01 · Dashboards lost

Every tool a small merchant buys arrives as another dashboard. Another login, another tab, another place where numbers are technically available and practically ignored. The data was never the problem. Sitting down in front of it was.

A question asked in a chat window has none of that friction. You ask while you are packing an order, and the answer arrives before you have taped the box.

02 · The gap is not information, it is attention

Shopify already knows your stock is at three units. Nobody told you, because telling you was your job. The useful layer is not another report; it is something that holds the whole picture and speaks up when it matters.

03 · Read before write

It is tempting to build an agent that edits prices, reorders stock and replies to customers on day one. We think that is how you lose someone's trust in a single afternoon. Answering questions is useful on its own, and it is safe: the worst case is a wrong sentence, not a wrong order. Write actions come later, one at a time, each with a confirmation.

04 · Small stores first

Enterprises already have an ops person who knows the numbers. A one-person brand has a spreadsheet and a memory. That is where an agent changes the day, and it is the segment every serious inventory tool prices out.

05 · What we do not know

• Whether sellers will trust a chat agent with store data at all
• Whether the answers stay accurate as catalogues get messy
• What the economics look like once model usage is not on us
• Whether agent-to-agent networks turn into anything real, or stay a demo

We will put the answers on this page as we get them, including the ones that make the idea look worse.

06 · Built in public, honestly

Plenty of projects in this space ship the website and the token first, and describe a product that does not exist yet in the present tense. We would rather have a smaller page that is true. If something here is listed as live, you can go and use it today; if it is not built, it sits under "Later" on the roadmap.