AI is a bicycle for customer discovery
When Steve Jobs called the computer ‘a bicycle for the mind’, he was talking about augmentation, not replacement.
Here’s how you can use AI to get closer to your customers rather than go round in circles.
How AI might actually get you closer to your customers
It sounds like a paradox: Use something distinctly non-human to create a more human product.
When we launched our new way of working with early-stage startups in September, our goal was to share the frameworks and structured sprints that can bring a founder from spark to scale fast. By that, we mean take them from having a promising idea into a scalable, intuitive, science-backed product.
Initially, AI came into this as a gentle footnote. It was a question of how we could use AI early in the soundboarding phase and balance its benefits against any issues further down the line.
Our conclusion was that if used strategically, AI might be the greatest thing a founder could do to hit the road and test their ideas early, rather than sit in a dark room for six months and create a product with little to no time talking to their end customer.
But, just like every time you get on a new bike, we’ll start with a safety warning…
Cycle safe: a note on what AI is and isn’t
A really good engineer can use AI extremely well, especially when they’re building a straw man that can be iterated on.
Yet, while AI is lowering the barrier to entry and making it possible for everyone to get started – no matter their background or coding ability – you’ll likely see very quick progress up to a point. Then, you’ll almost always hit a roadblock if you don’t understand the nuts and bolts of your code or how your finished product was built.
The end goal for any engineer is to create something that’s clean, easily maintainable, extensible, and scalable. That’s the art of software engineering.
But when you outsource that skill to AI, you might have to spend a lot of time backpedalling to understand why and how you got to your current product to make even a relatively small change like changing the colour of a button – let alone tackle more fundamental issues like:
How to scale to 10x the number of users
Reduce cloud computing costs
Raise the bar on security to satisfy a more demanding class of customer
Whatever you build, it’s important that you are fully responsible for all of it.
In our humble opinion, it’s always worth bringing on experts who can fill that delta between the product an AI can create and the production-grade software you’d be comfortable putting in front of a customer, particularly when your reputation is on the line.
Here’s how we’ve been using AI at Cirevo in the early stages of product development.
Generating ideas and getting unstuck
Soundboarding with a product like Claude, Copilot, Gemini, or ChatGPT can be a fantastic way of generating lots of ideas in one go. If you’re getting stuck, this can also be a super useful way to think through other approaches, ideas, and potential frameworks.
If you’re at the first rung of building and just figuring out what your product might look like, experimenting with detailed prompts can give you a sense of what you like and don’t like.
Two key things to consider at this stage:
Make sure you know when the discovery questions stop and the customer research begins, so you don’t end up in a rabbit hole of endless desk research.
Keep bringing in your own critical thinking, drawing on your knowledge and industry understanding to make your inputs (and therefore, outputs) as useful as possible.
2. Creating a scrappy prototype
For some concepts, particularly those in complex or niche areas like regenerative agriculture or soil health, you can use AI to create a quick prototype that can show customers, future talent, and investors that:
a) Yes, you really do know what you’re talking about
b) You have a concrete vision and are willing to show them something tangible
It’s far easier to say yes to something that you can grasp and clearly see how it might fit into your life or those of its customers.
When you use AI to do this, you have to accept that what you’re building is quick and scrappy. It’s there primarily to get early feedback, strengthen your thinking, and validate your problem statement. While it can be built on in some cases with heavy prototyping, you have to be ready to start again.
In this messy stage of AI prototyping, the gold is found by maximising on learning and not becoming too wedded to a specific solution.
When you’ve spent months or even years bringing a product into fruition, there’s a risk that you think “This has to work – or else”.
Now, from a product perspective, you can vibe code a concept in an afternoon, show it to customers, and if there’s energy – you can see how it might solve a problem or bring a benefit to their lives – you can confidently invest the time and resources into building it properly.
3. Strengthening (not outsourcing) your thinking
We love to use AI on smaller blocks of thinking rather than broader strategies.
This way, you’re not outsourcing your own critical thinking skills or taste and discernment. What’s more, you can be very transparent in how you’ve used AI and why. If you’ve applied AI to a small piece of the puzzle, you can easily mark up where it’s from (e.g. Claude) and check that you’ve tested it against real-world evidence rather than rely on assumptions.
4. Pair programming
Just like pair programming or pair coding, an agile software development technique where one engineer writes the code (the driver) and the other reviews, thinks ahead, and provides guidance (the navigator), you can use AI to perform a similar role.
When used responsibly and with focused prompts, the AI can help you figure out the syntax for what you’re trying to do, support with writing well bounded functions, and soundboard architectural thinking. The ultimate goal is to reach cleaner code, faster knowledge sharing, fewer bugs, and stronger team alignment
“Look Mum no hands”
Just like a bicycle isn’t sentient, AI also does not care.
Given how impressive AI outputs can be, you have to be disciplined not to hand over too much control and ownership.
AI’s external polish can be seductive and you might even start doubting your own ability to do a good job. But, that’s a tough place to go psychologically.
“The moment you start outsourcing your own ability and understanding, you’ve lost mastery over the software you’re building and ultimately, mastery over your product and its fate.”
Humans are amazingly creative and can join the dots between so many different things – optimising and innovating in ways that are genuinely new.
Outsourcing this future thinking to a machine at best, feels like a shame, and at worst, is a hindrance to inspiration and overall progression: both blocking growth at the personal and professional level as well as progression of software development as a whole.
This veneer of progression and progress brings its own set of existential risks of hiding actual, real-world substance, much like a shiny sustainability report that packages up good intentions with very little to no real-world impact.
Know where AI ends and your customer discovery begins
With all things considered, how can you be the captain of your ship and – as we totally overstretch this bike metaphor – the pedaller of a fantastic product for people and the planet?
Our conclusion?
Ultimately, AI is a bicycle that can bring you closer to your customers through early stage soundboarding, stronger thinking, and the creation of a quick, scrappy, and accessible prototype. It’s something that can help you not get too fixated on a specific idea, brings something quickly to validation, and enables you to choose the most promising route based on customer feedback.
We think that if used in this way, AI might successfully shorten the time between ideation and prototype to just weeks rather than months or years and create genuinely impactful solutions – the kind that can shape our lives and those of future generations.
The trick is knowing where AI ends and your customer begins, without getting too caught up in its all-too-polished spokes.
If you have an idea for a product in climate, nature and circular economy and need some help moving it from idea to workable product, you can schedule a chat with our co-founders here: Pete Redshaw and jon@cirevo.com.
More about Cirevo
Cirevo designs and builds digital tools for a regenerative future.
We’re a multidisciplinary team of software engineers, data scientists, and environmental experts who thrive in complexity. From idea to delivery, we partner with climate innovators, circular economy ventures, and mission-driven teams to turn big, systems-led challenges into practical, scalable technology.
Our work is grounded in science, shaped by real-world context, and built to last.
If you have a sustainable and regenerative technology project that you’d like help with, please get in touch today.
