I want to tell you exactly what happened, because the way most people describe these wins strips out the part that actually matters: the specific decisions.
The niche I picked was real estate agents. Not because it was the most exciting option, but because I knew two of them personally, which meant I could validate before I built anything.
The problem I was solving
Real estate agents spend a surprising amount of their morning on information gathering. What did property prices do yesterday? Which of their leads haven’t heard from them in a week? What new listings came on the market in their patch? They check three or four different places for this, every single morning, before they can start their actual work.
I’d seen OpenClaw’s cron and search capabilities and thought: this is a 30-minute setup, not a $300/month SaaS subscription. So I built it, showed one agent the output, and asked if they’d pay $199/month for it delivered automatically every morning.
They said yes before I finished the sentence.
What the setup actually looks like
The agent runs at 6:45am via OpenClaw’s cron scheduler. It does four things:
- Web search for property price changes in a specified postcode area (Rightmove, Zoopla, local news)
- Checks a connected Gmail for any unread emails from leads, flags which ones haven’t had a reply in more than 3 days
- Pulls the agent’s calendar for the day from Google Calendar
- Checks a Google Sheet I set up as a simple CRM — lists all active leads and when they were last contacted
The output is a formatted Telegram message. Clean, readable, and arrives before they’ve had their first coffee.
The whole config took about four hours to build properly, including testing. The Google Calendar and Gmail connections were the fiddliest parts. Everything else was straightforward OpenClaw skill configuration.
The cold outreach that worked
Once I had a working demo, I messaged 20 estate agents on LinkedIn. Not a bulk template. Twenty individual messages that referenced something specific about each person’s work.
The message I used (adapted per person):
“Hi [name], I noticed you’re covering [area]. I’ve built an AI briefing tool specifically for estate agents that delivers a morning summary of local market moves, lead follow-up reminders, and your day’s calendar — all in one Telegram message before 7am. I’ve got two agents using it right now. Happy to show you a sample? No obligation.”
I got 6 replies from 20 messages. Three of them asked for a demo. Two signed up. One asked for changes I didn’t want to build, so I passed.
Two clients at $199/month from the first weekend of outreach: $398/month. Not $2k yet, but the foundation was there.
Getting to $2k
The next four weeks I sent the same message to another 80 agents. Response rate held around 25-30%. Of those who replied, about a third wanted a demo. Of demos, about 40% converted.
By week six I had 10 clients. $1,990/month. The service runs automatically. I spend maybe 30 minutes a week on it: checking nothing broke, responding to the occasional question.
The costs are low. Each daily briefing runs about $0.05-0.10 in API costs depending on how much web search the agent does. At $199/month per client, that’s roughly $1.50-3.00 in costs per client per month. The margin is north of 98%.
What I’d do differently
I’d have charged more. $199 was a guess. The agents who signed up would have paid $299 without much resistance. The value they’re getting is an hour of their morning back every day. That’s worth more than $199/month to most of them.
I’d also have built the outreach system inside OpenClaw from day one. I was doing LinkedIn manually, which doesn’t scale. Now I have an agent that monitors LinkedIn for new estate agents in my target areas and sends me a weekly shortlist of people to outreach to. That’s meta, but it works.
The repeatable lesson
Find a professional group that does the same information-gathering task every morning. Show them that OpenClaw can do it automatically. Charge a monthly fee that’s trivially small compared to the time it saves. Iterate on the product using revenue from the first five customers.
The hard part isn’t the technology. The hard part is picking a specific enough niche that your pitch feels personal, not generic.