AdBeeCore

Features

One data model across every network

Four consoles, four ways of counting installs and four reports merged into one. That is not cosmetics: network comparison, automation and agent learning all start here.

A single layer over four networks

Adapters for Google Ads, Meta, Apple Search Ads and TikTok normalise data into one model: campaign, ad set, ad, keyword, metric. Everything else in the console works on that model.

  • Batched GAQL queries with quota control
  • One Meta call per account for all metrics
  • ASA API v1 with JWT and bids
  • TikTok throttling to respect rate limits

Agents with separated roles

Observer finds anomalies, Executor proposes and applies changes, Guard checks risk, Reflector scores the outcome after the attribution window, Curator updates the handbook. One clever prompt does not move money.

  • at most one change per campaign per 72 hours
  • daily budget caps per ad account
  • campaign and ad deletion forbidden
  • human confirmation for irreversible actions

Log and reproducibility

Every change comes with before/after values, author and time. You can reconstruct who raised a budget, by how much, and what came out of it.

  • change_log across all networks
  • hypotheses with an evaluation window
  • decision scores from −100 to +100
  • campaign config version history

ASO connected to paid

Keywords and rankings, custom product pages, review replies and the Apple Ads link: see where organic already covers a query and where paid traffic is required.

  • ranking tracking by country
  • keyword recommendations
  • CPP and listing previews
  • ASO and ASA without double-paying

Metrics the agents work with

One metric registry for all networks: same names, same calculation, currency converted on the way out.

Try it on your own account

Connect one account and switch observe mode on. Within a day you will see where the team is under-watching.