OpenAI just launched the most significant new paid ad surface since Google’s early search ads — and the window to lock in low CPCs before mainstream advertisers arrive is measured in days, not months. This week’s intelligence also covers Claude’s invasion of Slack, a real 30% API cost reduction, and hard data on which content format earns the most AI citations.
ChatGPT Ads Go Live With Self-Serve CPC Buying
ChatGPT Ads are officially live in the US with self-serve buying, CPC bidding, a Conversions API, and real-time contextual targeting appearing below AI-generated responses. Users seeing these ads are already mid-query with specific intent — this is search-quality intent without the keyword auction that has inflated Google CPCs for two decades. OpenAI building a Conversions API on day one signals this platform was architected for performance marketers, not brand advertisers.
Get into the ChatGPT Ads self-serve platform this week and launch one test campaign before mainstream advertisers discover it and CPCs spike.
Read the full story →
Try it yourself →
Join the discussion →
Claude Tag Embeds Full Claude Code Into Slack
Anthropic launched Claude Tag, bringing skills, connectors, memory, and scheduled tasks — the full Claude Code feature set — directly into Slack for team-wide AI collaboration. The strategic move is distribution: 99% of business users are already in Slack daily but never open Claude Code, so Anthropic is embedding AI adoption into existing workflows rather than demanding behavior change. The persistent memory layer is the real unlock, turning a chatbot into a context-aware team member across every conversation.
Pilot Claude Tag this week on one repeatable marketing workflow — brief generation, title ideation, or campaign review — where your team already collaborates in Slack.
Cut Claude API Costs 30% With Image Billing Arbitrage
Rendering large context windows as small-text images cuts Claude API input costs by approximately 30% because image billing is fixed by pixel dimensions, not token count — verified at $1.30 for text versus $0.69 for image on identical prompt contexts. The deeper architectural insight is that Fable orchestrating cheaper Opus or Sonnet sub-agents produces near-identical results to all-Fable pipelines at a fraction of the cost, meaning model routing is a productivity decision, not a quality tradeoff. As Karpathy’s framing makes clear: the model is not the moat — the process and system architecture are.
Switch high-volume Claude Code marketing workflows to Fable-as-orchestrator with Sonnet sub-agents now to capture the cost savings before the next pricing revision patches the image billing gap.
‘Best Of’ Listicles Earn the Most AI Citations
Glen Allsopp’s analysis of 750 ChatGPT prompts found that ‘best [category]’ listicles are the most frequently cited source type in AI-generated answers — the first hard, prompt-volume-backed data on which content formats earn AI visibility. This reframes the humble ‘best of’ post from a low-prestige SEO tactic into a high-value AI answer inclusion asset that content marketing teams should be prioritizing immediately. The warning from Ahrefs is equally clear: build it to genuinely help users, not to game citations, because AI systems will eventually detect and downrank content optimised purely for citation gaming.
Audit your existing ‘best [category]’ content this week and update it with specificity, recency, and genuine curation — these pages are your highest-leverage AI citation candidates right now.
Read the full story →
Join the discussion →
Google’s Open Knowledge Format: The Next AEO Primitive
Google published the Open Knowledge Format (OKF) under Apache 2.0 — a spec that structures website knowledge as a folder of small linked markdown files designed to be traversed by AI agents across organizations. OKF formalizes the CLAUDE.md memory-folder pattern that practitioners already built organically, which is historically a strong signal that a pattern is about to become infrastructure, much like schema.org markup became table stakes for structured search. Websites that implement it early give AI agents a machine-readable map of their expertise before it becomes a citation or ranking factor.
Read the OKF spec on GitHub this week and assess whether your website or product documentation can be restructured in this format before adoption becomes mandatory for AI discoverability.
Read the full story →
Try it yourself →
Join the discussion →
Cloudflare Forces AI Companies to Pay Publishers
Cloudflare’s new policy creates an infrastructure-level mechanism pushing AI companies to pay publishers for content crawled via its network, drawing a formal line between permissioned and unpermissioned AI training data access. Because Cloudflare sits in front of a significant share of the public web, this gives publishers a practical lever to monetize AI crawling without requiring individual negotiations with each AI company — accelerating the emerging ‘pay to train’ licensing market. Whether this specific enforcement mechanism holds or not, the direction is clear: within two years, AI training data will carry a price tag.
If you manage a content-heavy website, investigate Cloudflare’s AI bot controls this week — this is a live mechanism to block or monetize AI crawler access to your content right now.
Google’s Spam Detection Now Targets Account Networks
Google researchers published a methodology that detects AI-generated video spam by clustering coordinated account patterns rather than analyzing individual content pieces — flagging the network, not the video. This signals that Google’s anti-spam infrastructure is making the same shift Google made with link spam detection a decade ago: from content analysis to behavioral network analysis, a move that permanently changed the SEO landscape. High-volume AI content strategies across multiple accounts now face systemic account-level risk, not just individual post penalties.
If you or a client runs scaled AI content across multiple accounts or properties, consolidate under a single authoritative domain with genuine editorial differentiation before Google’s network-level detection matures further.
GPT-5.6 Arrives as Three-Model Suite: Sol, Terra, Luna
OpenAI is preparing GPT-5.6 as a suite of three distinctly named models — Sol, Terra, and Luna — mirroring Anthropic’s Haiku/Sonnet/Opus tiered strategy and confirming that multi-tier model suites are now the permanent standard across all major AI providers. By giving each tier a name rather than a number, OpenAI is creating distinct products that are easier to price differently and harder for users to compare on raw capability-to-price ratios. Practitioners now need model routing strategy — deciding which tier handles which task — as a core competency, not an optional optimization.
When GPT-5.6 launches, immediately map Sol, Terra, and Luna to your existing workflows rather than defaulting to the top model — cost-efficient task routing across tiers is where the real productivity ROI lives.
OpenAI’s 5% Equity Offer: Government Alignment as Strategy
OpenAI CEO Sam Altman has proposed donating 5% of OpenAI’s equity to a US sovereign wealth fund, framing AI wealth distribution as national infrastructure policy — which would give the US government a financial stake in OpenAI’s success rather than purely in its oversight. A government financial interest of this kind could structurally advantage OpenAI in federal procurement and create a precedent that pressures other AI companies to follow. For enterprise buyers, especially in the public sector, this signals both increased stability and potential policy capture risk worth monitoring.
Track this proposal’s legislative progress over the next 90 days — if it advances, it will reshape how enterprise AI contracts in the public sector are structured and evaluated.
Read the full story →
Join the discussion →
More from Rafal Reyzer
For deeper dives on AI and marketing strategy, visit my YouTube channel →
Rafal Reyzer
Hey there, welcome to my blog! I'm a full-time entrepreneur building two companies, a digital marketer, and a content creator with 10+ years of experience. I started RafalReyzer.com to provide you with great tools and strategies you can use to become a proficient digital marketer and achieve freedom through online creativity. My site is a one-stop shop for digital marketers, and content enthusiasts who want to be independent, earn more money, and create beautiful things. Explore my journey here, and don't forget to get in touch if you need help with digital marketing.