AI is changing how founders and teams work.

We’ve been hearing valuable insights from founders participating at some of our recent events about how they are utilizing AI. They shared specific AI workflows for improved engineering productivity, streamlining hiring processes, building out operations, creating custom AI tools, and the transformative effect of building an AI-first culture.

Here are some highlights.

AI for engineering productivity

Ivan Lee, founder and CEO of Datasaur, said he has seen a remarkable increase in engineering productivity across his team after implementing AI coding assistants.

“If you told me I could adopt one thing and get a 26% productivity increase, that is insane,” he said. “It is an incredibly valuable technology.”

The process was an evolution that required trial and error of various tools to see what his team would adopt. They formed a committee of a dozen engineers dedicated to adapting and learning the latest best practices and found that different engineers adapted to AI tools at varying rates. Some saw no change in productivity while those who were quick to adopt experienced a 30%+ improvement. They decided to host brown bags to teach each other how they should be using these tools, even though Datasaur is itself a state-of-the-art AI company. Learning how to best leverage AI tools is still a learning process for everyone.

Understanding the human behavior involved in adopting AI within their own engineering team was helpful in thinking through how their non-engineering clients would adopt their AI products as well.

Building operations with AI

Amir Elaguizy, cofounder and CEO of both Poker Skill and Cratejoy is at the far end of the AI adoption curve. He described his typical workflow, which involves multiple AI models running simultaneously across five monitors.

“I have a deep research one that’s going way ahead, [ones running on top of LLMs], I have a testing one that’s making sure the unit tests all passed and doing all the transactional crap, and then I have the one that I’m coding on real time, taking the research from something I synthesized over here.”

“We use AI to power how we teach you how to play [poker], but we also have used AI to build the whole company. I talk to AI all day, every day, multiple AIs at once, they’re first-class citizens at our company. We hire people explicitly to drive AI all day at both of my companies.”

At Cratejoy he’s enabled his small team to achieve what would normally require a much larger organization by using AI for customer service, marketing, merchandising and vendor management. Amir reduced headcount while improving business performance through AI-powered operations.

“The velocity is compounding to the point that it’s starting to be hard to bring anybody onto the team, because they’re entering this machine that’s just going faster and faster and faster and faster and we speak a foreign language inside the company.”

AI for recruiting

Siqi Chen, cofounder and CEO of Runway described how using AI could keep his company under a certain number of people with “the distance between an idea and the realization so much more compressed.” This has affected two things – how they hire and the types of people they hire. 

“We have an unusually senior team…and we have an unusually high agency team. So, one example is we have account executives regularly pushing code all the time. They hear a thing from a customer and say, ‘I’m just going to go build that’. You want to hire for that kind of personality.” 

Dmitry Alexin, founder and CEO of Handoff also detailed how Handoff has transformed their recruiting process from a three week process to 30 mins.

“Recruiting is one of the highest risks. The upfront thinking required to create a scorecard, think through competencies, develop an interview process that would normally require a whole recruiting team -we’ve automated that with AI. It builds your whole interview bank of questions, prioritizes them, you describe what you want…the job description is then built after with that information, making it much simpler. Our hiring bar is so much higher and we’re so much more precise about who we look for. Our hit rate for successful recruiting has drastically increased.”

An AI-first culture

Siqi talked about how they intentionally promote an AI first culture at Runway. 

“Model the behavior you want. I constantly try to create new tools. In fact, if you go to the GPT store, under Education, the number one or number two GPT is created by me called Universal Primer. We really encourage everyone in the company to build these tools and that’s one of the underrated things about AI, that tool creation has never been more accessible to everyone in the company.”

“Today there’s a bunch of tools that are commercially available. Two and a half years ago when we first launched there wasn’t as much of it and we built a lot of tools in house. We built an AI lead qualification tool. You can tell it – describe the kind of leads that are qualified – and in Slack it’ll automatically enrich it, decide whether it’s qualified or not.”

Dmitry described how his team uses AI to create an informative and fast feedback loop essential to building a successful business. 

“Feedback loops really get stuck in internal company processes. Every single person is not only allowed but required to run experiments within their domain and outside of their domain…to test their idea, whether it’s product, go to market, or something else. It takes time. You have to think about it. The setup for the experiment is the cold start problem and very often prevents people from doing it. We automated that with an AI template that helps think it through.”

Handoff now averages 2.7 experiments per week with a small team, has sped up the experimentation feedback loop, and the pace continues to accelerate.

Using AI intentionally

Everyone agrees that AI is fundamentally changing how startups operate, but taking advantage of it requires an intentional approach, from adoption to pushing the tools to do what you want and need them to do. 

As Amir put it, “People are waiting for AI to be packaged up with a ‘do my job’ button, and that’s not how it works… By the time somebody’s selling the ‘do my job’ button, you don’t have a job.”