Succesful AI Implementations Start Specific

Most executives I speak to think AI implementation requires technical expertise, complex coding, or specialized teams. Some of them are even hesitant to learn AI for exactly this reason. 

What I found out, however, is that non-technical leaders are sometimes getting the most out of AI. They are succeeding precisely because they're not technical. 

I’ll share my thoughts as to why:

Getting the most out of AI is treating it like a coworker: understanding what work needs to be done and then delegating it to AI. As discussed last week, this is why the elite among AI leaders treat AI like a thinking partner.

Implementing AI as a Non-Technical Leader

I see the same thing in deeper AI implementations. Last week, we kicked off a 3-week trajectory focused on implementing AI for Lead with AI graduates. 

While technologists often start with the tools and capabilities of AI, successful business leaders start somewhere entirely different – with their deepest understanding of where their time delivers real value.

Why Most AI Projects Fail

The traditional approach to AI implementation typically follows a familiar pattern: Hire consultants, plan massive transformations, and attempt to automate entire workflows at once. We see charts like the one below (an ‘automated social media content’ workflow) and think we need to master this to benefit from AI.

But this approach misses a crucial truth: the most valuable AI implementations aren't all-encompassing but laser-focused on solving specific, well-defined problems. 

By targeting a narrow domain and optimizing for clear objectives, our AI can get really good at something specific. This is way better than those big, complicated AI projects that try to do too much and end up not getting anywhere because they're too confusing and don't have a clear goal.

The Power of Being Specific 

Take an early-stage venture capital professional from our recent cohort. Instead of trying to automate her entire investment process, she focused on one specific pain point: Investment memos that were taking her team four days to write. 

By building a simple AI system that combines pitch decks with interview transcripts, she cut that time to 90 minutes. The key wasn't technical sophistication – it was her deep understanding of what makes a great investment memo and the workflows to get there.

Other examples from our community were equally clever:

• Sodexo’s Head of Future of Work Henrik Jarleskog transformed his property investment by feeding historical price data into an AI system that helped negotiate in real time – and land him the house of his dreams! 
• Marcus Bowen, co-founder of Work&Place, revolutionized his media company's editorial process, cutting production time from weeks to hours while maintaining quality.
• Work futurist Sophie Wade built an AI agent to review video course scripts, elevate key learning points, and make various supporting course materials.
• Virtual Work Insider founder Sacha Connor used a custom GPT to turn drafts of marketing posts into ones that reflect her voice.
• Follow FlexOS creator and proptech guru Antony Slumbers (check his course on AI for real estate!) religiously uses NotebookLM as a study guide, like understanding Deepseek’s technical papers and asking for key developments.  

What unites these success stories isn't technical sophistication – it's clarity about which specific tasks to transform, and then purpose-build your own AI toward it.

From Insight to Implementation

‍Through working with dozens of leaders, I've identified four crucial questions that build on my G.E.D. model that predict AI implementation success:

• Is the task general enough that a smart generalist could do it?
• Is it digital work?
• Does it happen frequently enough to justify the setup time?
• Is it work you'd happily delegate if you could clone yourself?

The highest-scoring tasks on these criteria consistently deliver the most value when automated, regardless of technical complexity.

‍Building Your Minimum Lovable Product

‍It’s important not to jump to the most technologically complex implementations. We may be dazzled by the content we see about crazy AI automation, but the truth is that we need to start with ‘Minimum Lovable Product,’ just like startup developers would. 

The most successful leaders will follow a clear progression:

1. Start with simple prompts to test their approach
2. Create a GPT or Project with specific knowledge
3. Add basic automation only once value is proven
4. Scale to more complex solutions based on results (and eventually even build agents)

The Bottom Line

‍As AI tools become more sophisticated, the advantage will increasingly go to leaders who understand their businesses deeply rather than those who understand the technology perfectly. The future belongs to those who can identify the right problems to solve, not necessarily those who can dream up the most technologically-forward AI solutions.

Focusing on specific, high-value tasks that you’d love to delegate, rather than comprehensive automation. This is how non-technical leaders will achieve remarkable results with surprisingly simple tools.

‍What specific task in your workflow could you transform this week? 

‍The tools are ready – you just need to choose the right target.