AI is moving from “ask me anything” to “let me handle that.” OpenClaw makes that shift feel surprisingly real: voice-first, practical, and close to the kind of personal AI assistant many of us have been waiting for. In this commentary, Viktor Šohájek, who leads the AI technology team at FLO, shares what the tool taught him about productivity, delegation, and the mindset teams need before AI becomes a natural part of everyday work.
What OpenClaw Taught Me About AI-First Work
When I first heard about OpenClaw, I was sold. At the beginning of 2026, Peter Steinberger built something where you could just describe what you wanted automated in your Telegram, and it would actually do it. I installed the “AI that actually does things” out of curiosity, and within days, I was hooked. No steep learning curve, just natural conversation to get things done.
Then came the magic moments. I'm at a car service, and the mechanic calls me later that day to say that he needs my technical documentation. I take a picture of both sides of the plastic card, attach a picture of his visit card, snap a voice message to my OpenClaw on Telegram: "attach these, draft an email, send it.” Done in two minutes as I’m brewing my coffee. Later that day, the same thing with my insurance assistant. That's when I realised what this was actually giving me: the personal productivity boost.
E-mail digests, invoicing, calendar scheduling, Slack automation, you name it. But here's the honest part. Every ten days, I'm SSH-ing into my VPS to refresh auth tokens. Jobs fail silently sometimes. Memory fails sometimes. My Slack digest stops working when the auth expires, and suddenly it's not just me, it's my whole team waiting for their automation jobs to get done. I'm paying a hundred dollars a month to keep it running smoothly, and I'm still babysitting it constantly.
That's when I realised the platform itself is amazing but the biggest gain was the mindset it taught me. I started asking different questions about my work. Which tasks drain me? Which ones can I hand over to AI? Which ones do I actually want to do myself? Reading through emails to find what matters… That frustrates me. Sending a calendar invite… That's boring. But drafting a thoughtful message or building something new? That powers me up.
Here's what I learned: this isn't really about OpenClaw. It's about the idea it proved, that you can just describe what you want done, and it gets done. What we need now is to take that idea from clunky personal automation to something reliable and shareable, for both personal and work life. And although I’m still keeping my OpenClaw for voice-first smaller tasks for a little longer, from what I'm seeing, Claude is moving in exactly that direction — built-in memory for Claude Code, Routines and Dispatch are starting to look like what I actually wanted OpenClaw or Hermes to become. Anthropic has shipped enough incredible things already that I trust they'll figure out this as well.
The real opportunity, though? It's not about picking one tool. It's about making teams - especially non-tech ones - develop this same AI-first thinking. Building an instinct to recognise what can be delegated, what shouldn't be. How to think differently about work itself. That's where the real value of my OpenClaw experience lives.
This article was written by Viktor Šohájek, Head of AI technology at FLO.
Ready to build AI-first habits across your team?
Tools matter. But the real shift starts when people learn to recognise which parts of work can be delegated, automated, or redesigned with AI. At FLO, we help teams move from experimentation to practical, secure, and scalable AI adoption.
