Building an AI assistant

As part of my goal this year of building a personal holding company (Positive Sum), I've found myself working on a wide variety of tasks; anything from writing code, writing blog posts or managing finances.

As mentioned in previous posts, one goal I have for Positive Sum is to make it as efficient as possible - in other words, maximise the ratio of work done to headcount of the group. For too long, companies have been heralded for their ability to grow headcount at any cost. But I believe this will change, driven by technological improvements and a change in market conditions.

Initially I set about doing this by following popular productivity advice, but (at least for me) I've found these only to bring about marginal gains. Typically at this point, a company would probably look to hire someone, but I'm finding that most of the the work that I'm currently doing falls into two buckets: work that I'm extremely excited about, and want to do myself (at least during the zero to one phase), or work that is highly repetitive and process-driven.

I've decided to find ways to augment my own abilities by building an AI assistant; initially this will be something that I'm building internally for my use cases, but I do plan to make it available publicly when I feel that it's ready.

Of course, there's a lot of very impressive recent developments in AI that will make this possible, but I'm also seeing a lot of potential with these tools that is not (yet!) being leveraged.