Eight lessons from building an AI product

This article is part of our Opinions section.

When ChatGPT exploded into the public domain, something magical happened. For me, that’s what every interaction with new technology should feel like, and there’s a reason that everyone – from the top CEOs to marketing powerhouses to your average Joe on social media – couldn’t get enough.

Having access to artificial intelligence in this way, something that Sundar Pichai, CEO of Google described in 2018 as being “more profound than electricity or fire”, was unlike anything we had experienced before. ChatGPT was the innovation of the year in 2023 for good reason, however these capabilities have been around for a long time. Think back to GitHub launching its AI copilot back in 2021. 

Enter the AI copilot for product managers

I’ve been interacting with AI since 2017, when it was only just crossing the chasm from early adopters, and always wondered when there would be an AI copilot for product managers. I thought about this for some time – AI and LLMs were so rare that this felt demonstrably distant – but in 2021, I built and launched one myself.

I learnt a lot in that time, and continue to do so every day. The AI space is enticing and ever-changing, and just like Sundar Pichai, I consider it to be the most profound thing I’ve worked on. As good as AI is, this space isn’t going to develop without the creativity of people, so I want to share my top lessons learnt in building an AI product.

Understand the why

Before you start building an AI product, define your purpose. The ‘why’ matters, and whether you’ve been inspired by a personal pain point or from hearing someone else’s, there are many areas still to be improved through the power of AI. Generative AI is an instance of technology that is guaranteed to make solutions faster and cheaper, and redundant and mundane tasks are an easier area to start with, as they impact us all.

Your audience cannot be everyone

This space is competitive, and AI is resolving new problems daily. For that reason, pick your niche. There will be buyers of your product, users of your product, and challenges to overcome, but diluting your business model to try and cover every possible persona means you’re actually turning away the customers you have the capacity and knowledge to serve. 

If we’ve learned anything from the widespread tech layoffs and company dissolutions over the last 18 months, it’s that the market is cut-throat, and the capacity to stand out is a necessity for survival. In this landscape, the importance of selecting your niche cannot be overstated. It represents not merely a choice, but a strategic decision that will define the trajectory of your business.

Celebrate being different

Trust me, being different is better. This is particularly crucial in industries led by AI and tech advancements, where every customer seeks solutions that are not just effective but also attuned to their unique challenges. Above all, knowing your niche means you know how to allocate your resources, from market research to product innovation, in ways that resonate deeply with your audience.

Keep it simple, and deliver

Trying to “catch them all” isn’t a sustainable business strategy, and reduces how effective your business is for a particular customer. Don’t commit to more than what you can do, and if you have big ideas but no idea how to get there, work towards them. Deliver what you actually can today, and keep the promises for when you are truly 90% ready to deliver. Don’t fall into the trap of being an “us too!” AI product. If you want to have engaged customers who keep coming back, stick to areas where you can genuinely deliver, and do it well.

Focus on the user experience

People who are using an AI product need to know that clicking will equal magic. Users should see the impact of AI in your product, and feel satisfied that you’ve taken away their challenges and delivered them a solution beyond their expectations.

My entire entrepreneurial life has been built on the importance of customer feedback, and this is just as true for developing an AI product.

While AI is still in its early days, trailblazers in the space are responsible for totally upheaving customer expectations. For that reason, the user experience, and addressing subsequent customer feedback, is vital to the development of any AI product.

Get started with open-source

When developing an AI product, cost is always going to be a big factor. Ideally, use open source, as this not only minimizes the financial burden (which is key for any budding creator) but brings you to the helm of community collaboration and innovation. Doing so means you aren’t compromising quality, giving you the tools to access cutting-edge technology and deliver products that are built on a robust, peer-reviewed foundation.

Safety first!

You’re creating an AI product, and take it from me, it’s an incredible time to be a part of this scene. But there’s one thing that cannot drop off your to-do list – it has to be safe. Data governance is essential, especially if you’re trying to build an AI product destined for longevity. Ensure that NDAs are in place, and security checks are consistent throughout. 

You also have to stay informed. This is an exciting time to be involved with AI because it’s quick-moving and full of the brightest minds in our industry looking to make their mark. However, it’s that particular component of AI – the rapid pace – that means anyone looking to create an AI product must keep themselves abreast of regulatory changes and safety guidelines.

Rome wasn’t built in a day, brace yourself

The incredible thing about developing an AI product is that it allows you to truly unleash innovation in the tech space. Ten years ago, the average person wouldn’t have had a generative AI image creation tool on their wish list, but today, users of DALL-E are everywhere. Essentially, creating something great takes time, and sometimes, you’re lucky enough to capture lightning in a bottle and come up with an idea that your audience doesn’t even know they want yet.

If building a great product or solution was easy, everyone would do it. Nurture your creativity, listen and learn, be an active part of tech communities, get your hands dirty, and above all – brace yourself! There will be challenges along the way, but for each you overcome, you’ll know you’re on the cusp of building something great.