Finding the right person to build a business with is no easy task, but programming one could be.
Between the working world marching in AI’s direction and AI replacing humans in almost every capacity — as therapists, concierges, government ministers, even lovers — it begs the question: Could the tech make a good co-founder?
To find out, Microsoft conducted a semester-long study with NYU Stern’s Tech MBA program, and the results were… not exactly shocking.
The methodology
Students were divided into startup-esque teams, given access to Microsoft’s AI agent Copilot, and asked to build “Frontier Firm” companies, AKA AI-first startups that “don’t just adopt AI, but are built around it,” per Harvard Business Review.
- Students simulated launching a startup using the AI to assign roles, outline plans, generate content, and test workflows.
- Then they mapped out AI-first organizational models, imagining how Frontier Firms could operate with AI at every level.
The results
HBR breaks it down into four parts:
- AI as a “first hire” — teams used AI for key roles like strategy and analysis, helping them move faster with fewer people and focus on higher-value work. Instead of wondering who to hire first, they started asking how far AI could take them and what roles they really needed to fill.
- Dialogue over documents — instead of starting from scratch, students talked to AI to generate first drafts (of docs, logos, pitch decks, etc.), then refined and contextualized the outputs.
- The role of human knowledge — AI made decision-making faster and cheaper, letting students explore more “what ifs” and focus on judging ideas instead of generating them. But! While it filled in knowledge gaps, it still needed human judgment to catch errors and add nuance.
- Reshaping team structures — humans became orchestrators of AI agents that handled tasks across domains. Basically, the tech became a digital workforce managed by humans, rather than a tool.
What does this mean?
In theory, employing AI as a foundational team member would save companies time and money by making organizations leaner and passing grunt work onto LLMs.
- Established companies have hyped AI adoption for the same benefits, though their “plug-and-play” approach has largely failed to deliver.
So, having an AI co-founder might not be a bad idea — it would just require radically changing the nature of hiring, workplace dynamics and workforces, and how we value ourselves…
What, like supposed “the biggest shift of our lifetimes” wasn’t going to come with some major shakeups?
Ai
