I have a specific ChatGPT chat that rewrites my posts really well every time; I usually just give it a rough draft and let it clean things up. But it was pretty interesting to realize how good these LLMs are at actually finding mistakes rather than just rewriting things. The point about not letting the model influence your voice was also very, very relevant.

One of the most useful use cases for me, especially since I work in GTM, is keeping track of new ways to get replies from prospects, whether through cold outbound or things like SEO/GEO optimization. I read at least 2-3 articles a day on this, and it genuinely helps me figure out which ideas are worth trying because I can now keep track of everything I’ve read.

Customers often know the outcome they want to achieve, but have no idea what steps they need to take to get there. For example, they might want to launch a campaign, but don’t know where to start or what they need to configure.

Having something like an AI assistant that understands the product and guides them through every step of the process could be really helpful. It could essentially take the user from the outcome they want to achieve and walk them through the entire workflow.

With coding agents, what I’ve noticed is that a simple task can often be handled with a fairly simple harness. But the hidden cost is really around context. One of the more interesting things I’ve seen is that two different harnesses can make a similar number of model calls while consuming a very different amount of context.

I think I recently came across a paper comparing Claude Code and Pi that touched on this. More context, more tooling, focused context, simpler loops, all of these can lead to very different costs and performance, even when the number of model calls looks similar.

More and more non-technical people can now build things they’ve always wanted to build, including their own websites.

As a non-technical person myself, I built my own website simply because I’ve always loved building things, but never really had the medium or technical skills to do it.

Now I do and the medium is Claude code.