A VC Take on Ethan Mollick's New Book, Co-Intelligence: Living and Working With AI

Lately, I have been obsessed with the role AI will play in the lives of startup founders. AI tools are allowing founders to more efficiently and effectively find product market fit and scale their ventures. To put a founder’s spin on a recently popular saying:  AI may not replace founders. But founders who use AI will replace founders who don’t.

It is with this perspective that I read Wharton Professor Ethan Mollick’s terrific new book, "Co-Intelligence: Living and Working With AI," which offers helpful tips and practical frameworks that any leader can apply to harness the power of AI in their organization. I read the book through the lens of a founder and kept asking myself, “How should I apply these frameworks to be more effective in building and scaling my startup?”

Mollick has become famous for his newsletter on AI called One Useful Thing which seems particularly targeted to educators. Everyone in my Harvard Business School community loves his newsletter for its high-level observations as well as tactical suggestions about how to incorporate AI into the classroom. He is also thoughtful about the impact AI is going to have on the world at large. Thus, much of his new book is about AI in education and society.

For founders, the relevant material comes late in the book when Mollick introduces an approach to categorizing tasks within the workplace, distinguishing between "Just Me" tasks, "Delegated" tasks, and "Automated" tasks. This framework is useful for founders aiming to maximize efficiency without compromising the unique value of human insight.

Here's a breakdown:

• Just Me Tasks are those that critically require human judgment and personal expertise—areas where AI cannot replicate the nuanced understanding and emotional intelligence of a human. For founders, hiring is a Just Me Task. Founders may able to use modern AI tools to provide candidate screening and interview intelligence (e.g., using our portfolio company BrightHire's AI-powered interview platform to build job descriptions, interview plans, and take interview notes) but a hiring decision is ultimately a Just Me Task that a founder must make on their own.
• Delegated Tasks are those that you assign an AI, perhaps to get you started, but carefully check before completing. Blogging and content marketing have become a Delegated Task for founders. Most of my sophisticated portfolio companies are using AI to automatically generate a ton of content for marketing campaigns and Search Engine Optimization. Each individual piece of content is not super important and it is relatively tedious to create all the various permutations. Another portfolio company has created a tool to automatically generate draft responses to RFPs, having trained the model on previous RFP responses that they have provided. Founders are doing thoughtful quality control on this AI-created content before sending it out to make sure there are no errors (aka hallucinations) or wonky-sounding phrases.
• Automated Tasks are ones you can leave completely to the AI and don't even need to check on and provide quality control. For example, one popular Automated Task founders are using is to point LLMs to their customer feedback (through all the various channels) to identify and summarize themes that require product feature adjustments and then communicate those feature priorities to the members of the team who set the product roadmap. LLMs are very skilled at summarizing large volumes of content.

The more time I spend playing around with these tools, the more I am convinced that every founder should be using them to assist them in their journey to find product-market fit and to scale their startups. Reading Mollick's book and studying his tasks framework will get founders thinking more deeply about what processes they might delegate and automate, speeding up their iteration cycles and experimentation.