The temptation is real.
You’ve seen the headlines. You’ve heard the buzz. And now your exec team wants a custom AI solution that transforms the business, reduces headcount, and delivers triple-digit ROI—yesterday.
But here’s the truth most consultants won’t tell you:
Most custom AI projects fail.
The smartest AI strategy might already be sitting inside the software you use every day.
Let me explain.
🤖 The 3 AI Paths – And Why They Matter
There are three distinct ways organizations are adopting AI in 2025:
1. Enterprise AI (Custom AI)
Building proprietary solutions in-house with data scientists and engineers. High potential – and high risk.
🛠️ Think: training your own models, developing internal tools from scratch, or orchestrating large-scale AI infrastructure.
2. In-App AI (Embedded AI)
Leveraging AI features built into the platforms you already use – like Microsoft 365 Copilot, Salesforce Einstein, or Autodesk’s AI tools.
⭐️ Think: AI-enabled suggestions, automations, document generation, analytics, or summarization in existing tools.
3. Everyday AI
Using consumer-facing tools like ChatGPT, Claude, Perplexity, and others to enhance individual productivity.
🧠 Think: knowledge work augmentation, ideation, writing, research, and basic automation.
So, where should you start?
With the second two. Every. Time.
💥 The ROI Reality: Built-In Beats Custom (At First)
Here’s why.
🧾 Custom AI projects have a terrible success rate.
- A 2023 IBM report found enterprise-wide AI initiatives delivered only 5.9% ROI – while consuming nearly 10% of capital budgets.
- 70–85% of AI projects either underperform or never get out of pilot.
🚀 Embedded and Everyday AI are delivering results – fast.
- Companies using generative AI report an average ROI of $3.70 for every $1 spent, often through simple content, code, or workflow automation.
- Microsoft and Salesforce users are seeing value in weeks or months, not years.
- No-code “citizen developers” are achieving automation ROI up to 300% – often without writing a single line of code.
And perhaps most importantly:
🧠 You don’t need to hire a data scientist to get started.
You need curiosity, experimentation – and a willingness to click the damn button.
🎯 The Barry Honey Principle: Just Click the Damn Button
Barry Honey is the CIO at pitt&sherry, an engineering consultancy based in Melbourne.
He’s not waiting for a custom AI strategy to fall from the sky. He’s not spinning up a lab of PhDs.
Instead, he’s telling his team: Just click the damn button.
“When you see those diamonds, stars, or AI icons in your software – click it.
Try it. Use it. See what it can do.”
It’s a simple mantra. But it reflects a much bigger shift in mindset:
✅ Don’t build from scratch when the value is already embedded.
✅ Don’t let fear of imperfection block experimentation.
✅ Don’t wait for permission to start using the tools you already have access to.
That’s how innovation spreads.
⚠️ When Is Enterprise AI Worth It?
To be clear: custom AI still has a role.
But it’s not step one. It’s step five – after you’ve done the foundational work.
Here’s the smarter sequence:
1. Map your highest-impact use cases
Focus on the business problems that – if solved – would create measurable value.
2. Prove value through embedded or everyday AI
Use what’s already available in your tech stack or on your desktop. Let the wins speak for themselves.
3. Build internal fluency and trust
Train your people. Encourage experimentation. Show them AI is an enhancement – not a threat.
4. Identify capability gaps off-the-shelf tools can’t fill
If there’s a strategic need that no tool covers – that’s your potential build.
5. Make sure your data is clean and complete
Because no custom model can outrun bad inputs. Enterprise AI lives and dies on data hygiene.
Unless you’re solving a truly unique problem at scale, there’s a good chance someone else has already built the thing you’re about to spend thousands to millions recreating.
🔁 Start Small. Move Fast. Learn Loudly.
“Start where the value is already waiting for you.”
That’s the mindset that sets future-fit leaders apart.
The companies winning with AI in 2025 aren’t necessarily the biggest, richest, or flashiest. They’re the ones who embed AI into everyday work, empower their people to experiment, and scale what works.
Start with the tools you already have.
Encourage your team to play.
Track what delivers value – and build from there.
💬 Over to You
Which of the three AI paths is your company currently exploring?
Are you embracing the tools you already have – or chasing something bigger before you’re ready?
Let’s compare notes 👇
Are you interested in learning how I can help you implement AI the right way? Contact me about:
- Team sessions and/or one-to-one coaching to build AI Fluency
- Advisory services to help you understand the when, what, why and how of AI Implementation
