On Taking an Even Bigger Risk

At the start of July, I started work on a project that I've been hinting at for a while. I started work on an agentic productivity platform whose design popped into my head about the time when LLMs became truly useful and the Age of Agents began. Like I do with most of my ideas, I wrote it down and went back to work.

The way my brain is wired is that when I encounter a problem, a solution appears almost simultaneously sorta like a movie on fast-forward or like an exploded parts diagram in my imagination. I can see the pieces and I can feel the relationships between the pieces.

This was the case for me when I saw Claude Sonnet 3.7 and began using Claude Code. A number of things became self-evident that we, humans, would need to make use of the tooling. I'm sure others with experience building software anticipated the same needs.

Some of the items on that list:

• We would need a way to share SKILL.md files.
• We would want to run agent sessions each in their own computer.
• We would want to interact with agents in ways beyond the TUI.
• We would want headless agents with the ability to attach a variety of UI clients.
• We would want to define agents in a way that's durable and repeatable.
• We would want to avoid extra typing by having reusable prompts.

It's fun to have insights. It's what you do with them that really matters. Prior to July, my attention was focused on shipping Shopwatch, the changelog for your Shopify store. With that project launched, my attention was available.

In July, I decided it was time to take the risk and build the platform that's been sitting in my imagination and notebook. Lots of people are building agent tools, but nobody is solving the problem in quite the way I imagined. It seemed that my take would provide value. It's a tool I want for me, for Loam, and for the people I work with. So I jumped in and started building.

I was working quickly to get to parity with what I saw on the market. I know I need to have answers to the most obvious, "Can I do ____" questions people will ask. Fortunately, with a fleet of agents this is completely tractable. In the old days, it was not. Not unless you raised a big round. It's crazy now what someone with runway for living expenses and $300/month plan can do. You don't need to hire a big team. I made a ton of progress in the 6 weeks I've been working. I have a working early version of the platform.

When you're self-funded, aka spending your family's savings, whether or not you're heading in the right direction becomes something you reflect on daily. I took a break to go camping with the family the first week of August, so it's been a natural time to reflect as I get back in the flow. I asked myself: I've been at this a little over a month. What are my chances of succeeding?

My chances are strong. I think my vision is clear. If I can bring to life what I see in my mind, I think people will find great value in it. The main question will be whether I run out of money before I have enough to sell. It's the question every founder faces. The second question I've asked myself: Is what I have in mind differentiated enough that people will be willing to stop using the tools they are currently using and adopt my platform?

If I launch with an opinionated, curated take on table stakes, I don't think it is. I need to launch with some of the weird, fun, and powerful features that are on the roadmap. I need to find a way to get those things out earlier.

Aside from a few peeks, I have been off social media and not tracking industry news. The last couple days I have spent more time on X to catch up with the latest news. Yesterday, with the launch of Slack Code, I think my conclusion is sound.

The greatest force I have to compete with is people using a tool like Slack and it implementing a version of the features I envision. Even if Slack's version is not quite there, its existence lowers the motivation to switch. It's less disruptive to people to use a "meh" version of a feature than to adopt a new platform. The only thing that will entice them to uproot themselves is wanting a feature set that's disruptively better than what Slack will eventually build. It has to break the existing paradigm. I use Slack as an example, but this principle applies to any platform.

I need my Sketch to Figma moment.

Or I have to go after people who haven't chosen any platform yet. This cohort is larger than my X feed would indicate. Us software people are way ahead. The average business owner likely hasn't even begun to think about leveraging agents. If I make it good enough for the people and companies using tools like Slack, I know I can land these customers with a bit of sales effort.

So, then what should I do?

It's time to embrace the weird.