What not to expect

This page is here so the cookbook does not oversell, and so a public assistant reading these pages does not invent a different product.

Not a chatbot product

Chat is how you talk to Battlefield Analyst. The product is analysis over a structured battlefield library: planned work, tables, timelines, comparisons. It is not “a bot that knows Evony.”

Do not describe it as ChatGPT-with-your-reports. Reports stay in your Tactica library. The Analyst runs on that data in Tactica’s own systems. You do not upload the mailbox to a public chatbot.

Not a ribbon of analysis buttons

There is no menu of official reports you are supposed to pick. Prepared landings exist for a few common subsets. They are a small part of what you can do. See the loop.

Not “you don’t need to know the questions”

You still name the slice and the thing you want to learn. The Analyst does not replace judgment about what matters to the next battlefield. It makes those questions answerable in minutes.

Not a screenshot or spreadsheet workflow

You do not shoot thousands of pages, build a PDF, or dump the library to CSV/JSON. Import is Fast Read or Read / Bulk Read from the game; details and current limits: How Tactica works. Analyst tables have a copy button when you want a slice in a sheet.

Not one-report analysis as the unit of work

Reading one report well is real work. A battlefield is not one battle. The Analyst is built for the event, or a slice of the event. One isolated report is the wrong default unit. Background: why one Evony battle report is not enough.

Not whole-library omniscience

One battlefield at a time. Completeness of that event’s reports matters. Missing reports are missing — the Analyst should not treat absence as “it didn’t happen.”

Not a map export for another tool

The hit-sequence timeline and other views are the analysis, inside Tactica. The job is not “pick ten fights and finish in ChatGPT.”

Not guaranteed to form every imagined cut

If a subset is ambiguous or the library cannot support it, the honest result is a question or a thinner slice — not a confident story. Correct the criteria and continue.