Everyone is obsessed with hiring expensive AI talent. I get it. You see the headlines, the massive salaries, and you think the only way to win is to get a team of ex-Google Brain researchers. I’m here to tell you that’s a trap. After two exits and over 200 angel investments in AI, including companies like Anthropic and OpenAI, I’ve seen firsthand what works and what doesn’t. And the single biggest predictor of success isn’t the prestige of your data scientists’ PhDs.
At my last company, RemoteTeam, before we were acquired by Gusto, we spent a solid six months digging into data from over 500 AI teams. We looked at everything. Team size, budget, the academic background of the engineers, the tools they used, you name it. We were trying to find the magic formula. The one thing that separated the teams that shipped amazing products from the ones that just… spun their wheels. The answer was not what we expected, and it’s something I’ve seen play out again and again in my own companies and the startups I invest in. It’s not about the talent, the budget, or even the data. It’s about a specific role that most companies completely overlook.
The Common Traps I See Every Day
I get pitched by AI startups constantly. And I see the same mistakes over and over.
The Myth of the AI Unicorns
So many founders I talk to are chasing “AI unicorns.” They’re convinced they need to spend a fortune on a handful of star researchers. I saw one startup burn through $10 million in seed money in less than a year. Why? They hired a team of brilliant, but purely academic, data scientists who couldn’t ship a single product. They built incredibly complex models that solved problems no customer had. It was a classic case of a solution looking for a problem. They were so in love with the technical challenge that they forgot to ask if anyone would actually pay for it.
Data Hoarding vs. Smart Data
Then there's data paralysis. Teams that are so proud of their terabytes of data, but have no idea what to do with it. They spend months cleaning and labeling, waiting for the perfect dataset. I call it "analysis paralysis." They’re so focused on achieving 99.9% accuracy in a lab that they miss the market entirely. A scrappier competitor with a fraction of the data, but a clearer problem to solve, will lap them every time. I’d rather have a team that can take a messy, incomplete dataset and find a real customer problem to solve with it, than a team that waits for perfection.
Tool Chasing
And the tools. Oh, the tools. Every week there’s a new vector database, a new framework, a new model that promises to be the magic bullet. I’ve seen teams rebuild their entire stack three times in a year, chasing the latest trend on Hacker News. They’re busy, for sure. But they’re not making progress. They’re confusing activity with achievement. The best teams I’ve seen are tool-agnostic. They pick a stack and they stick with it. They’re focused on solving the problem, not on using the shiniest new toy.
The #1 Factor: The "Translator" Role
So what is the secret? What was the one thing that our analysis of 500 teams pointed to? It’s the Translator.
This is the person who lives in the space between the business and the technology. They don’t have to be the best coder, and they probably don’t have a PhD in machine learning. But they have something far more valuable: deep customer empathy and a product-focused mindset. They can talk to a customer, understand their real-world problem, and then turn around and explain that problem to the engineering team in a way that they can actually solve. They are the bridge between the customer and the code.
I remember at MovieLaLa, my second company, we had an engineer who was a good, but not great, coder. But he was obsessed with movies. He spent his weekends talking to other movie fans, trying to understand how they discovered new films. He was a film nerd first, and an engineer second. He became our de facto Translator. He would sit with the marketing team and then with the AI team, literally translating their needs and constraints to each other. He was the reason our recommendation engine felt so personal, so human. He was the reason we were acquired by Gfycat. He wasn’t the best engineer on paper, but he was the most valuable person on the team.
Another example is from one of my portfolio companies, a startup building an AI-powered sales tool. They were struggling to get traction. Their model was technically brilliant, but sales reps weren’t using it. They hired a former sales manager, someone who had carried a bag for 10 years, to be their Head of Product. He didn’t know how to code, but he knew the sales process inside and out. He spent his first month just talking to users. He came back with a list of a dozen small, simple changes to the product that made it fit into the existing workflow of a sales rep. The changes were easy to implement. And the result? A 300% increase in user engagement in one quarter. That’s the power of a Translator.
How to Cultivate Translators in Your Organization
So how do you find these people? You probably already have them. You just need to know what to look for.
Promote from within: Look at your product managers, your sales engineers, your customer support leads. Who are the ones who are always tinkering, always asking smart questions about the technology? Who are the ones who are obsessed with your customers? Give them a path to grow into this role. Send them to a coding bootcamp. Pair them with a senior engineer. Invest in them. They are your future AI leaders.
Hire for it: When you interview AI talent, don’t just ask them about their Kaggle scores. Ask them to explain a complex technical concept to a non-technical person. Ask them about the last product they used that they loved, and why. You’re looking for a passion for the product, not just the algorithm. I’d rather hire someone with a B+ in computer science and an A+ in customer empathy than the other way around.
Structure your teams for it: Don’t silo your AI team. That’s the kiss of death. Embed them with your product and business teams. Make them go on sales calls. Force them to sit in on customer feedback sessions. The magic happens when the people building the AI are the same people who are talking to the users. At RemoteTeam, we had a rule: every engineer had to spend at least one day a month on the customer support queue. The engineers hated it at first. But then they started to see the patterns. They started to understand the real-world problems our customers were facing. And they started to build better products.
The Future is Translated
Look, the AI hype is deafening. It’s easy to get caught up in the race for bigger models and more impressive benchmarks. But the companies that will win in the long run are the ones that remember that AI is a tool to solve human problems. And to do that, you need more than just brilliant engineers. You need Translators. The companies that win in this new era won't be the ones with the biggest models, but the ones with the best Translators. The choice is yours. Don't just hire coders. Hire people who can connect the code to the customer. That's the secret. That's how you win.
I can't stress this enough. I've seen it in my own companies, and I've seen it in the 200+ startups I've invested in. The pattern is always the same. The teams that succeed are the ones that have strong Translators. The ones that fail are the ones that don't.
Think about it. Who is going to have more impact? A team of five PhDs who spend a year building a model that is 99.9% accurate but solves a problem nobody has, or a team with one good engineer and one great Translator who ship a product in three months that is 80% accurate but solves a real, painful problem for customers? It's not even a contest.
I remember one of my early angel investments. A brilliant team out of Stanford. They were building an AI-powered tool for legal research. The tech was incredible. They could parse legal documents faster and more accurately than any human. But they couldn't get any law firms to buy it. They were about to run out of money. I sat down with the founder and asked him to walk me through a demo. The product was a command-line tool. You had to be a programmer to use it. I told him, 'Lawyers don't use the command line. They use Word and email.' He looked at me like I was crazy. He said, 'But this is so much more powerful!' I told him, 'It doesn't matter how powerful it is if nobody can use it.'
They ended up hiring a paralegal who had taught herself to code. She became their Translator. She redesigned the entire user interface to look and feel like a Word plugin. It was a huge step back, technically. But it was a massive leap forward for their users. They closed their first six-figure deal a month later. Today, they're the market leader. That's the power of a Translator.
So, my advice to you is this: stop chasing the AI unicorns. Stop hoarding data. Stop chasing the latest tools. Instead, look for the Translators. They are the real unicorns. They are the ones who will make the difference between success and failure in the age of AI. Find them, empower them, and get out of their way. You’ll be amazed at what they can do.
Frequently Asked Questions
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.
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.