Halka Arz Haber — AI-Assisted IPO Tracker

A public-offering calendar is only useful while it is current, and a reader arriving at the moment a listing is announced wants more than the announcement — they want to know what to make of it. Writing a balanced assessment of every offering by hand does not scale, and letting a model write about finance freely is how an app ends up inventing a ratio or nudging someone into a trade.

A Cloud Function produces the evaluation: company data, financials and documents go to the model, which drafts a balanced assessment of the offering. The system prompt is editable from the admin panel, so the tone and the limits of the text are set by an editor rather than by a developer. The prompt’s rules are the substance of it: speak only from the data supplied, never invent a figure or a ratio, state a gap explicitly as "no data provided", treat upside and risk separately, and never steer the investor to buy or sell. Everything it produces is a draft, and an editor decides what is published. Around it, search visibility and ad revenue were designed in from the first release rather than retrofitted, with content structured for snippets and Discover.

Every offering gets an evaluation that reads as considered rather than promotional, and the boundary is structural rather than editorial goodwill: the model cannot produce a figure it was not given, and cannot tell anyone what to do. On the distribution side the app has produced substantial AdMob revenue on Android and iOS off steady visibility in search snippets and Discover.

A Flutter application on Firebase that follows Borsa İstanbul public offerings, with AI-drafted evaluations produced by a Cloud Function against an editor-controlled system prompt, plus a subscription tier.