Skip to content

Google Ads AI Agentic System

Taboola joins the Google system’s documented deployment lanes.

Version
v1.3.0
from
v1.2.0
Released
Type
Feature

This August release added a seven-skill native-ad pipeline, adversarial creative review, AI Max workflows and advertorial presells.

What to inspect first.

Suggested first inspection · based on this dated release
Your next stepWork and review
Inspect
Read the documented Taboola pipeline and choose one advertorial brief to inspect.
Retain if you try it
Retain its audience evidence, page direction and creative-review findings.
Review before use
Confirm which providers and approvals that lane requires; a documented skill is not an account already connected or authorized.

Source: The Google Ads AI Agentic System v1.3.0 · dated published release · Reading guidance from the published release; no exercise, generated artifact or customer result is claimed here.

Use the instructions and compatibility guidance for the version and access you actually have. Open my system materials →

What changed

  • Taboola is now a first-class deploy surface (7 new skills)

    prep, mine, launch, expand, harden, autogrow, and report. Includes publisher screening with tiered auto-blocking, day-1 realtime launch watch, and a pre-spend reach estimator. Together with the existing Google Ads engine and Microsoft Ads lane, the system now runs three deploy surfaces from one folder.

  • Adversarial roast panel (4 new skills)

    roast-copy, roast-image-ad, roast-video-ad, and roast-presell put every asset in front of a hostile expert panel before money touches it.

  • AI Max search (2 new skills)

    launch and autogrow Google's AI Max search campaigns alongside your classic Search builds.

  • Advertorial presell pipeline (2 new skills)

    write and deploy editorial advertorial presells end to end, the highest-performing page class in the theme library.

  • Deeper feed + cross-channel engine

    brand shopping feeds, category testing, feed sharpening, PMax consolidation, Meta-to-Google creative push, competitor spy briefs, audience sync, angle-performance mining, and campaign-level optimization.

  • Two editions, one upgrade for existing owners

    the product now ships as Blueprint ($4,997) and End-to-End Codebase ($19,997). Every Agent Codebase owner has been upgraded to the full End-to-End system at no charge - pull the latest from your private repo.

Release notes

This release turns the system from a Google Ads machine with side lanes into a three-surface paid acquisition engine - Google Ads, Microsoft Ads, and Taboola native - and grows the End-to-End Codebase from 149 to 173 skills. Nothing was removed from your workflow; one legacy skill (run-presell) was consolidated into the presell gate chain.

The headline: Taboola, end to end (7 skills)

The complete native-advertising pipeline we run in production:

  • taboola-prep / taboola-mine - account and audience preparation plus demand mining before a dollar is spent.
  • taboola-launch - guarded launches with a day-1 realtime watch (site-level and item-level learning states, account-timezone windows).
  • taboola-expand / taboola-autogrow - reach estimation before scaling and weekly autogrowth loops after.
  • taboola-harden - publisher screening with a canonical pattern allowlist and watchlist, tiered auto-blocking for fraud-pattern publishers, and a combined cap guard.
  • taboola-report - performance reporting wired into the same brand reporting spine as Google.

The roast panel: adversarial review before spend (4 skills)

Copy, image ads, video ads, and presell pages each get a dedicated hostile-review skill. The panel attacks claims, hooks, congruence, and compliance, and returns a ship / fix / kill verdict - the same gate discipline the launch engine already applies, now pointed at creative quality.

AI Max search (2 skills)

search-aimax-launch and search-aimax-autogrow bring Google's AI Max campaigns into the system with the same guarded-launch and weekly-growth discipline as classic Search.

Advertorial presells (2 skills)

presell-write-advertorial and presell-deploy-advertorial ship the editorial advertorial format - research-grounded story pages with the full editorial hat - as a first-class pipeline from copy to deployed HTML.

Deeper engine across the board

  • Feed: feed-brand-shopping, feed-category-test, and feed-sharpen extend the Shopping feed machine with brand-line feeds, structured category testing, and title sharpening.
  • PMax: pmax-consolidate merges overlapping asset groups without losing learning.
  • Cross-channel: meta-push-google carries winning Meta creative and audience signal into Google campaigns; brand-audience-sync keeps audience definitions aligned across surfaces.
  • Intelligence: spy-brief compresses competitor research into an action brief; brand-mine-angle-performance mines which angles actually convert; brand-optimize-campaign adds campaign-level optimization passes; verify-work adds a self-audit gate to every long-running job.
  • Reference layer: the per-skill reference library has grown to over 1,600 documents - every phase of every skill now carries its own execution reference, so the system stays reliable as it grows.

Two editions, and what it means for you

The product now ships in two editions: Blueprint ($4,997) and the End-to-End Codebase ($19,997). The middle Agent Codebase tier has been retired - and every owner of it has been upgraded to the full End-to-End system at no charge. If you owned the Agent Codebase, your private repo access now includes the complete 173-skill system: pull the latest and you are current.

By the numbers

  • End-to-End Codebase: 149 → 173 skills
  • New deploy surface: Taboola (joining Google Ads and Microsoft Ads)
  • Reference layer: 1,600+ execution reference documents
  • Blueprint curriculum: unchanged

Existing owners pull the latest from your private codebase repo - no upgrade charge. Pricing is unchanged.

Review current system scope