What if I told you the key to unlocking massive productivity with AI was actually doing less? It sounds crazy, but the results speak for themselves. After years of trial and error, two exits, and investing in over 200 companies, I’ve seen what works and what really, really doesn’t when it comes to building a culture that thrives with AI. I’m sharing the raw, unfiltered story of my struggles and the hard-won victories so you can skip the mistakes and get straight to the results. It wasn’t easy.
When we were building RemoteTeam, we were obsessed with efficiency. Every new tool, every new platform, we had to try it. Especially when the AI wave started to crest. We thought more AI had to equal more output. We were wrong. It just created more noise, more distraction, and more busywork. It took me a while to see the pattern. The real breakthroughs came from a place I didn’t expect.
They came from subtraction, not addition. From focusing on people, not just the tech. And from embracing some seriously counterintuitive ideas.
Lesson 1: The “Shiny Object” Syndrome Will Kill Your Productivity
I remember the week our engineering team’s output dropped by 30%. Thirty percent! We were a startup; that kind of drop is terrifying. I dug into the data, looked at our project management boards, and couldn’t find a reason. No one was on vacation. There were no major bugs. The code base was stable.
Then I started talking to people. It turned out a new AI-powered code completion tool had just been released. It was the hot new thing on Twitter, and half the team was trying to integrate it into their workflow. The other half was debating if it was better than the other AI tool we were already paying for. Instead of writing code, they were benchmarking AI. They were configuring plugins. They were arguing in Slack channels. They were completely distracted by a shiny new object.
That’s when it hit me. The problem wasn’t the tool; it was the culture of chasing. We had created an environment where everyone felt they needed to be on the bleeding edge of every single development. It was my fault.
So we made a change. We implemented a “boring tools” policy. We’d pick one, maybe two, solid, reliable AI tools for a specific job and stick with them for at least six months. No exceptions. No testing the new flavor of the week. The goal was to make the AI so integrated, so boring, that it became invisible. Like electricity or the internet. You don’t think about it, you just use it.
Productivity shot back up. Why? Because the cognitive overhead of choosing and learning new tools vanished. The team could focus on what they were actually hired to do: build a great product. The lesson was clear: stop chasing and start integrating.
Lesson 2: Your Best AI Prompt Engineer Might Be a Poet
When we started to get serious about using large language models for content and marketing at MovieLaLa, my first instinct was to hire a bunch of data scientists and engineers. I thought we needed people who understood the models, the parameters, the technical jargon. I was half right. We needed some of that, but our biggest breakthroughs came from an entirely different place.
Her name was Sarah. She was a creative writing major we had hired to help with social media. She knew nothing about APIs, token limits, or temperature settings. But she knew how to tell a story. She understood nuance, tone, and voice better than anyone on the team.
One afternoon, we were struggling to get our AI to generate compelling movie descriptions. The engineers were feeding it keywords, genres, and actor names. The results were bland, generic, and lifeless. "A thrilling action movie starring Tom Cruise as a spy who must save the world." You’ve read that a thousand times.
Sarah walked over, read the prompts, and said, "Can I try something?" She ignored the keywords. Instead, she wrote a short, evocative paragraph setting a scene. She described the feeling of the movie, the emotional state of the main character, the core conflict in human terms. She gave the AI a feeling, not just a set of instructions.
The output was stunning. It was witty, it was exciting, it had a unique voice. It was everything we had been trying and failing to produce. We quickly realized that the most important skill for working with generative AI wasn
's wasn't technical proficiency; it was creative and linguistic dexterity. It was about crafting the right narrative to guide the AI.
We completely changed our hiring strategy. We started looking for people with backgrounds in journalism, fiction writing, and even poetry. We created a new role: "AI Storyteller." These weren't just prompt engineers. They were the bridge between human intent and machine execution. They were the ones who could take a complex business goal and translate it into a story the AI could understand and expand upon. The impact on our marketing was immediate and dramatic.
Lesson 3: AI-Powered Performance Reviews Are a Trap
This one was a painful lesson. As an engineer and a data-driven guy, I love metrics. When AI tools emerged that promised to analyze employee activity—code commits, Slack messages, document edits—and generate performance insights, I was all in. It felt like the ultimate management hack. Objective, data-backed, and scalable. What could go wrong?
We rolled out a system at RemoteTeam that tracked dozens of digital signals. The idea was to create a fair and unbiased way to see who was contributing the most. In the first quarter, the reports looked amazing. We saw charts, leaderboards, and productivity scores. It felt like we had a god-like view of the company.
Then the culture started to rot. People weren't collaborating as much. They were gaming the system. Engineers were making dozens of tiny, meaningless commits to boost their numbers. People were having long, pointless conversations on Slack just to increase their activity metrics. The focus shifted from doing good work to looking like you were doing good work. The metrics were measuring activity, not impact.
One of our best engineers, a quiet but brilliant developer who would spend days thinking and then write a few hundred lines of perfect code, was flagged as a low performer. It was absurd. He was solving our hardest problems. But the system couldn't see that. It only saw his low commit frequency.
I pulled the plug on the whole experiment. It was a disaster. We went back to the old-fashioned way: managers talking to their people. We used AI to support these conversations, not replace them. For example, we used AI to summarize project progress from our task manager, which gave managers a good starting point for a discussion. But the judgment, the context, the human element—that was left to humans.
AI is a powerful tool for surfacing data, but it’s a terrible judge of character and impact. Trust your people, not the numbers on an AI-generated dashboard. True performance is nuanced and often invisible to a machine.
Lesson 4: Don't Automate Onboarding. Personalize It.
With a remote team, onboarding is everything. It’s your one shot to make a new hire feel like part of the team. My initial thought was to automate the heck out of it. Why have a human walk someone through the company handbook when an AI chatbot can do it? Why schedule a dozen intro meetings when an AI can create a personalized video playlist?
We built the most efficient, automated onboarding system you can imagine. New hires would get a series of emails, links to our knowledge base, and access to a chatbot that could answer any question they had. It was a well-oiled machine. And it was completely soulless.
Feedback from new hires was polite but lukewarm. They said the process was “efficient” but they didn’t feel connected. They didn’t know who to talk to with “dumb” questions. They felt isolated. We were optimizing for information transfer, but we were completely failing at human connection.
We scrapped it and started over. This time, we put a human at the center. Every new hire was assigned a “launch buddy”—an experienced team member from a different department. Their only job for the first week was to be the new hire’s guide and friend. They had coffee chats (virtual, of course). They did a screen-share walkthrough of our internal tools. They introduced them to people.
Where did the AI fit in? It became the launch buddy’s superpower. We built a tool that would give the launch buddy a daily digest of suggestions. “Hey, it looks like [New Hire] hasn’t set up their design software access yet. You might want to ping them.” Or, “Based on their profile, they’re interested in photography. You should introduce them to Maria from marketing, who runs our company photo club.”
The AI provided the prompts, the nudges, the personalized suggestions. But the human built the relationship. It was the perfect blend of high-tech and high-touch. Our retention rates for new hires improved by over 50% within six months.
The Real AI Revolution is About People
I’ve spent years and millions of dollars chasing the promise of AI. I’ve seen it do incredible things. But the biggest lesson I’ve learned is this: the goal of AI shouldn’t be to replace human judgment, but to augment it. It’s not about building a team of robots. It’s about building a team of super-powered humans.
Stop chasing every new tool. Focus on a few that work and integrate them deeply. Hire for creativity, not just technical skills. Don’t let AI replace the difficult, messy, but essential human work of management and connection. Use it to make those human interactions smarter and more meaningful.
The future of work isn’t a fully automated office. It’s a place where technology and humanity enhance each other. It took me a few painful failures to figure that out, but it’s a lesson I’ll carry with me to my next 200 investments and beyond.
Frequently Asked Questions
Can I implement all of these at once?
I'd strongly recommend against it. Pick the 2-3 items that resonate most with your current situation and focus there. Trying to do everything simultaneously is a recipe for doing nothing well.
How do I know which items apply to my situation?
Start by honestly assessing where your biggest bottleneck is right now. The items that address that specific constraint will give you the highest return on your time and energy.
Are these recommendations still relevant in 2026?
Absolutely. While specific tools and tactics change, the underlying principles remain consistent. I update my thinking regularly based on what I'm seeing in the market and across my portfolio companies.
How were these items selected?
Each item on this list comes from direct experience, either from building my own companies or from patterns I've observed across the 200+ startups I've invested in. I prioritize practical, actionable items over theoretical concepts.