I’ve been in affiliate marketing for over 20 years now, and these days a big part of what I do is help companies grow faster by working with their teams to build automation that removes bottlenecks. That’s really what AI is great at right now: repetitive tasks that need clean, exact execution without a middle man constantly watching over them.
I’m writing this because I think the virtual assistant world is about to get hit hard, and being Filipino myself, that hits close to home. So much of the Philippines’ economy runs on VA work. But I don’t think the answer is to pretend this isn’t happening.
The answer is understanding how to use AI to extend what your assistant can do, before the work extends past them entirely.
Where I’ve seen it happen
In affiliate marketing, one of the biggest sources of human error has always been process. Not a lack of skill, just the natural way humans handle repetition: attention drifts, steps get skipped, someone’s having an off day. Automation rules built by AI don’t have that problem. Once they’re set up, they follow the process exactly, every time.
There’s no error, just a learning curve while you figure out how to use the tool. And once that curve is over, the repetitive task the VA used to own is basically obsolete. That’s the uncomfortable part: obsolete for the task usually means obsolete for the person doing it.
Three examples from my own work:
Affiliate network signups and follow-ups. A lot of networks now have API and MCP access, which means AI can actually interact with these systems directly instead of a person copy-pasting between tabs. That access lets the automation think through its own next steps instead of waiting on a human to tell it what to do next.
There’s usually still a bit of human involvement, someone reviewing an email response or approving something that needs a judgment call, but that’s maybe 10% of the work now. The other 90% runs on its own, and the human just clicks a button or two to keep the flow moving.
Creative approvals. I built this out for a network, and it’s honestly one of my favorite examples because of how direct it makes the relationship between advertiser and publisher. There’s a learning curve on both sides while people adjust, publishers submitting creatives, advertisers leaving notes on what’s approved and what’s not, but the best part is what happens after that curve.
Instead of someone building out internal approval guidelines by hand, the AI compounds the advertiser’s feedback over time and writes its own guidelines from it. Then it uses those guidelines to check every new creative that comes in from the publisher. No internal team second-guessing whether something passes. No back-and-forth waiting on a person to make the call. The system already knows the standard because it built the standard.
Creative creation. This one goes a step further than approvals, because it’s not just checking work, it’s making it. If you already have a proven video creative flow, one that you know converts, you can give AI the steps to follow: the angles to test, the hooks, the pacing, the format.
From there it can generate images and video following that exact flow, step by step, and just keep producing variations. Compare that to how it usually goes with a human video editor or graphic designer.
They look at their own work and say “this doesn’t look the way I want it,” and that’s not a flaw in them, that’s just how creative people work, they second-guess, they chase a feeling. AI doesn’t have that problem. It removes the emotion from the process entirely. It just follows the proven flow and outputs the next version.
Other places this is showing up
Those three are just what I’ve built directly, but I’m seeing the same pattern everywhere in this space:
- Reporting. Pulling performance numbers into a weekly report used to be a standing VA task. Now that’s a scheduled pull, formatted and sent automatically, no manual compiling required.
- Reconciliation. Matching payouts, invoices, and network statements is exactly the kind of repetitive, error-prone task automation handles better than a person checking rows by hand.
- Fraud and compliance flags. Instead of someone manually scanning traffic reports for anything that looks off, the system flags anomalies as they happen and routes only the real judgment calls to a human.
- Internal documentation. This is the one I think gets slept on the most. VAs often end up as the keepers of internal SOPs, updating the wiki, writing the how-to doc, training the next hire. AI can maintain that documentation itself, updating it as the process changes instead of someone remembering to go log it after the fact.
Where this leaves the virtual assistant
There are a lot of ways automation is quietly taking over work that used to belong to virtual assistants, in the Philippines and everywhere else this industry operates. And I won’t pretend it isn’t sad to watch smaller countries end up on the losing side of this shift. But adoption is what makes us strong, not resistance to it.
Learning has to be ongoing in this space. Affiliate marketing changes fast, and if you’re not adapting to the tools shaping it, you’re not just behind, you’re behind by years before you even notice it happened. The task is going to become obsolete either way. The question is whether the person doing it becomes obsolete along with it, or becomes the one who knows how to run the automation instead.
