'''

The Unspoken Rules of AI Consulting

Is it 1999 all over again? I see the same wild-eyed optimism, the same valuation madness, the same fear of missing out. As an angel investor in over 200 companies, including some of the foundational players like Anthropic, OpenAI, and Scale AI, I have a front-row seat to the AI gold rush. And let me tell you, it feels eerily familiar.

Everyone is throwing money at anything with ".ai" in the name. The hype is deafening. But I’ve learned a thing or two from my past exits and my portfolio. Yes, it's a bubble. But not everything is a bubble. The trick is to know the difference.

As an investor, I'm not chasing the shiny objects. I'm placing my bets on the "picks and shovels" of this gold rush. The unsexy, behind-the-scenes infrastructure that makes the entire AI revolution possible. While everyone else is looking for gold, I’m selling the shovels.

The "Picks and Shovels" Play

When everyone is rushing to find gold, the surest way to make money is to sell picks and shovels. In the context of AI, this means investing in the fundamental infrastructure, tools, and platforms that everyone else needs to build their own AI applications. I’m talking about companies like Scale AI, which provides high-quality training data for AI models, or Hugging Face, which has become the GitHub for machine learning.

These companies aren't building the flashiest consumer-facing AI products. They are building the essential infrastructure that the entire ecosystem relies on. They are the ones providing the picks and shovels to the gold miners. And they are the ones who will be standing long after the gold rush is over.

Corporate AI Adoption is Harder Than It Looks

I’ve seen it a dozen times. A big corporation decides they need an "AI strategy". They hire a team of expensive consultants, buy a bunch of software, and a year later, they have nothing to show for it but a massive bill. Why? Because they focus on the technology, not the business problem.

They get so caught up in the hype of large language models and generative AI that they forget to ask the most important question: "What problem are we actually trying to solve?" Adopting AI isn't about plugging in some magic black box. It's about deeply understanding your business processes and figuring out where AI can provide real, measurable value. It’s about having a clear strategy and a way to measure ROI.

I remember one of my portfolio companies, a B2B SaaS startup, was struggling to get traction. They had built a sophisticated AI-powered analytics tool, but nobody was buying it. I sat down with the founder and asked him, "Who is your customer, and what is their biggest pain point?" He couldn’t give me a clear answer. They had fallen in love with their technology but had forgotten about the customer. We spent the next three months talking to potential customers, and we discovered that their biggest pain point wasn’t analytics, it was data entry. So, we pivoted the product to focus on automating data entry using AI. The company took off.

The Unspoken Rules of AI Consulting

Which brings me to the core of this article. If you're an AI consultant or an entrepreneur in the AI space, here are some unspoken rules to live by. These are the things that the glossy brochures and the keynote speakers won

't tell you.

1. Sell Outcomes, Not Technology

Nobody cares about your fancy algorithm. Seriously. They don’t care about transformers, or recurrent neural networks, or your proprietary model that’s 0.5% more accurate than the state-of-the-art. What they care about is what it can do for their business. Can you reduce customer churn by 10%? Can you automate 50% of their manual data entry? Can you increase their sales conversion rate by 5%? Those are the questions that matter.

I learned this the hard way with my first startup, MovieLaLa. We had built this amazing recommendation engine, a true work of art from a technical perspective. We were so proud of it. We would go into meetings with movie studios and talk all about the intricacies of our collaborative filtering algorithm. They would nod politely, and then they would ask, "So, how does this help us sell more movie tickets?" We didn't have a good answer. We were selling technology, not outcomes. It wasn't until we pivoted to focus on a concrete business problem – helping studios target their marketing campaigns more effectively – that we started to get traction. We were eventually acquired by Gfycat, but that early lesson stuck with me.

2. Your First Pilot Will Probably Fail

And that's okay. In fact, it's expected. The goal of a first AI pilot is not to achieve a massive ROI. The goal is to learn. You are testing a hypothesis. You are gathering data. You are figuring out what works and what doesn't. If you go into it with the expectation of a home run, you're setting yourself up for disappointment.

I always advise my portfolio companies to frame their first AI projects as experiments. Be transparent with the client. Tell them, "We believe we can solve this problem with AI, but we need to run a small-scale pilot to validate our approach. We'll define clear success metrics, and at the end of the pilot, we'll have a clear go/no-go decision on whether to proceed with a full-scale implementation." This manages expectations and creates a collaborative, learning-oriented relationship.

One of the biggest mistakes I see is companies spending a year and millions of dollars on a massive, all-encompassing AI project, only to see it fail spectacularly. Start small. Be agile. Iterate. The goal is to fail fast and cheap, learn from your mistakes, and then double down on what works.

3. Data is Everything

This is probably the most important rule of all. Your AI model is only as good as the data you feed it. Garbage in, garbage out. It's a cliché, but it's true. You can have the most sophisticated algorithm in the world, but if your data is a mess, your results will be a mess.

Before you even think about building a model, you need to have a deep understanding of your data. Where does it come from? How is it collected? Is it clean? Is it labeled? Is there bias in the data? I’ve seen projects get derailed for months because the team didn't do their data homework upfront. They would build a model, get terrible results, and then have to go back to square one to clean up their data.

This is why companies like Scale AI are so valuable. They provide the clean, high-quality, labeled data that is the lifeblood of any AI application. If you're a consultant, a huge part of your job is going to be data janitor. It's not glamorous, but it's absolutely essential. Don't underestimate the amount of time and effort that will be required to get the data into a usable state. It's often 80% of the work.

4. The Last Mile is the Hardest

So you've built a great model. It's accurate, it's robust, and it's solving a real business problem. You're done, right? Wrong. The hardest part is often the last mile: integrating your model into the existing business processes and workflows. This is where so many AI projects fall apart.

It's not enough to just have a model that spits out predictions. You need to figure out how those predictions are going to be used by the people on the front lines. How do you present the information in a way that is intuitive and actionable? How do you get them to trust the model's recommendations? How do you handle cases where the model is wrong?

This is not a technical problem. It's a human problem. It requires a deep understanding of user experience design, change management, and organizational politics. You need to work closely with the end-users to understand their needs and their pain points. You need to be a translator between the world of data science and the world of business. This is a rare and valuable skill. If you can master it, you will be unstoppable.

5. Don't Boil the Ocean

I see this all the time with ambitious founders and consultants. They want to build a single, monolithic AI platform that does everything for everyone. It's a recipe for disaster. The most successful AI companies I've seen all started with a narrow focus. They picked a specific industry, a specific problem, and they went deep. They became the best in the world at solving that one problem.

Once you've established a beachhead, you can start to expand. But if you try to do everything at once, you'll end up doing nothing well. The AI landscape is incredibly competitive. The only way to win is to be the best at something. What is your niche? What is your unique insight? What is the one problem that you can solve better than anyone else?

Find your niche, dominate it, and then expand from there. It's the same strategy I used with RemoteTeam, which was acquired by Gusto. We didn't try to build a full-stack HR platform. We focused on one thing: making it easy for companies to hire and pay remote workers. We became the best in the world at that one thing, and that's what made us successful.

The Real Opportunity

The AI revolution is not about building sentient robots or artificial general intelligence. Not yet, anyway. The real opportunity, right here, right now, is in the boring, unsexy, and incredibly valuable work of applying AI to solve real-world business problems. It's about automating tedious tasks, providing better insights from data, and creating more personalized customer experiences.

It's not as glamorous as the stuff you see in the movies. But it's where the real money is being made. And it's where you can have a real impact. So, if you're an entrepreneur or a consultant in the AI space, my advice is simple: ignore the hype, focus on the fundamentals, and go solve some real problems. The picks and shovels are where the fortunes will be made in this gold rush. Don't be the one left holding a bag of worthless gold.

Frequently Asked Questions

What experience informs this perspective?

This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.

Do all experts agree with this view?

No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.

How has this view evolved over time?

My thinking on most topics has changed significantly over the years. Early in my career, I held many conventional views that experience proved wrong. I try to update my beliefs when the evidence changes.

What's the most common pushback you get on this?

People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.