The best week to reach someone about fractional AI ops is the week they decide to hire an operations person. Tomorrow, my AI sales agent starts watching for exactly that. Of course, this is all theory until we find traction, but why not go public about our ideas of prospecting, ICP and why.
It goes live with one job: outbound prospecting, hooked straight into Apollo. Not a newsletter, not idea generation, not anything else. It finds the right companies and writes the first draft to them. Something short, personal and having to do with the job they're posting, and why they might consider a fractional role for the position.
It does not spray a cold list
Apollo holds a database of millions of companies and the people who run them. Most outbound tools treat a number that big as a licence to blast. I want the opposite. The agent filters that database down to my ideal customer profile, profitable small companies with a non-technical owner who is open to AI, and then it waits for one buy signal: that company is actively looking to hire an AI person, an AI-ops person, or a general operations person.
That signal is what it's all about though. A company posting for an ops hire has already admitted the pain out loud, and has already set aside the budget to fix it. They are about to spend six figures and months of recruiting on a full-time seat. That is the exact week to arrive with a fractional AI-ops team that gets to the same outcome for less, and sooner. I am not interrupting a stranger over cold outbound garbage that we all hate. I am showing up the week that person will not be annoyed, but intrigued by our offer.
Out of a database of millions, that is a small list on any given morning. Small is the point. Five to fifteen sharp drafts to people already raising their hand beats five hundred to a cold list that never asked. Been there, done that. Don't like it.
The one thing it cannot do is send
The morning a signal shows up, the agent writes a tailored draft to that company in my voice. Then it stops. It does not send. The draft is a plain text file marked "pending review," and it waits for the human in the loop to review. Jan is going to take that part over since he's more technical than me and can tweak the ai employee, and I'll take the leads. True collab style.
That job should be about five minutes. I open each draft, read it, and approve or discard. Approve and it loads into the Apollo sequence and the first touches go out on their own. Discard and nothing happens at all. Each draft a single decision, done before the coffee goes cold.
Why are we doing this through an openclaw fork instead of just doing it in Apollo itself? This will be more automated, and it will go through the signals daily on its own, making the tedious work of finding leads hands-off.
This is the AI salesperson I started building on Day 52, finally close enough to switch on. Back then it was a plan. Tomorrow it writes to its first real buy signals. That's the plan at least.
The boring part is the safe part
The loop that checks whether I have approved a draft runs every morning, and there is no AI in it. It opens each file, reads the status line, and asks one thing: does this say approved? If yes, it runs a permission check, makes the Apollo API call, and writes a line to an audit log. If no, it moves on. The model never touches that loop. The only place the model works is writing the drafts, which a human still has to read.
That split is deliberate. The thinking can be a model. The action I cannot take back needs a human eye on it, always, before anything sends.
I wrote almost the same sentence yesterday, about an AI reconciliation that ran clean and still missed half the statement. The lesson came from the books that time, and it is the same one here. It is the trust ladder pointed at sales instead of a ledger. AI proposes, a human disposes, nothing goes out on the model's confidence alone.
What changes tomorrow
For two years outbound was the work I avoided. I have been paying for the prospecting tools and barely opening them. Tomorrow a server watches a database of millions for the few companies about to hire the exact thing I do, writes to them the morning their hand goes up, and leaves fewer than twenty drafts in a folder for me.
I am not handing it the keys. I am handing it the first draft and the timing, and keeping the one decision that matters. Another daily automated sequence that can be done during the morning dirty chai, keto style.
Monthly Revenues $11,000 | Clients 2 | Prospects (AI outbound agent goes live tomorrow) | Team: Me + part time Jan (CTO)
Day 76 of 365.