1 Adobe stuffed 70+ tools into a single ChatGPT plugin
On August 6, Adobe launched a unified plugin inside ChatGPT that bundles more than 70 tools from Photoshop, Firefly, Express, Acrobat, Lightroom, Illustrator, InDesign and Adobe Stock — replacing the handful of separate connectors it shipped through OpenAI’s Apps SDK late last year. Users type “@Adobe” and the plugin auto-coordinates the right tools behind the scenes; it’s live globally on web and mobile, with Slack and Gemini integrations coming later in August alongside expanded access in Claude and Copilot.
The lesson isn’t about Adobe — it’s about distribution. When your customers live inside an assistant, your product has to become a callable tool in that assistant. “Where does our product show up when the user just describes what they want?” is now a distribution question every marketing software vendor has to answer.
2 Optimove: 72% of US consumers now shop by asking AI
Optimove Insights’ 2026 Holiday Shopping Report, released August 5, found that 72% of US consumers regularly or occasionally consult ChatGPT, Claude or Gemini for shopping ideas and gift recommendations. Product-recommendation AI is used by 55%, chatbots by 41%, and visual search by 31% — and nearly three-quarters trust AI recommendations as much as or more than other sources. Meanwhile, 53% say they’ll buy only from stores they purchased from last year.
Read those two numbers together: discovery is moving into AI answers while loyalty is hardening. If you’re not in the consideration set an AI assembles — or the customer’s last-year purchase history — you may not be in the holiday funnel at all. AI visibility work has a deadline, and it’s Q4.
3 KIBO collapsed nine commerce agents into one
On August 5, KIBO Commerce announced general availability of KIBO AI, a redesigned agentic layer that unifies commerce and order management into one interface, one conversation, one intelligent agent — with a coordinated network of specialized agents operating invisibly behind it. It replaces KIBO’s initial offering of nine purpose-built agents, and is built on a unified data model, a Bring Your Own Model approach that avoids LLM lock-in, and KIBO’s proprietary agentic framework.
Agent sprawl is ending as fast as it started. Twelve months ago vendors bragged about how many agents they shipped; now they brag about hiding them behind one conversation. When you evaluate agentic platforms, the questions that matter are orchestration quality and model portability — not agent count.
4 Bloomreach turned the search bar into the whole storefront
Also on August 5, Bloomreach launched an “Ask Me Anything” capability that makes the site search bar a single entry point for everything: keyword search, conversational shopping questions, and even customer service on orders and returns via third-party service-agent integrations. Its Loomi conversational agent acts as the orchestration layer, reading intent and routing each request to the right experience in real time.
Search, merchandising and support just collapsed into one surface. That breaks a decade of e-commerce org design where those were three teams with three tools and three dashboards. If shoppers stop navigating and start asking, your category pages — and the KPIs built on them — quietly stop describing how your store actually works.
5 PubMatic shipped guardrails for autonomous ad buying
PubMatic launched a customizable governance framework for its AgenticOS on August 5 — a five-step guardrail architecture covering configurable boundaries, inherited business rules, pre-approved asset libraries, one-time-passcode approval workflows, and audit trails with drift detection. Unlike bolt-on governance, it’s embedded at the point of execution. Rise, a Quad agency, is the first pilot partner; AgenticOS has run 80+ autonomous campaigns across 100,000+ properties with 4,000+ direct deals since its January debut.
Autonomy scales exactly as far as accountability lets it. The vendors racing to automate the buy are now racing to prove the automation can be constrained and audited — because that’s what procurement and legal will demand before real budgets move. Ask your DSP and SSP where their guardrails live: inside the execution path, or in a slide deck.
6 The IAB published the rulebook for AI visibility measurement
On August 3, the IAB released “Measuring Visibility in the AI Era,” a framework for measuring brand and publisher presence inside AI-generated answers across ChatGPT, Gemini, Copilot, Perplexity and Claude. It defines the “4 P’s” — presence, prominence, portrayal, persuasion — and two quality tiers, with “decision-grade” measurement (rigor across query volume, sample size, prompt coverage, cadence, reproducibility) required before making budget decisions. Context: 20+ vendors now sell AI-visibility tools with conflicting methodologies, and only 16% of brands track AI visibility at all.
The AEO/GEO gold rush just got a referee. If a vendor’s numbers can’t meet the decision-grade bar, they’re directional at best — so audit the methodology before you reallocate SEO budget based on a dashboard. And if you’re in the 84% not tracking AI visibility, item 2 above is your reason to start.
7 ChatGPT ads got closed-loop call attribution
CallTrackingMetrics announced a native OpenAI Ads integration that closes the attribution loop for ChatGPT advertising — when a ChatGPT ad drives a phone call, CTM captures and attributes the lead, then automatically pushes qualified conversion events back to OpenAI. It’s the same bidirectional data flow performance marketers rely on for Google, Meta and Microsoft campaigns, now extended to an AI channel for the first time.
The performance plumbing around AI channels is being built at remarkable speed — conversion bidding last week, closed-loop call attribution this week. The “it’s experimental, we can’t measure it” excuse for ChatGPT ad spend is expiring. Treat AI channels like any other: demand closed-loop measurement before you scale, and test while auction prices are still early-adopter cheap.
8 PulsePoint launched the first agentic platform for healthcare marketing
On August 4, PulsePoint launched Hatch, billed as the first agentic AI platform built for healthcare marketing. Marketers, planners and traders can ask questions, size audiences, monitor delivery, and act on live campaign data through natural conversation — the platform was designed around conversational AI at the center rather than bolted onto existing workflows.
When the most compliance-constrained category in advertising goes agentic, the technology has crossed a maturity line. Healthcare’s privacy and regulatory requirements make it the hardest place to automate — if conversational campaign management works there, vertical-specific agentic platforms are coming for finance, insurance and pharma next. The DSP interface of 2027 is looking a lot like a chat window.
9 IDC: AI agents are already shopping, and merchants aren’t ready
A new IDC InfoBrief sponsored by WooCommerce found that AI agents shopping on behalf of buyers are already reshaping discovery — and most commerce platforms aren’t prepared. IDC projects AI will replatform $500 billion in digital spending by 2030, with agent use among Global 2000 companies growing 10x by 2027. Brands that aren’t ready stand to lose access to 25% of their market; 80% of agentic use cases will require real-time contextual data access by 2027, and 65% of digital leaders cite legacy platform rigidity as the top barrier.
Agents don’t respond to brand storytelling or emotional creative — they evaluate structured data, verified inventory and pricing accuracy. The unglamorous work of cleaning product data is quietly becoming the highest-ROI marketing project on the roadmap. Audit your feed before your competitors’ agent-readable catalogs eat your discovery share.
10 Optimizely: 75% of consumers say your marketing is irrelevant
Optimizely’s UK research, released August 3, surveyed 1,000 consumers and 100 marketers: 75% of consumers regularly receive irrelevant marketing, 69% get duplicate messages from the same brands, 61% feel overwhelmed by volume, and 42% disengage when content feels irrelevant. On the marketer side, 60% admit launching campaigns without the time or data to optimize them, and 66% say managing multiple tools creates unnecessary work.
This is the uncomfortable counterweight to everything above. AI has been deployed to accelerate content production, not improve relevance — more of the wrong thing, faster. While the industry builds agentic everything, the basic promise of the last decade of martech remains unkept. The brands that win the chat-box era won’t be the ones producing the most; they’ll be the ones whose data infrastructure makes the message actually land.