What Comes After AI Efficiency Gains
Stopping at efficiency leaves value on the table. The real opportunity is knowing what to build next, and having the judgement to direct AI rather than just deploy it.
The first series in AI for Humans was about getting started: figuring out where you stood, getting your data into shape, running a first pilot, scaling what worked. If you haven’t done that work yet, it is still worth your time.
Once you have, a different question tends to surface. Not “how do we use AI?”, but “what should we be doing with it that we are not?” That question is what this series is here to address.
The ceiling on efficiency
Using AI to accelerate what you already do, like generating proposals or speeding up research, makes sense as a starting point. There are real gains in it. But there is a ceiling.
When a genuinely new capability arrives, the instinct is to bolt it onto the existing workflows and measure the time saved. The problem is that every other business in the sector is doing the same thing with the same tools, and when everyone gets faster at the same work the advantage disappears into price.
The businesses that pull ahead are the ones using AI to do different things, not the same things cheaper.
What has actually changed
For a long time, business worked on a fairly simple assumption: talent is scarce, execution is expensive, so you delegate carefully and protect your best people’s time.
AI has quietly upended that. The ability to execute well-defined tasks (generating content, summarising research, analysing data) is now abundant and cheap. You can get in minutes what used to take a team days.
Which raises a question that not enough people are sitting with: if a tool can produce the output, where does the value actually live?
The scarce resource is no longer execution. It’s knowing what to ask for in the first place.
The bottleneck has moved. The constraint used to be how much could be produced. Now it is what should be produced in the first place, and whether anyone has the judgement to know when AI should lead, when it should support, and when it should stay out of the way.
Automation versus application
Automation is the same work done faster. A process becomes cheaper or more consistent. It is measurable, it is sensible, and it is the default starting point for good reason.
Application is something else: work that was not possible before. Services that would not have turned a margin, analysis nobody had the weeks to do, ways of operating that needed a team you did not have. Harder to pin down on a slide, but where lasting advantage actually lives.
The first series in AI for Humans sat mostly on the automation side: readiness, data, pilots, scaling. That work still matters. This second series is about the application side. It is about moving past “how do we speed this up?” and onto “what could we be doing that we are not?”
Sustainability check
Efficiency gains don’t always mean less impact. Sometimes they just mean more output, more compute, more energy. As you move from automating to applying, it’s worth asking: does this new thing we’re building reduce our footprint, or quietly grow it? The businesses thinking clearly about AI and sustainability aren’t treating them as separate conversations.
The case for judgement
When machines can handle everything that fits neatly into a brief, value starts to concentrate around the things that don’t.
That is what judgement looks like in practice: knowing when an output is technically correct but does not fit the situation, sensing when the brief itself is the problem, spotting the question a customer has not yet thought to ask.
It matters more now than it did a year ago, and not because the tools are bad. It matters more because they are good. The better AI gets at producing competent output, the wider the gap becomes between the businesses directing it well and the businesses running it for the sake of running it.
Across sectors, the work is moving in the same direction: from making things directly to designing the systems that make them. The prompts, the workflows, the scaffolding around the AI. That work rewards the businesses that can design those systems thoughtfully, and that can tell the difference between doing the work faster and doing better work.
What this series will cover
Spotting New Opportunities: Identifying the work AI has newly made viable, and the questions that surface it.
When to Delegate, When to Lead: Building judgement about where AI should run, where it should support, and where it has no business being.
Designing the Judgement Layer: Teams, workflows and systems that develop better judgement over time, and turn that into value your competitors cannot easily replicate.
Who this is for
This series is for people who have started using AI and are already wondering what the next move looks like. The reader I have in mind is a leader who can feel that “a bit faster than last year” is not going to be enough to compete on, and is starting to ask harder questions about where the business should actually be going next.
If you’re still getting started, our first series is there for you and it’s a good foundation to build on. This one picks up where that leaves off.
If your team is past the efficiency stage and wondering what comes next with AI, we work with businesses at exactly this point through our Digital Product and AI service.