Hiring for an AI-first startup. How to maximise revenue per employee.
What’s your startup’s revenue per employee? If you haven’t been thinking about this metric recently, you should be. Six practical tips for how to maximise it.
If you’re short on time, read the 30 second version of this post.
What’s your startup’s revenue per employee? If you haven’t been thinking about this metric recently, you should be. The smartest founders are re-evaluating how they build teams and staying as lean as possible. How we work faces the biggest disruption since the Industrial Revolution, so it’s time to tear up the rulebook for building high-performing teams. A small employee base leads to more runway and less need for consensus building. The outcome? Faster execution.
It’s not simply hiring fewer people or firing your existing team. A thoughtful piece I read this week challenged some of the current narrative around AI’s impact on jobs.
In fact, as I’ll argue, the widespread conviction that AI’s basic function is to replace human workers might be holding back the technology itself. AI entrepreneurs and engineers might be so focused on the vision of human-replacement that they might be ignoring much more productive — and much more lucrative — business opportunities. In fact, this could have parallels to the way entrepreneurs initially tried (and failed) to use electricity to power their factories in the early 20th century.1
A recent tweet from Gergely Orosz also raises the topic of entry-level candidates being more fluent with AI than their senior counterparts2.
So it’s not a simple case of reducing the number of employees. What’s clear is that we’ll need to adapt our approach to building teams to:
Setting up an AI-first culture
Hiring for curious entry-level talent, the tinkerers
Rethink the role your experienced staff play
We sat down with David Watkins, Co-founder and CTO at DASH Rides, a marketplace startup that allows individuals to buy cycle-to-work scheme-eligible bicycles and cycling apparel, to explore each of these points in more detail.
If you’re short on time, check out the Practical Tips ✨ sections below. This list is in no way exhaustive, so if you have other tips, please share them on LinkedIn.
1. Setting up an AI-first culture: Stretched & shipped is better than bloated & perfect
AI tools can be a game changer for your startup's execution speed, but only if your team fully embraces it. Startups are deliberately stretching themselves to figure out pain points and instead of hiring, they are automating and solving problems through rapid iteration. This means you need to give your team the freedom to experiment with the AI tools at their disposal:
Marketer? Try copy.ai.
Data analyst? Generate SQL with ChatGPT.
Engineer? Claude + Cursor.
For example, David recently had an entry-level colleague learn SQL from ChatGPT, which provided valuable insights into their email campaigns.
We generate rough versions at speed with AI and then incorporate feedback from the team to iterate to a shippable version. The beauty is in the iteration. We can now discuss what to fix in an analysis or what to change in a Figma design, meaning friction is reduced between functions. We start speaking the same language, whether visually or in code since we have something tangible to iterate on. I call this reducing ‘time to insult’, where everyone can tell me why my design is rubbish, which I can then improve.
Practical tips ✨
Tip 1: Force yourself and the team to stretch the limits of what's possible using AI. You should feel pain before you consider adding another team member.
Tip 2: Encourage rough drafts to centre conversations around tangible outputs. Ensure the correct expertise is brought in to iterate.
2. Hiring for curious entry-level talent, the tinkerers
Consider what being an entry-level candidate must feel like in 2024. They don’t know what the workplace looked like before AI came into our lives. And that’s a good thing. In the same way that many of us might be considered digital natives because we grew up with smartphones and the internet, we should consider entry-level candidates as AI natives or AI-first. These candidates think about how AI can simplify tasks and improve their output by default. As a result, they increase your startup’s efficiency. So how do you go about finding these young cyborgs?
David tests candidates with real work instead of hypothetical tasks. At DASH rides, they invite candidates for a paid day’s work and put them to the test.
We evaluate their performance, seeing how they work with the team and how they use tools to solve genuine problems. We assess their ability to collaborate, to push back and to get sh*t done.
Technically, this isn't a new way of assessing candidates. Linear – a startup with an inspiring approach to talent – does 2-5 day work trials, for example. What’s different now is how much you can expect an individual to achieve in less time (e.g. a day). Short work trials feel like a part of the future of how we hire, that is, as long as they are paid.
Paid work trials are further down the funnel though, and you still need signals from earlier in the candidate’s application. In David's case, he highlights examples of candidates showcasing their use of AI in their academic or side projects. Candidates showing initiative is a signal that has always been useful and will continue to be helpful in the future; how they show initiative is the only thing that changes.
Practical tips ✨
Tip 3: Use short paid work trials, not hypothetical tasks, to evaluate how candidates gel with your team.
Tip 4: Let entry-level candidates showcase their use of AI in their academic or side projects. The curious tinkerers will emerge.
3. Rethink the role experienced staff play: They’ll need to embrace the change
Senior staff naturally come with a wealth of experience. This can make them reluctant to try new tools and experiments. As such, you need curious minds looking to 10x not only their output but also the output of the people they work with.
They'll need to offer feedback to entry-level hires in their specialisation and broaden their skillsets in other specialisations. We have the tools now for a designer to pick up some code, an engineer to put some designs together, etc.
As mentioned, this is not to provide a final output but to facilitate collaboration around a rough (and quickly developed) first draft. When they feel they have reached their limits, they can draw on their experience or that of their colleagues to understand what additional resources are needed.
For example, David used ChatGPT to expand his coding skills. He writes code for the front and back end of his product. It wasn’t a perfect development process, but he got to workable solutions by tinkering and experimenting. His company kept progressing without immediately needing a new hire, and he's been able to 10x his output.
It’s also about soft skills. The ability to handle pushback, set a fast pace and nurture entry-level hires is more critical than ever. Covid & remote working have eroded basic communication skills, and we have a part of the workforce that has not experienced in-person communication five days a week. For senior hires, you'll need to understand the character outside of their hard skills deeply. What motivates them? What leadership traits do they exhibit? For reference, Amazon’s leadership principles have some (yes, some) excellent guidelines for startups.
Practical tips ✨
Tip 5: When hiring senior staff, focus on their character rather than just hard skills. Ask questions about conflict resolution and how they set their team's pace.
Tip 6: Hire seniors who can use their network to bring in specialist expertise, for when you reach the limits of what AI can help with.
Conclusion: Hire and create an AI-first culture
How do we stay lean and mean, then? Here are some of the tools that we recommend for keeping headcount low and velocity high.
Perplexity: This tool is fantastic for research and discovery. If you need reports or want to dig deep into a topic quickly, it is the tool for you.
ChatGPT or Claude: Swiss army knives. Use them as sounding boards, to write emails or generate SQL. Make sure you ask these foundational models before your colleagues.
Cursor: Once you have the base code generated from other models, Cursor can help non-engineers contribute to coding by simplifying the debugging process. However, ensure you have a senior in the loop to review this code.
Midjourney: Reduce the need for creative agencies to create assets for your company. Or at the very least gives you something to get them up to speed with what you’re looking for faster.
To effectively use these tools, they must be embraced deeply in an AI-first culture. AI first does not mean we lose the human touch. It means we use AI, like any other tool, to build and have rich conversations on works in progress. It gives us back our time from doing grunt work to focus on the more essential things. Blending this with a small team of tinkerers will allow you to produce more with less and keep your headcount low, maximising revenue per employee.
Finding these cyborgs is not easy. You should disrupt your interview processes to capture softer skills like initiative and collaboration. We have seen work trials starting to emerge as a remedy for assessing these people, and can’t wait to see what other new ways founders and hiring managers use to find good talent.