LionClaw
When I was around 11 years old I used to have recurring daydreams about one day having my own robot army. You remember that cartoon Voltron? I think that cartoon and similar ones were the inspiration. Of course, that dream was never actualised and my life took a completely different turn.
So the rise of AI agents has been particularly interesting to me. I have been a software engineer for more than a decade now so some of my interest comes from that perspective. It is indeed mind-bending to watch my own skills acquired over decades quickly become commonplace. Pure existential dread. What will happen by December? Will I be selling maize by the roadside in 2030? Heh!
However, I also find it interesting to be alive today. Think of the possibilities. Before AI agents building certain kinds of software required hundreds of well paid software engineers and some budgets could approach the GDP of small countries. But now? Now all you need is an idea and the ability to describe in detail what you want and you can have it. The age of big tech may be over. Everyone can make their own personalised tools in a weekend.
However, big tech will not give up control that easily. Indeed, right now AI is mostly their forte. Mega-corporations are established to seize even more control. Before they controlled things like Gmail, Whatsapp, Facebook, Instagram, etc (which is basically everything "tech"), but if they end up controlling AI they control everything. Not only the tools you use, but also your deepest thoughts and desires. I find it terrifying.
Since anyone can build their own tools right now, I believe that we - together or as individuals - can fight this. I would like the future to be one where we are all super-empowered and have AI assistants (think of how everyone had a personal terminal connected to one of the AI minds in the Culture series of books). Your AI assistant would allow you to take on tasks big and small and you could enhance it using AI Agent skills or whatever replaces them in the future. You would not be tied to Google or ChatGPT or whatever. You would be able to use any tool and drop it when you are done or when it does not work as you want.
You see, I believe that the future should be one where you control your own tools and how they work. No one should be able to listen in unless you allow it. No one should be able to subliminally control (or police) your thought processes. And no one should be able to ban you from their platforms and leave you completely helpless.
We cannot sleepwalk into a future where a handful of large companies own the agents, the memory, the tools, the workflows, and eventually the shape of the work itself. We have seen this movie before. It was called Web 2.0, and it ended with everyone renting their own digital lives back from platforms.
I am building LionClaw as a small refusal of that future.
LionClaw is a local-first control plane for AI agents. The underlying model may be commercial. The agent harness may be Claude Code, a Gemini CLI, or something better that appears next month. That is fine. But the boundary around the agent - the project it runs in, the state it sees, the credentials it uses, and the audit trail it leaves behind - belongs entirely to you. LionClaw keeps that boundary local, explicit, and swappable.
The future won't be dominated by just one agent; it will be an ecosystem of runtimes coming and going. LionClaw provides a stable, highly auditable home for them. It is intentionally small so that one person can deeply understand and quickly change it within a single weekend.
That is the bet: real agents, local boundaries, user-owned control. Whether that means an agent sorting your email receipts for tax season, or a local interface securely managing your business contacts over WhatsApp, the foundation remains yours.
LionClaw is free. It is open and available for tinkering here: https://lionclaw.co.ke