FAQ
Q: What kind of companies do you work with?
A: Marketing agencies running multiple clients across multiple paid channels. That's where scattered data and lost context cost the most. In-house teams with meaningful multi-channel spend fit too.
Q: We already have a data warehouse.
A: We've never seen an agency warehouse that actually works. Naming conventions nobody respects, pipelines broken since the last account migration, historical data gone. If yours genuinely works, great: we audit it, keep it, and build the context layer on top. If it doesn't, we fix it rather than rebuild from zero.
Q: We already document everything in Notion / ClickUp / Asana.
A: Good, you're ahead of most. But documentation a human can browse is not context an AI can use. We harvest what you have, structure it by account, date, and type, and wire it into the intelligence layer. The tool you write in stays; the structure changes.
Q: Will the AI touch our ad accounts?
A: No. The intelligence layer is read-only. It analyzes data and context, explains what changed and why, and recommends what to do. Your team decides and executes. Activation can come later, when you're ready. The foundation is the same either way.
Q: Why not just connect ChatGPT to our exports?
A: Two reasons. Scale: 20 clients at real spend produces more data than any chat context window can hold; the AI needs a warehouse and an operations registry that tells it exactly which tables to query. Context: without knowing the tracking broke last Tuesday or the client launched a TV campaign, even the best model gives generic recommendations. That's the difference between AI on top of exports and AI on top of a context warehouse.
Q: Are AI companies training on our data?
A: We're not Anthropic or OpenAI. We don't train on your data and we don't store it: everything lives in your warehouse, in your own cloud project. Which model reads it is your call. Copilot under your existing Microsoft license, an open-source model on a private cloud, or a frontier API once legal has read its terms. The LLM is swappable. The foundation is yours.
Q: Who owns the infrastructure?
A: You do. The warehouse runs in your own cloud project wherever possible, the data is yours, the dashboard is yours. We build it and optionally maintain it. No lock-in by design.
Q: How is this different from hiring a data engineer?
A: A data engineer builds what you can spec. The hard part is knowing what a marketing agency actually needs: which metrics matter, which context to capture, what the AI should read to produce recommendations a media buyer respects. We bring 15 years of paid media from the inside plus the engineering bench. You get the foundation without building a data team.
Q: Do you manage our ad campaigns directly?
A: No. We build the foundation and the operations automation around it. Your team stays in control of strategy, creative, and execution.