that were previously impossible.
Horizon 3 is Transformation. You're not doing the old work at all. You've invented entirely new workflows, new business models, new ways of creating value that didn't exist before AI.
Most companies get stuck on Horizon 1. They optimise themselves into a corner, then wonder why things just pretty much feel the same - no huge productivity breakthroughs. No obvious competitive advantage.
So we've been working with clients on how to set their sights on the next horizon in front of them.
Horizon 1: The Efficiency Trap
You're in Horizon 1 if your AI pitch deck uses phrases like "time savings," "cost reduction," or "productivity gains." Not that these are bad. They are the natural starting point. They're just… limited.
Here's what Horizon 1 looks like in practice: You use ChatGPT to write your marketing emails faster. You use an AI transcription tool to summarise meetings. You automate data entry. You speed up invoice processing.
All good. All useful. All completely replaceable by your competitor doing the exact same thing next quarter.
The problem with efficiency is that it has a ceiling. You can only make a process so much faster before you hit the laws of physics. And even if you could make it infinitely fast, you're still just doing the same work everyone else is doing - you're just doing it cheaper.
Which means you've entered a race to the bottom.
How to know you're stuck in Horizon 1: Your AI projects are all about "doing X faster." You measure success in hours saved or FTEs reduced. Your team celebrates when they shave 10% off a process timeline. You're optimising, not reimagining.
How to escape: Stop asking "How can AI make this faster?" Start asking "Why does this process exist at all?"
That's the question that breaks you out.
Take meeting summaries. Everyone's using AI to transcribe and summarise meetings now. Horizon 1 thinking: great, now I can skim the summary instead of attending the meeting (although, arguably, nobody even looks at it again!)
But the better question is: why are we having this meeting?
If the meeting exists to share information, you don't need AI to summarise it - you need to kill the meeting and use async communication instead. If the meeting exists to make a decision, you don't need a summary - you need a decision-making framework that doesn't require eight people in a room for an hour.
So the question is, can AI help with those things? Can it produce and share information automatically through ingesting async comms and inferring where things are at? Can it take inputs, apply a decision-making framework, and give you recommendations ranked against a defined rubric to determine the best decision to make?
Here's your escape hatch: for every AI efficiency project you're running, ask yourself what would happen if you had infinite speed and zero cost. If the answer is "we'd do the same thing, just more of it," you're trapped. If the answer is "we'd stop doing this entirely and do something else," you've found your exit.
Horizon 2: Augmentation (Or: Giving Humans Superpowers)
Horizon 2 is where it starts getting interesting.
You're not replacing the human. You're not speeding up the task. You're fundamentally expanding what the human can do.
I saw this click for a sales team last year. They'd been using AI to auto-generate email templates (Horizon 1). Saved them maybe 20 minutes a day. Fine.
Then they tried something different. They built an AI copilot that could instantly pull up every interaction their company had ever had with a prospect - emails, calls, support tickets, website visits, everything - and synthesise it into a real-time brief while the salesperson was on the phone.
Suddenly a junior rep could sound like a senior rep. They had perfect recall. They knew the prospect's pain points, their buying history, their org chart, their budget cycle. Things that used to require years of experience were now available to everyone.
That's not efficiency. That's augmentation.
The work changed. The junior rep wasn't just doing the same job faster - they were doing a different job entirely. One that required less memorisation and more strategic thinking.
What Horizon 2 looks like in practice: A customer support agent who can now handle complex technical questions because an AI copilot is feeding them the answer in real-time, pulling from documentation they'd never have time to read.
A designer who can iterate on ten concepts in an hour instead of one, because the AI is handling the mechanical execution while the human focuses on creative direction.
A financial analyst who can model scenarios that would've taken weeks in a spreadsheet, testing hundreds of variables in minutes, because the AI is doing the computational heavy lifting.
The pattern: the human is still in the loop, but they're operating at a level that was previously impossible.
How to get to Horizon 2: Stop thinking about tasks. Start thinking about capabilities.
Don't ask: "What tasks can AI do for us?"
Ask: "What could our people do if they had perfect information, infinite memory, and instant analysis?"
That reframe changes everything.
Let's say you're in HR and you're hiring. Horizon 1 thinking: use AI to screen resumes faster. Horizon 2 thinking: give your recruiters an AI that can analyse every hire you've ever made, identify the patterns of who succeeded and who didn't, and flag candidates who match the success profile but might not have the "right" resume.
Now your recruiter isn't just processing applications faster - they're making better decisions. They're seeing things they couldn't see before.
The trap to avoid: Horizon 2 can feel like the finish line. You've given people superpowers. You've expanded capability. You're getting real value. But you're still operating within the old structure. You're still doing the same type of work - just a much better version.
Horizon 3: Transformation (Or: Building the Thing That Doesn't Exist Yet)
Horizon 3 is where you stop optimising the old game and start inventing a new one.
This is hard. Not because the technology is hard (though it can be), but because it requires you to admit that the thing you've been doing - the thing that's "always worked" - might be obsolete.
Let me give you an example that clicked for me.
There's a company that used to sell software for managing email marketing campaigns. You know the type: build your list, design your email, schedule your send, track your opens and clicks.
Horizon 1 thinking: use AI to write better subject lines, optimise send times, personalise the content.
Horizon 2 thinking: give marketers an AI copilot that suggests campaigns based on customer behaviour, auto-generates variants for A/B testing, and predicts which segments will respond best.
Both good. Both valuable.
But Horizon 3? They killed the concept of a "campaign" entirely.
Instead, they built an autonomous agent that monitors every customer's behaviour in real-time - website visits, product usage, support tickets, social media activity - and delivers personalised content the instant it detects an intent signal. No batches. No schedules. No "campaigns."
The AI is always on. Always learning. Always adapting.
That's not a better email tool. That's a different business model.
What Horizon 3 requires: You have to be willing to kill your sacred cows. The processes that "work fine." The workflows that "everyone understands." The org structures that "make sense."
Because Horizon 3 isn't about making the old thing better. It's about asking if the old thing should exist at all.
Here's a test: if you were starting out today and removed all constraints - technology, budget, regulation, culture - what would you build?
Not "what would you improve." What would you build from scratch?
That's your Horizon 3 target.
A real example: I read about a logistics company that was using AI to optimise delivery routes (Horizon 1) and predict demand spikes (Horizon 2). It was very useful.
But when asked the Horizon 3 question - "What would you build from scratch?" - the answer was wild.
They realised that the entire concept of "delivery routes" was a relic of human limitations. Humans need routes because we can't hold 500 addresses in our head and dynamically re-route in real-time based on traffic, weather, and new orders. But an AI can.
So they built a system where there are no routes. Just a fleet of vehicles and a swarm of AI agents that constantly re-allocate based on real-time conditions. A driver might get rerouted mid-delivery because a new high-priority order came in two blocks away. The system is always optimising, always adapting.
They didn't make routes better. They eliminated routes.
That's Horizon 3.
How to Actually Move Through the Horizons (The Practical Bit)
Okay, so you understand the three horizons. Now what?
Here's the thing: you can't skip Horizon 1. You need those early wins to build momentum, to prove value, to get budget for the bigger stuff.
But you can't - or shouldn't - stay there.
Step 1: Audit where you are
Look at every AI project you're running right now. Be honest. Which horizon is it in? If 80% of your projects are Horizon 1, you're in danger. You're optimising yourself into irrelevance.
Step 2: Set a portfolio strategy
You need projects in all three horizons. Not equally - that's not realistic - but you need a mix. A rough guide: 50% Horizon 1 (quick wins, build momentum), 30% Horizon 2 (expand capability), 20% Horizon 3 (place bets on transformation).
The Horizon 1 projects fund the Horizon 3 experiments.
Step 3: Change how you measure success
Horizon 1: measure time saved, cost reduced. Horizon 2: measure capability unlocked (what can people do now that they couldn't before?). Horizon 3: measure new value created (revenue from new products, customers you couldn't serve before, markets you couldn't enter). If you're only measuring Horizon 1 metrics, you'll only get Horizon 1 results.
Step 4: Protect your Horizon 3 projects
Horizon 3 projects will fail more often than Horizon 1 projects. They're riskier. They're weirder. They threaten the status quo. Which means they need protection.
You need a sponsor who's willing to let them run even when the ROI isn't obvious in month two. You need a team that's comfortable with ambiguity. You need to insulate them from the "we've always done it this way" crowd.
Step 5: Kill your darlings
This is the hardest part.
You're going to have processes that you love. That you built. That you're proud of. That "work fine."
And Horizon 3 will require you to kill them.
The expense report process you spent two years perfecting. The approval workflow that "ensures quality." The review cycle that "maintains standards."
If you can't let go of the old thing, you can't build the new thing.
So where are you?
Horizon 1, optimizing the old game?
Horizon 2, amplifying what's possible?
Horizon 3, inventing the new game?
You don't have to jump straight to Horizon 3. But you do have to know which game you're playing.
If you're ready to break out of Horizon 1 or want a hands-on guide for navigating the next horizon, or simply curious to learn more, that's where we come in. We work directly with leaders and teams to find opportunities to get more done. Whether you're looking to amplify capability or invent the new game entirely, we can help you chart the course, build the strategy, and get unstuck for good. Reach out - we're always open for a casual conversation to learn more.