How to Start Building Apps with GenAI

Without the proper use case, your AI plans will crash and burn.

It’s tempting to think big, but don’t fall into that trap. When you start too big, you risk failing big and losing credibility and confidence from your decision-makers.

Pick a small project that provides a basis for success. When you have a quick win, you can shout it to the rafters and take your lessons learned to the next step.

Goals like using AI and generative AI to build software or create reports are excellent. Things like AI-generated customer experiences? Keep it on your roadmap, but way down the path.

Think about the data in your organization. Where is it coming from, and what is the quality? How much if it is siloed? How consistent is the formatting? How much can you reliably get to? How much are you using, and for what?

Data drives everything. “Good inputs lead to good outputs” has never applied so perfectly. To build an application with AI and genAI, you need:

1. 1. 1. Large datasets
      2. Consistent, accurate data
      3. Data you can get to

Because AI is so data-reliant, risk and compliance are critical. How are you keeping your sensitive customer data private and secure? Is that data part of your AI use cases? Are you sticking to legislation like the GDPR and EU AI Act?

Establishing data hygiene practices like securing your data and consistent formatting are good starting points, as is looking for ways to combine your data.

According to a recent AWS survey, 93% of respondents agreed that a good data strategy is vital to driving value from AI.

There is a skills gap in IT. Developers and data scientists must learn how to use, build, and implement AI.

But it also applies to the business side. Those are your stakeholders. It doesn’t help anyone if you’re asked to build a rocket ship when you only have the skills to build a model. Here are a few focus points.

Gartner experts believe that AI adoption may lead to productivity boosts of nearly 25%.

Finding common ground is difficult if your business is way over here and IT is way over there. It can feel overwhelming, but here are the three keys to bringing everyone together:

1. 1. 1. Prioritization: How do we take everyone’s needs and develop the best roadmap?
      2. Stakeholder management: How does IT engage stakeholders early and often to gather insights and ensure buy-in?
      3. Clear roles and responsibilities: What do the teams look like? What are the roles? Who’s building what? Are you utilizing business technologists?

In the middle of all this is where you want to be. Your roadmap needs to reflect all your stakeholders, and your teams need to reflect your roadmap. That shared sense of purpose sparks more discussion, more thoughts, and, ultimately, more innovation.

According to Gartner, business technologists make up 28-55% of the workforce, depending on the industry. Of those, almost a third are involved in an application’s entire capability life cycle.

AI software development is a never-ending cycle, so set yourself up for the long haul. You’ll constantly move from ideas to experimentation to outcomes, then take your learnings and start the whole process again.

To be clear: When it comes to AI, you will fail. The evolution is too fast, and the world is shifting too much to get it perfect on the first try. From there, you have two options:

1. 1. 1. Grumble that the tech’s no good, it’s not worth it, you don’t have the expertise, etc.
      2. Fail fast, take learnings, and go again.

Option 1 is where good ideas go to die. Option 2 is how you learn, innovate, grow, and win.

Getting your AI lifecycle right is part technology, part mindset, and part process. Keep it straightforward, especially in your early days with brainstorms, fusion teams, and continuous collaboration. Put yourself in a position to succeed early so you don’t have to worry about building the basics later.

And when you succeed, even in the most minor ways, celebrate! AI solutions can be incredible, and accomplishments feel like big deals because they are. Even failures can bring new ideas that bear fruit down the road.

Some estimates of AI project failure are as high as 80%. So when things go wrong, don’t stress. You’re in good company.

To build an application with AI and genAI, the most important thing is to start experimenting. You could waste months, quarters, and years setting yourself up, but by the time you’re ready, the market’s evolved, and you need to go again.

However, these steps provide a robust enough starting point to protect your organization, regardless of the future.

The spark of inspiration is just the beginning. Now you’ve got some momentum. Take those ideas, experiment with them, and turn them into your desired outcomes.