1 Alphabet’s ads hit $81.6B — AI didn’t kill search after all
Alphabet reported Q2 2026 revenue of $119.8 billion, up 24% year over year and well past the $116.9 billion analysts expected. Google’s advertising business — Search, YouTube, and Network combined — rose to $81.6 billion, with Search growing 17% in the middle of the AI Overviews era and YouTube ads climbing 13% to $11.1 billion, paced by live sports. Cloud surged 82%.
The “AI will cannibalize search ads” thesis has now missed for several consecutive quarters. Google changed the interface and kept the monetization — which means budget plans built on the assumption that search spend is a melting ice cube need a rewrite. The risk to watch isn’t Google’s revenue line; it’s your click volume, which is a different number entirely.
2 ChatGPT’s UK ad auction swung 278% — and nobody could say why
As ChatGPT’s UK ad auction opened and new advertisers rushed in, brands watched CPCs swing between 39% and 278% with no explanation and no visibility into who had entered the auction against them — OpenAI’s Ads Manager reports only your own numbers. In response, search intelligence firm Adthena and dentsu launched a UK partnership around a new product, Decision Intelligence, that maps which competitors entered specific prompt categories, when, and what it did to prices. Dentsu is one of OpenAI’s four UK agency launch partners, with live campaigns for brands including Hilton, and three dentsu clients already rank among the UK’s top nine ChatGPT advertisers.
This is the most reliable pattern in ad platform history repeating on schedule: a new auction launches with zero competitive intel, and within weeks a third-party visibility layer appears — exactly how the SEO and PPC tool industries were born around early Google. Conversational ad auctions are now real markets, volatile enough that visibility itself is a product. If you’re buying ChatGPT ads, assume your competitors can see you before you can see them.
3 HubSpot shipped Agent Hub — one console for agent sprawl
HubSpot launched Agent Hub and Agent Builder in public beta on July 23 for all Professional and Enterprise customers — a single place to build, monitor, and audit AI agents that share customer context. The launch targets a failure mode most teams already recognize: a prospecting agent emailing an account the same week a service agent is handling that account’s open complaint, with neither aware of the other. Early customer Ignite Reading built an agent that parses school-district calendars and expects to reclaim roughly 350 hours a year.
The interesting shift is that the product isn’t the agent — it’s the coordination. Every vendor has shipped agents; almost nobody has shipped the layer that keeps them from stepping on each other. Capability stopped being the bottleneck months ago. If your stack has more than two agents touching customers, the audit trail is now the feature to buy for.
4 Meta is running brand ads for AI itself
Meta released “The Future Is for Everyone,” a 45-second film that opens on black-and-white doomsday AI headlines and shifts to color scenes of people connecting across WhatsApp, Instagram, and Messenger. It was produced in-house, narrated by a Meta employee, announced by Mark Zuckerberg on July 23 — and set, as TechCrunch noted, to David Bowie’s “Five Years,” a song about the end of the world. It’s the first installment of a broader paid and earned campaign.
The company that sells more AI-powered ads than anyone is now spending media dollars to sell AI itself — which tells you trust, not capability, is the adoption constraint Meta’s own data is surfacing. For marketers the lesson is in the backlash mechanics: audiences read the Bowie subtext within hours. When you advertise optimism, the internet fact-checks the soundtrack.
5 The creator economy grew a middle class
Bloomberg reported July 24 that a growing tier of microinfluencers is earning mid-level salaries from small audiences — stacking affiliate links, brand deals, and community revenue rather than chasing viral reach. Young workers increasingly treat creator work as a plausible career pivot, even as softening consumer spend makes brand deals less predictable and pushes creators toward revenue diversification.
Most media plans still over-index on mega-creators, but the durable inventory is thousands of small, trusted audiences — the same math driving this week’s industry debate about whether it’s smarter to reach a million people once or a thousand people who already trust the messenger. For brands entering the US market, the creator middle class is the affordable, high-trust way in — and it’s getting more professional every quarter.
6 Affinity’s new agents live inside Claude and ChatGPT
CRM company Affinity unveiled Ascend, an agent platform that lets investment firms automate workflows across the deal lifecycle — with agents deployable not just from Affinity’s own interface but directly from Claude, ChatGPT, and custom AI environments, all running on the same data foundation.
Read past the private-capital niche and this is a distribution story: vertical software is starting to ship agents that live wherever the user already chats, instead of forcing users into the app. “Bring the agent to the interface” is quietly becoming the default playbook — expect marketing platforms to follow, and expect “works inside your AI assistant” to become a standard line on martech pricing pages.
7 Observe.AI signed a multi-year AWS pact — the cloud is the new sales channel
Agentic CX platform Observe.AI announced a multi-year strategic collaboration agreement with AWS to help enterprises deploy AI agents across customer service operations at scale — resolving customer needs directly, supporting frontline teams in real time, and improving performance across conversations. The platform is also now available on AWS Marketplace.
The pattern to note is procurement: agentic software is increasingly bought through the cloud bill, not a separate vendor negotiation. For martech and CX vendors, hyperscaler marketplaces are becoming the distribution channel that app stores were for mobile — and for buyers, committed cloud spend just became leverage for AI tooling you were going to buy anyway.
8 Algolia’s filters now build themselves
Algolia launched Dynamic Facets, a capability that automatically surfaces the most relevant search filters based on what shoppers are actively looking for — adapting in real time to query context and behavior instead of showing the same fixed refinement options for every search, and cutting the manual configuration work merchandising teams normally carry.
Product discovery is quietly becoming self-assembling. The merchandiser’s job shifts from configuring the shelf to auditing what the machine assembled — the same configure-to-audit shift already underway in media buying and creative. If your e-commerce search still runs on hand-built filter trees, this is the direction of travel.
9 SurveyMonkey relaunched GetFeedback — VoC in one line of code
SurveyMonkey reintroduced GetFeedback as a simplified self-serve product that lets marketing, product, and CX teams collect feedback directly on their websites — sign up, choose a plan, create a form or button, and add it with a single line of code. The pitch: capture input at the exact moment visitors hesitate, hit friction, or abandon.
The interesting part isn’t the survey — it’s the timing. Feedback captured at the moment of friction beats a quarterly NPS email by miles, and the self-serve packaging drops voice-of-customer tooling to SMB price and effort levels. For teams optimizing conversion, this is the cheapest new data source of the week.
10 CallRail wired AI phone agents into the trades
CallRail deepened its partnership with field-service platform Jobber, shipping a bidirectional integration between Jobber and Voice Assist — CallRail’s AI voice agent that qualifies leads in real time and transfers intelligently — alongside three new integrations with JobNimbus, JobTread, and Workiz for contractors and home-services businesses.
This is what the agentic era actually looks like on Main Street: not a demo, but an AI that answers the phone for a roofing company, qualifies the lead, and books the job into the field-management system. The unglamorous vertical plumbing is where agent adoption is compounding fastest — and where the ROI case is easiest to make.
The throughline: the money in AI advertising is now indisputable — Alphabet’s $81.6 billion quarter settled that. What’s missing is visibility: into who’s bidding against you in a ChatGPT auction, into what your own agents are doing to the same customer, into what the machine-assembled shelf is showing shoppers. This week, that gap became the product — Adthena, HubSpot, and a dozen others are building the dashboards the AI stack outgrew. Buy capability second. Buy visibility first.