The Tiny Teams Era Is Here and AI-Powered Startups Are Winning It

Quentin Peccoux vibe codes every day. He’s one of seven full-time employees at an AI-powered startup. Initially, he feared the technology would replace him. Now, he says it “feels like a superpower.”

He’s not alone in boasting about AI’s impact. Shivam Sagar, one of nine full-time employees at another company, said that AI agents can do the work of two to three additional engineers. The productivity boost is invigorating, but work-life balance is still tough to achieve — for his first six months in the job, he felt like all he did was eat, sleep, and code.

For better or worse, the Tiny Team era is here. Modern-day startups are proving that they can scale more quickly, reduce spending, and thrive against competition with only a handful of employees.

So what is it really like working on a team with only a few people who are human beings?

We asked founders and employees from startups with small staffs of fewer than 10 people about working alongside AI agents. We talked about what they like, what stresses them out, and what skills are AI-proof. These are the experiences they shared with us. Quotes have been edited for length and clarity.

‘As a team of 7, we can work like a team of 50 using AI’

Quentin Peccoux, 28, based in France, works as head of AI products and partnerships at Arcads AI, an advertising company founded in 2023 that creates AI-generated video ads. Arcads AI has seven full-time team members.

I worked for Arcads AI as a freelancer until I joined full-time in January. I led AI content operations, and I just became head of AI products and partnerships.

I used to work as an SEO strategist, so I can read HTML code, but I’m not a developer.

In AI content operations, I managed the gap between the product team and the tech team. I monitored what AI models were coming out and how we could use them. I also worked with the tech team to build and develop features.

The team uses AI to optimize the code base, diagnose what is happening in the code when there is a bug, and when we need to fetch any sort of data. I also use it for internal communication, writing, and for help with ideas.

I don’t see any part of the business where there’s no AI involved.

I vibe code basically every day. When I first tried vibe coding, it was to test some features, see the AI workflow, and understand what it could do locally on my device. But using an LLM, like Cursor, makes things a hundred times faster.

When I’m vibe coding, I feel like I’m right next to a senior developer and telling them what I need. It’s like having a superpower. But people who use this tool need to know how to read code to use it properly.

Even those of us who use a lot of AI in our daily workflows are still at 1% of what we could be doing with it. As it evolves, I think the strength of teams will shift toward people with ideas rather than people with hard skills.

‘It’s a heavy lift and a lot of pressure’

Shivam Sagar, 27, is based in northern India and is a senior software full-stack engineer at Aragon AI, an AI headshot generator company founded in 2022. Aragon AI has nine full-time team members and two offshore members for customer support.

It’s been around 10 months since I joined Aragon AI as a senior full-stack engineer, and during my first six months here, I struggled a lot with the lean business model. At my previous job, there were around a few dozen engineers on a team, so it was a huge transition to go to a team of nine people.

At a larger company, there are teams for the product, design, front-end, and back-end. For example, when you have a bigger team, there is a dedicated person working on the product planning who guides you through the flow. Here, we have to attend product meetings, understand the project, code it, and design the user experience and user interface ourselves. We have ownership of everything from start to finish, but our team has a tech lead to consult with if we get stuck.

The work-life balance is not very smooth. I’ll only get to spend time with my family and friends on the weekend. I don’t have much time during the week because of how fast we are moving.

Now that I have a better idea of the workflow, using AI tools has drastically improved my productivity. It can do the work of two or three engineers by assisting with research, coding, and reviewing.

‘Each one of us has the opportunity to play a huge role in the company’

Raul Alcantara, 25, based in Los Angeles, is a tech lead at Aragon AI.

I started at Aragon AI as a technical lead in June, and my job is to ensure that the code we push and what we deploy are up to a certain standard. I graduated with both my undergraduate and master’s degrees from Massachusetts Institute of Technology and had been working on my own startup for about a year before joining Aragon AI.

The expectation for speed using AI on a lean team gets intense. Previous biweekly sprint goals have turned into weekly sprint goals.

Each one of us has the opportunity to play a huge role in the company, which is exciting, but the velocity expected is hard. I’m trying to improve my ability to know when to take a break. This is not an eight-hour-a-day job, but that doesn’t mean it should be a 24-hour job.

‘We need to vet candidates more closely now because results don’t necessarily mean skill, with the way AI can build’

Sidhant Bendre, 25, based in New York, is the CEO and cofounder of Oleve, an AI-driven consumer software portfolio company founded in 2024. Oleve has six full-time employees and, in the last four months, has hired seven additional contracted employees for support needs.

AI existed when my partners and I started our company, so every job we’ve hired for has been augmented with it. Forty percent or above of everyone’s workflow is augmented or completely owned by AI, and that’s a conservative estimate that varies by person on the team.

Keeping the team small at first was, in some ways, driven by profitability, because profit gave us the power to keep exploring options and do what’s best for the company. The contractors we hire are all engineers for support, with the exception of one, who we use for marketing.

I try to hire specialists in one system and then expand their capabilities with AI to achieve more.

The rate at which we scale up and the pattern we follow will be quite different now with AI involved. It’s going to be seasonal in the sense that every time we discover new processes, we will put people in place to do those, but once they’ve figured it out, we can automate and augment the systems so they can do more than just that one process.

For example, when I hire an executive assistant, my expectation is that instead of doing the same tasks over and over again, my goal is that they work with me to build a system that leverages AI to do some of the work so they can move on to other systems that I need.

There’s a new lean startup operating set of principles, and that includes a world where we can start scaling whole teams of agents to work and be commanded by one person. Looking for employees also becomes about their capability to command.

‘AI has never fully run the show, but we’ve been able to leverage it’

Anada Lakra, 33, based in New York, is the CEO and cofounder of BoldVoice, an AI accent coaching app founded in 2021. BoldVoice has seven full-time team members.

The benefit of being a small team is that we don’t have bureaucracies, processes for the sake of processes, or meetings for the sake of meetings. We’ve been able to leverage AI to do the work in days that would have taken weeks.

In addition to myself and my cofounder, we have one product designer, two full-stack engineers, and two machine learning engineers. We also contract two voice coaches as needed. The engineers report to him, and I manage the designer and coaches.

Leveraging AI and staying lean is a superpower because it keeps costs down as you scale the product. It’s allowed us to grow very fast and, as of this year, become cashflow positive. We’re in a very good position where we don’t need outside funding.

If we had double or triple the team size, there would be lots of coordination needed, people issues coming up, and middle management would be needed to keep everything working in the same direction. It would become easy to lose priorities that way, and we just don’t need any of that right now.

When you have a very small team, there will always be limitations. Everyone’s stretched fairly thin, and there are always things that might pop up that are tempting. We’re not able to do them right away because we don’t have enough people, which can be frustrating.

But the days when you needed a team of a hundred to achieve $10 million annual recurring revenue are gone. Now, we can do it with a team of fewer than 10.