Building My First AI Team

Recently, I’ve been using the new Grok Bot product from SpaceX AI via the Cursor team. The big idea for Grok Bot is that it’s a dedicated computer in the cloud that can do actual work by way of chatting with it. Of course, it has all of the power of AI and large language models, so the possibilities are endless.

I started by following a few different guides online to get going with Grok Bot and quickly set up a Chief of Staff bot to interface with the other bots. Then I set up individual bots to start doing work around some of my common day-to-day activities.

The easiest way to start with it is to think of it like a system that would be an employee. What would you assign as a task to an employee? Please build this spreadsheet. Please assemble the latest metrics and analysis for this report. Please review and provide recommendations for our next social media post. Please analyze what our competitors are doing in other markets. Please research the latest trends and best practices in our industry and summarize them daily.

Again, the key is to go from doing something like research or analysis to actually producing work from it and then putting that work product in the right place. The work product could be something like a new Google Doc with research. It could be a new Google Sheet with financial analysis. It could be ten new email drafts written in your tone and voice right inside your Gmail drafts folder.

Next, think of it like an employee where you give it work, it produces output, and it sends it to you, puts it in the cloud, sends it to your colleagues, or posts it in a Slack channel. It’s truly an AI teammate.

For me and my Grok Bots, I followed some of the best practices online and went with the Chief of Staff bot to interface with the other bots. Then I went with individual bots for things like monitoring all of the startups that I work with and sending me a weekly summary of anything new related to them, anything new on their respective websites, and any new comments or sentiment online about them. Stick it in a Markdown file and summarize it with a weekly email.

This example is similar to different products out there on the research side that will research things on a recurring basis and send you a summary. Nothing too fancy, but again, having the power of a large language model and AI means this can be made much more powerful over time.

As a simple example, anytime one of the startups that I work with issues a new press release, I can have the Grok Bot queue up an email draft in my Gmail account saying, “Congrats on this,” include a link to that press release, and then notify me that the draft is ready for my review.

Yes, I could do this by hand using some other tools. I could copy and paste the email draft by hand. I could have these pieces as part of my manual workflow. But by using AI plus a computer in the cloud, it’s always running. It never forgets. It does more research and has more access. The ultimate result is that it’s a repeatable workflow that slots right into my day-to-day. I get tremendous value from it, and it was low effort to set up and configure.

The next bot that I set up was the local startup activity bot. In a similar fashion, it reads all the websites that cover the local startup scene, summarizes what’s going on, and then sends me a weekly email with that information. Nothing fancy, but it’s an area of interest and focus of mine.

The next bot that I’m working on is around the individual startups that I work with. It aggregates all of their board decks, weekly updates, monthly updates, investor presentations, strategy docs, and anything and everything that I’m provided. It then uses the information from these proprietary sources and looks for its own trends and analysis, along with its own opinions on what it thinks the entrepreneur is doing well, what it thinks the entrepreneur is not doing well, its opinion on the market they’re in and the segment of the market, and any other nuggets or tidbits that it comes up with.

You could think of this as an analyst employee on the team, but again, constantly monitoring everything, constantly analyzing, constantly doing research, and then making it all actionable and feeding it up to me on a silver platter with minimal effort.

Entrepreneurs would do well to experiment with Grok Bot, start using their own AI employees, think through the rest of the team having AI as a team member, and implement this into the daily workflow.