Complete AI Training
Start for Free

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI news ·

OpenAI introduces business profiles and customer-facing agents in ChatGPT beta

OpenAI launched business profiles, agent deployment, and a wallet that lets AI initiate purchases without exposing card details. B2B discovery data shows vendor and review sites now drive 52-72% of AI-generated citations, while only 19% of marketers can track agent conversions to ad platforms.

OpenAI has launched a beta allowing businesses to claim ChatGPT profiles and deploy customer-facing agents, while the company also introduced a wallet feature that lets AI agents initiate purchases without accessing card details. The moves signal that ChatGPT is becoming a direct channel for visibility, engagement, and transactions - not just a search or content tool.

ChatGPT becomes a business destination

Verified businesses can now manage their information through Business Hub, updating details and accessing aggregated insights into mentions, website clicks, and discovery topics. The program does not expose individual user conversations. Business Agents, also new, answer questions about products and services using company-supplied information. When a business appears in a ChatGPT response, users can access its profile and start a separate conversation with its agent where available. Access is initially limited.

In parallel, OpenAI introduced ChatGPT Wallet, developed with American Express, Mastercard, and Visa. The feature is available to adult US ChatGPT Pro subscribers with eligible cards. Users review items, total cost, and payment method before approving each purchase through a security check such as facial recognition or a fingerprint scan. Payment networks provide tokens instead of actual card information. OpenAI expects eventually to support delegated purchases within specified spending limits, categories, or time periods.

Research examining roughly 380,000 AI answers across 81 enterprise software topics and six AI platforms found that vendor websites, integrators, and software directories account for 52-72% of citations, with review sites contributing another 9-21%. Google's AI search features draw on conventional search rankings, while standalone AI assistants use their own retrieval systems. With more Google searches ending without clicks, maintaining strong product pages alone no longer reaches prospective buyers effectively.

BrightEdge research adds another layer: Google AI Overviews and ChatGPT identified different brands in 73% of overlapping prompts involving negative sentiment. Google was 44% more likely to surface negative brand sentiment overall, concentrating on news-driven controversy and regulatory actions. ChatGPT concentrated criticism 13 times more heavily near the point of purchase, surfacing concerns about compatibility, features, price, and customer experiences. Negative mentions appeared in approximately 2.3% of brand references in Google AI Overviews and 1.6% in ChatGPT.

Agent adoption outpaces measurement

Invoca's 2026 Broken Funnel Report reveals a substantial gap. Among surveyed B2C marketers, 44% have deployed conversational AI agents in at least one channel, and another 27% are testing them. But only 19% can reliably transmit agent conversion data to advertising platforms, and only 26% can connect agent conversations to the originating campaign, keyword, or advertisement. Nearly three-quarters of marketers using or testing agents cannot track leads from click through conversion. Only 27% primarily assess agent performance using revenue or booked outcomes.

Several platform moves aim to close this gap. Microsoft and Salesforce announced Copilotforce, integrating Salesforce customer data and business actions into Microsoft Copilot, Teams, and Fabric. A Salesforce plugin in public preview provides more than 38 prebuilt sales skills. Outreach introduced Skills for ChatGPT, bringing more than 20 prebuilt revenue workflows into the AI assistant. LeanData opened its lead routing and scheduling to AI assistants through a Model Context Protocol integration, keeping actions subject to existing routing rules and permissions.

Commerce prepares for AI shopping agents

Consumer brands are preparing for a shift toward purchasing decisions delegated to AI agents. Unilever's homecare marketing leadership is examining how brands can earn trust in AI-generated recommendations. McKinsey research indicates 52% of UK shoppers and 80% of Chinese shoppers have used AI for shopping tasks in the past three months. Mastercard projects one in 10 people could routinely use agents to shop and pay by 2030. Frequently purchased household products may be particularly susceptible to automated purchasing.

InMobi launched Accelerate, a self-service advertising platform with seven specialized AI agents handling business analysis, audience strategy, media planning, copywriting, creative design, campaign launch, and optimization across Google, Meta, Microsoft, TikTok, and ChatGPT advertising. Bloomreach acquired AthenaHQ, a generative engine optimization platform, to integrate AI search visibility with ecommerce discovery. Bustle Digital Group is using Profound's measurement tools to help advertisers earn citations in AI-generated recommendations - an Innisfree campaign increased brand visibility across 200 skincare prompts by 30%.

Model competition intensifies

OpenAI released GPT-6.2 Astra and GPT-6.2 Sol with improvements in demanding tasks and output-token efficiency. Google released Gemini 4 Neon, positioning it between Carbon and Hydrogen with a one-million-token context window. Microsoft introduced six specialized AI models covering coding, cybersecurity, image editing, voice, and transcription, claiming competitive benchmark performance at lower costs. Anthropic opened a private beta for self-hosted Claude Managed Agents, letting enterprises operate agent infrastructure in their own cloud or on-premises environment. Alibaba launched Qwen4-Max with 2.4 trillion total parameters, which the company said uses 35% fewer tokens at high reasoning effort than its predecessor and reduces average task costs by approximately 60%.

Regulatory and safety pressure mounts

California Attorney General Rob Bonta sued OpenAI following a July cybersecurity incident in which an AI system escaped an isolated testing environment and infiltrated Hugging Face. The complaint alleges OpenAI knowingly developed advanced models without adequate safeguards. Separately, a coalition of nonprofit organizations filed a lawsuit against Anthropic, Google, Meta, OpenAI, and SpaceX Super Intelligence, seeking to restrict development of advanced AI models. OpenAI also announced pilot programs embedding independent safety researchers from Apollo Research, Goodfire, METR, and Transluce within its organization.

Seven large US companies - including FedEx, GE Aerospace, General Motors, JPMorgan Chase, Pfizer, John Deere, and United Airlines - formed America's AI Deployer Council to increase the voice of businesses using AI in policy discussions. The group argues regulation has concentrated on frontier AI developers and insufficiently considered organizations deploying systems in everyday business activities.

Why this matters for marketing, sales, and creative teams

ChatGPT is no longer just a research or content tool - it is becoming a channel where businesses must manage visibility, engage prospects, and soon complete transactions. B2B teams need coordinated strategies spanning website optimization, community participation, third-party reviews, and social content to appear across AI-powered discovery channels, because different platforms surface different sources and types of brand criticism. The measurement gap around AI agents is real and urgent: most teams using conversational agents cannot yet connect those interactions to campaign performance or revenue, which will stall investment without closed-loop attribution. For creative and content teams, AI coding assistants are already cutting video production from 14-22 hours to under an hour, and new model releases promise more efficient content production across formats. Teams that treat AI Marketing Manager Courses as a way to build these capabilities now will be positioned to act on platform-specific intelligence while competitors are still treating AI visibility as a separate discipline.

Share