Headless HQ

Why I’m building this

Running paid acquisition across more than one platform is a tedious problem. Meta, Google, and TikTok each have their own dashboard, their own definition of a conversion, and their own idea of what a good ad looks like. You end up tab-switching between three consoles, exporting CSVs, and trying to reconcile numbers that were never meant to line up.

That’s the version of the problem I hit first, because I’m the one running the ads for my own products. But it’s a specific case of something more general: a small company has a lot of operational work that is well-specified, repetitive, and spread across a dozen SaaS tools, and the only reason it takes a person is that nothing can hold the whole picture at once. That’s changed.

What it actually is

Headless HQ is a platform for specialized agents that do operational work across the tools a business already uses (30+ integrations today). An HQ agent routes each piece of work to the right specialist team. Every write action passes an approval gate the customer sets to full review, a spending threshold, or autonomous. That control is the part that makes it usable rather than terrifying.

Marketing shipped first, because it’s the one I needed. It runs campaigns across the Meta, Google, and TikTok ad APIs from one system, so launching, pausing, and moving spend doesn’t mean logging into three different places. The part I care most about is the creative analysis: most ad tooling only looks at the metrics, and this reads the creative itself (the copy, the framing, the hook) and connects it back to performance. Instead of “this ad spent $400 and got 12 conversions,” you get a read on why one variation is beating another, and what that suggests for the next one.

M&A diligence is the second deep team. It’s the same shape of problem as marketing: a lot of context spread across sources, and a set of judgments that benefit from something that can hold all of it at once.

FP&A and support come later. Support will get RAG-based auto-resolution against your own documentation. FP&A will take on the planning work that currently lives in a spreadsheet someone maintains by hand.

The architecture bet

The interesting problem isn’t any single agent. It’s cross-agent intelligence. The marketing agent knows what acquisition costs. An FP&A agent would know what the runway looks like, and a support agent what’s actually breaking for customers. Those three facts are related, and in most companies nothing connects them because they live in three different tools owned by three different people.

Building them on one platform, over one integration layer, with one approval model, is the whole reason to do this as a platform rather than separate products.

Where it’s going

I’m building it against my own products first, because I’m the one living in those dashboards and I’d rather find the failure modes myself than ship them. It’s free during beta while that’s true. Once the marketing agent reliably handles real spend inside the limits I set, the same bar applies to the next one.