My $450/mo AI setup just broke. So did my Meta account in 2018

If you haven’t been living under a rock, you probably know what openclaw is. Last week Anthropic released that they’re moving forward and disabling OAuth token usage for 3rd party tools and that openclaw will continue to run only if you change to API token usage. Which in reality is still fine but for power-users it’s a nightmare.

My ~$450/mo setup had to change and I had to think about how… more on that later in this email, but first

your stupid-simple high ROI insight.

Similar to Anthropic, a lot of my experience is with Meta, and when you’re platform-dependent you almost always have to have a plan B (even plan C)… With my historical affiliate marketing knowledge, I was experiencing a lot of Meta bans… and when I say a lot I mean it… At some point I had more than 500 ad accounts and most of them were banned.

Ad rejections. Profile restrictions. Even complete profile bans. My personal profile was forbidden to run ads for 3 years, from 2018-2021. I had to use my brother’s business manager to operate on client accounts. Thank you BRO!

But I learned something walking through all that hell so I thought it’s helpful to share it with you.

the structure

Every profile can create up to 2 Business Managers. Every Business Manager starts with 1 ad account, but if you add an app or verify your business Meta can unlock 3-5 additional ad accounts. Behave for a while and you can get up to 10 ad accounts per BM.

So the first thing I do when I work with new businesses is figure out how to implement a structure where I have:

• 2+ Business Managers
• 2+ ad accounts
• 2+ pixels

Here’s why this matters.

why 2+ business managers

Most people run everything through one BM. All their ad accounts, all their pixels, all their pages. One BM.

That’s one point of failure.

If Meta flags your BM for any reason… policy violation, suspicious activity, sometimes literally nothing… everything under it goes dark. Ad accounts. Pixels. Pages. All of it. Locked. And if you’ve ever tried to get Meta support to un-restrict a BM, you know how that goes. You submit a form. You wait. You submit again. Sometimes it comes back. Sometimes it doesn’t.

With 2 BMs, you have a backup that’s already warm. Already verified. Already has its own ad accounts and pixels. If BM1 gets flagged, you’re not starting from zero. You migrate the page to BM2 and you’re running again within a day, not a month.

why 2+ ad accounts

This isn’t about splitting your testing from your scaling. For most brands, one ad account works fine. Keep everything consolidated, let the algorithm learn from all your data in one place.

The second ad account is your backup. That’s it.

If your primary account gets flagged, restricted, or hit with a policy review… you’re not starting from zero. You have a warm, verified account ready to go on your second BM. You migrate your page, upload your audiences, and you’re running again.

The only other time I’d use a second ad account: if you want to test something completely out of the box. Weird angles, aggressive messaging, stuff that might catch a reviewer’s eye. Keep that away from your main account. If it gets flagged, your revenue campaigns don’t feel it.

Otherwise? One account. Keep it clean. Don’t overcomplicate it.

why 2+ pixels

Same logic. Your backup pixel runs in the background collecting the same data as your primary. You never optimize against it. It just sits there learning.

BM gets restricted? Pixel locked inside it. Accidentally delete your pixel? (I’ve seen it happen.) All that learning data, gone.

With a backup pixel installed on your site firing the same events, connected to your second BM… when something breaks, you switch. Your backup already has months of event data. You’re not starting from scratch.

the 15-minute version if you’re starting today

Not everyone needs the full setup right now. But here’s what I’d do minimum:

• Check if you have a second Business Manager. If not, create one. It’s free. Go to business.facebook.com/overview, click Create Account. Takes 2 minutes.
• Verify your business on both BMs. Upload your business docs. This unlocks additional ad accounts and makes your BMs harder to restrict. Meta treats verified businesses differently.
• Create one ad account on BM2. You don’t need to run anything on it yet. Just having it there, warm and verified, means you have a backup if BM1 gets flagged.
• Install a backup pixel. Add a second pixel to your site. Set it to fire the same events. Connect it to BM2. Let it learn in the background.

That’s the minimum. 15 minutes. And you’ll thank yourself the first time Meta does something unexpected to your primary setup.

back to the openclaw situation

So that was my Meta plan B lesson. Here’s how it connects to the AI situation.

I was running two AI agents on openclaw (BOSSMODE & BOSSMADRE)… two VPS servers at ~$7mo each, two anthropic max subscriptions at $200 each. ~$420/mo… automated workflows, content pipelines, client data pulls, ad creative generation, lead processing… everything working perfectly

Then the OAuth announcement.

API pricing for the same usage? BOSSMODE alone: $600-$1,000/mo. BOSSMADRE: ~$450. Total: $1,050 to $1,450 for the same thing.

Same lesson. Different platform. Platform dependency without a plan B.

I moved BOSSMADRE to codex (chatgpt) for now and spent a week extracting every piece of knowledge from both agent systems into a portable wiki. 150+ files. 65 structured pages. Every playbook, every framework, every client strategy. The agents will come back on API when the revenue justifies the cost. But the knowledge layer doesn’t depend on any single platform anymore.

Whether it’s Meta, Anthropic, Shopify, or whatever tool you’ve built your operations on… the question is always the same: if this changes overnight, what do I have?

Talk soon,

Kris

Kristijan Arapov is a DTC growth consultant who has managed $100M+ in Meta ad spend. He helps consumer brands break through revenue plateaus with AI-enabled marketing systems. Work with Kristijan