UiPath's CMO says the person who already ran paid media built an in-house agent that pulls ad data from platforms they don't own, analyzes it, and makes low-cost campaign changes around the clock so the team can shift money in real time instead of waiting on vendor reports.
Why it matters: The first agent worth standing up is the data-pull replacement written by the operator who already ran the process.
Google started rolling a Search terms and ad combinations view that ties each AI Max query to the exact headline, description, landing page, match source, and whether the copy was uploaded or generated. Common Thread's ecommerce playbook is sort by cost, flag mismatches, pin winners, and snapshot the top 50 rows before BFCM scale.
Why it matters: Yesterday's $0.67 AI Max receipt needed this report. Open it on the highest-spend Search campaign before you add more budget.
SALT's Reza Moaiandin ran 1,650 Claude answers across six buying scenarios and five languages. Same London–Tenerife flight: British Airways won 52.5% of picks in Persian and zero in Russian, while easyJet flipped from 12.5% to 72.5%.
Why it matters: If you sell outside English, the AEO scorecard is a language set, not a ChatGPT screenshot.
Topics Portfolio groups Copilot and Bing citations into themes and, for the first time, surfaces opportunity queries where other sites are cited and yours isn't. Volumes are ranges, and branded competitor queries can show up as "opportunities."
Why it matters: Treat it as a gap list for the content calendar, not another share-of-voice dashboard.
Mike King flags Anthropic's Nov 12 usage policy: using Claude to seed "independent" sources that manipulate AI answers is now an explicit ban, and you're on the hook for what your agents do through tools and browsers.
Why it matters: Audit every GEO tactic that only works if the model can't tell the source is you. Disclosure beats a throttled Claude seat.