Most people ask which tasks AI can partly handle. Wrong question. Start from: all digital work eventually gets automated. The direction is not in question anymore. The one thing that doesn't get automated: relationships. That's where you win now.

The wall that stopped most people from starting wasn't the idea. It was every gap around it. The code. The design. The ops. The reporting. That wall is gone. One person agents now out-produces a whole sequential team. The boom isn't coming. You're in the first hour of it.

Auto-pulled MRR into my dashboard. 5 lines of config. Stripe restricted API key. Read-only. Hits Stripe directly. No webhooks. No ETL. Most Stripe builds go heavyweight because the API could do damage. Read-only keys remove that risk. We wire these via Omni. DM.

My CRM is a folder. 234 companies in companies.md. 38 partners in partners.md. Call notes in call-notes/. Every change is a commit. I can grep my pipeline. Git blame tells me why a deal moved. Build via Omni. DM.

Replaced my email send layer in an afternoon. Old: a marketing tool I had to log into. Slow. Charged for things I didn't use. New: Resend API 200-line helper. Agents compose, schedule, ship directly. Most SaaS-tier features are a weekend project. Build via Omni. DM.

Every agent that touches the outside world clears with me first. Email? Show me. CRM update? Show me. Schedule a call? Show me. Sounds slow. Isn't. Agent does 95%, I tap one button. I run more agents than I'd otherwise trust. This is what Omni wires up. DM.

Spent years filing CRM notes in the wrong place. Now I drop a free-form note. Claude reads it, figures out the company and context, files it correctly. Asks only when genuinely ambiguous. The 80% CRM problem was the friction of choosing where. We build via Omni. DM.

Ctrl Shift \ pops a floating widget showing everything I'm working on as a tree. Python script reading one JSON file. Click an item, opens the right file in VS Code. Regenerates a printable A4 version on demand. 200 lines. We build these for Omni customers. DM.

Smallest thing I've built this year, the most used. Type `capture "Idea: ..."` from anywhere. Claude figures out where it goes. Filed automatically. Replaced "I'll write that down later" with thoughts that don't disappear. We wire these for Omni customers. DM me.

Calendly Standard only checks one calendar. Multi-calendar fix is $96/yr per seat. Built it into Omni in an evening. Sync mirrors busy blocks across calendars as opaque events. Calendly sees them. Conflict gone. Most upgrade-tier features are weekend builds. DM.

Put my own performance review on a cron. 5pm Friday a Slack message reads my week against my goals. Opens with "you said 2 demos. You shipped 1." 40 lines of script. Replaces the coach I thought I needed. We build accountability agents like this for Omni customers. DM me.

Simple software that are just front ends being controlled by Claude Code will become a new form of software. Already seeing it.

byo header smoke test 3 1777346299

Been running OpenClaw for a week and honestly it's wild how much it just handles on its own

When tools like Claude Code, Cursor, and Lovable emerged, most people saw them as new products. I saw them as signals. They weren’t the end state. They were the trigger for something much bigger. They lowered the capabilities required to build software. Once that happened, experimentation exploded. Internal tools replaced vendors. Small teams started shipping what used to require departments. That’s the real shift. And I still think we’re in the early innings here. In the last year the ease for anyone to work with code has become easier and easier.

This might sound counterintuitive, but I think we’re going to see more free software, not less. When building becomes cheap, the real constraint isn’t code it’s distribution. If you can build quickly and cheaply, giving the product away becomes a rational move. You capture attention. You embed into processes. You gather data. You position yourself around the outcome. The monetisation doesn’t disappear. It just moves outward into services, optimisation, performance-based pricing, and vertical solutions. In that world, software isn’t the product. It’s the entry point. There’s plenty of new opportunities here if you look close enough.

It's amazing to think my entire day is spent using agents now. This is the future. Once you start it's hard to do anything any other way. Honestly it's better work. You feel like you're achieving more, doing more and being more productive. I sit in Claude code for 3 to 4 hour sessions and it's unreal what can be achieved in that time. I even got OpenClaw (I call mine EDNAclaw now) to create a 20 page ebook on its source code the other day. Did it in about 5 minutes. Perfectly formatted, saved to my downloads folder on my computer. Wild times.

I set up my forked version of OpenClaw properly this past week. Cron running, browser access on, APIs wired in. I’ve even rebranded it and created a new UI that I like more. But wow it’s impressive. It just wakes up, checks things, pulls data, compares it, mades decisions, logs it, moves on. It’s working on a variety of tasks for me now without me having to do anything other than review the end output. No hand-holding. It just works so seamlessly. I’m now spending most of my time not “doing” work in the traditional sense, but designing the system, connecting the pieces, setting constraints, reviewing what the agents are doing. That’s basically my job now. It all started with Claude Code but it’s evolving even more now into almost all work, not just with code, that’s what OpenClaw is signalling to me. If this becomes standard inside organisations, and I don’t see why it wouldn’t, then a lot of what we call knowledge work today turns into managing and monitoring agents. More system design.

A lot of people assume SaaS companies are going to suddenly implode. I don’t think that’s how this plays out. I think what we’ll see is slower growth, more pricing pressure, longer sales cycles, and more internal rebuilds. When a customer can spin up a “good enough” version of your product using tools like Claude Code or Cursor in weeks instead of years, the negotiation dynamic changes. They don’t need to leave you immediately. But they don’t need you in the same way anymore either.

OpenClaw might not be the end destination, but it certainly is the new direction. This is the new wave of AI to be riding for sure, and I think it's going to change a lot quite quickly, even from what we've been used to over the last few years. One of the big insights for me is how many apps this actually takes away, or lowers the need for. You just need OpenClaw, the right skills, and the right API calls. It's also so wrapped around your daily workflow because you can access it so easily across all devices. It's not that simple to set up from my experience, but it still is wildly powerful once you can get it working.⁣