Here’s How Isaiah Obi Obasi Built a Network Where AI Does the Work

Isaiah Obi Obasi started Syncmedia the way most affiliate networks start: with a long list of operational tasks that had to get done before any real work could begin.

Link formatting was first. His setup uses direct linking, which means every offer link has to be built manually: correct subparameters, correct structure, ready for bulk upload. Across 50-plus offers, that’s a full workday of work that produces nothing except a list of correctly formatted links.

He gave it to Claude Code instead. He hasn’t done it manually since.

⚡AI Stack

Isaiah's AI Stack

Claude Code Sheets Everflow

The Part That Used to Take Hours

Isaiah has been in the affiliate space for ten years. He came up on the partner side, running traffic for major networks: Madrivo, DFO, and others, all on Everflow. When he decided to build his own e-commerce affiliate network, he chose Everflow because, in his words, it was "the golden standard when it comes to affiliate network platforms." He'd already been using it for years. He just switched sides.

He gave Claude Code access to his Everflow tabs and a spreadsheet, explained what the links needed to look like, and let it work.

"I did over 50 offers in about an hour, because I was able to format all the links, and it would download into a CSV, and I would just bulk upload the additional links to Everflow."

The postback configuration he had running during the interview is the same approach — an agent working through each offer and event while Isaiah talked, paused for Zoom and still queued up when they finished. Hundreds of configurations, no developer setup, no API credentials: just instructions and browser access.

He's clear that this isn't advanced usage. He draws a contrast with more automated setups where the AI gets its own computer profile and can control everything on screen. His approach is more basic: Claude Code in a browser tab, following instructions the same way a trained team member would, just faster.

"Instead of hiring an assistant, or an Everflow expert, I can just give the AI the instructions, and it'll get the job done every time, and fast as well."

Creative Production at Scale

Take AI out of Isaiah's operation and the biggest gap isn't the link formatting. It's the creative side.

Good media buying requires test volume: multiple campaigns, multiple hooks, enough variables to actually learn something from your data. Before AI, producing one solid campaign with real A-B variation could consume most of a workday, and that assumed everything went smoothly.

Isaiah's team runs it differently. AI writes the script, handles the voiceover, sources the B-roll. The campaign body comes together in about 20-25 minutes. Generating 30 different hooks for testing takes another 20-25 minutes. Multiple campaigns, fully testable, in two to three hours.

"One good campaign might take the majority of the workday. With AI, you can pump out multiple campaigns, multiple A-B tests, in just a span of two to three hours."

Voice consistency isn't an issue. Isaiah uploads past creative as reference, and the AI matches the tone. If a revision is needed, it's a single instruction: change the energy, tighten the hook, adjust the framing. The AI understands what's already been done and applies the change consistently.

Where the Human Stays In

Isaiah is direct about where the line is.

Affiliate networks run on relationships. The trust between a network, its affiliates, and its advertisers isn't a soft metric. It's the operating layer the whole model depends on. Partner outreach, networking, business development: these are human conversations, and automating them carries a cost he's not willing to pay.

"I would not go as far as building an AI agent to communicate with partners, or to automate any type of networking, because you don't want to sound too robotic when you're reaching out to certain people."

He's seen the skepticism building on both sides of a deal. People are increasingly alert to whether they're talking to a person or a bot. Deploying AI into that space, he argues, risks exactly what a network depends on.

His frame: save AI for the technical, repetitive, and mundane so your team has more time for the work that actually requires a human. The goal isn't to remove people from the equation. It's to get them out of the postback configuration.

What He's Building Toward

Isaiah is vibe-coding a platform that connects to the Everflow API and generates weekly publisher reports. The idea is to take the campaign-level data already inside Everflow, the kind most affiliates never look at, and compile it into a clear document that goes out automatically every week.

The data point he's most focused on: events. E-commerce offers have multiple drop points in the funnel: advertorial, product page, checkout. Some creatives bottleneck users at the editorial stage. He wants to surface exactly where that's happening for each publisher, automatically, before they have to ask.

The longer-term vision connects to Everflow's API and the Facebook Ad Library: spot which products are building momentum for an upcoming season two or three months out, and get publishers on Syncmedia's network activated on those offers first.

One Bottleneck at a Time

The numbers are concrete. Fifty offers configured in about an hour instead of a full workday. Multiple campaigns with real A-B test variables in two to three hours instead of most of a day for one. Isaiah runs a lean team — and at a volume that would take a significantly larger operation without the AI layer absorbing the repetitive work.

Isaiah's advice for anyone who hasn't started isn't about AI in the abstract. It's operational.

Check the documentation for whatever platform you're running. A well-built platform will have already outlined what can be automated and how. Everflow's OpenAPI documentation is a concrete starting point. That's a faster on-ramp than experimenting blind.

Beyond that: look at what people in your specific field are already doing on X and YouTube. The workflows are there. Find what applies, test it, and optimize from there.

"Don't try to reinvent the wheel too early. Figure out what people are already doing, and then optimize from there, versus trying to build something completely from scratch."

The thing Isaiah keeps coming back to is simple: almost every bottleneck in your operation has a fix. He's been finding them one at a time since college, and he's still going. The agent he paused for the interview finished while you were reading this.