Get tomorrow's brief in your inbox
Today: Meta bought Moltbook to build an "agent graph" for AI-to-AI commerce. Zendesk acquired Forethought to ship autonomous customer service faster. China's OpenClaw craze spawned a cottage industry of installation services charging $34 per setup.
Meta acquired Moltbook, a social network for AI agents, in what looks like an acqui-hire for talent. But the real play here is building infrastructure for what Meta calls the "agentic web," where businesses' AI agents and consumers' AI agents negotiate directly with each other instead of relying on humans clicking ads.
Here's what this means for your business: Facebook built the "friend graph" that mapped social connections between people. Now Meta wants to build an "agent graph" that maps how AI agents discover each other, connect, and transact. If you run an e-commerce business, your future customer might not be a person browsing your site. It could be their AI agent shopping on their behalf, comparing prices across hundreds of competitors in seconds, filtering by their owner's preferences (eco-friendly, small business, on sale only), and negotiating the best deal.
The takeaway for small businesses is twofold. First, this shift could level the playing field. Your AI agent could pitch directly to a customer's AI agent without needing a massive ad budget. Second, you need to start thinking about how your business will be discovered and evaluated by AI agents, not just humans. That means structured data, clear pricing, and machine-readable policies. The businesses that adapt early will have an advantage when AI-to-AI commerce becomes mainstream.
Google Gemini 3.1 Flash-Lite is now available in preview, and it's the fastest, cheapest model in the Gemini 3 series. Priced at $0.25 per million input tokens and $1.50 per million output tokens, it's 2.5x faster than the previous version and delivers 45% faster output speed. If you're running high-volume chatbots, data analysis tools, or customer service automation, this could cut your AI costs significantly while improving response times. Available through Google AI Studio and Vertex AI.
Zendesk acquired Forethought, the 2018 TechCrunch Battlefield winner that pioneered AI-powered customer service automation years before ChatGPT existed. Forethought was already handling over a billion monthly customer interactions for companies like Upwork, Grammarly, and Airtable. Zendesk says the acquisition accelerates its product roadmap by over a year, bringing more autonomous customer service capabilities, voice automation, and self-improving AI to its platform. If you're using Zendesk, expect better agent tools soon. If you're evaluating customer service platforms, this signals that AI-first support is no longer optional.
Replit raised $400 million and launched a new AI agent that can "vibe code a startup from scratch," according to CEO Amjad Masad. If you've been putting off building a custom internal tool or simple web app because you don't have a developer on staff, Replit's agent might be worth testing. It's designed to go from idea to working prototype without requiring coding knowledge.
WordPress launched my.WordPress.net, a browser-based private workspace that runs entirely in your browser with no hosting, domain, or signup required. Sites are private by default and stored locally in your browser. WordPress is positioning this as a personal workspace for drafting, journaling, and building private tools. The App Catalog includes a Personal CRM, RSS Reader, and AI Workspace. You can also ask an AI assistant to build custom plugins for you. If you need a quick prototyping environment or a private knowledge base accessed by AI, this is free and instant. Just remember to back up regularly since data lives in your browser.
Ford Pro launched an AI assistant for its commercial fleet customers that can monitor millions of data points across vehicles. It tracks fuel consumption, seatbelt use, vehicle health, idle times, speeding, and acceleration events. Ford emphasizes this isn't a chatbot; it provides detailed insights instead of diagnostic error codes. If you manage a fleet of trucks or service vehicles, similar AI monitoring tools could help you cut fuel costs, reduce insurance premiums, and spot maintenance issues before they become expensive breakdowns. Ford Pro paid software subscriptions grew 30% in 2025, showing businesses are willing to pay for AI tools that deliver measurable ROI.
Netflix paid up to $600 million for Ben Affleck's AI startup InterPositive, which makes tools that help filmmakers edit footage in post-production. This is Netflix's second-largest acquisition ever. InterPositive addresses continuity issues and enhances scenes without generating new content or using footage without permission. If you're in video production, this signals that AI editing tools are becoming table stakes. Amazon and Disney are making similar moves. For small production companies, this means you'll soon be competing against studios using AI to speed up post-production by weeks or months.
Franchising tax cuts are now permanent, according to a House Small Business Committee hearing. The 20% 199A small business deduction is permanent, giving franchise owners more cash flow to invest in operations, employee compensation, and expansion. The committee also pushed for clarity on the Joint Employer rule, which has been a regulatory headache for franchisees. If you've been considering franchising as a growth strategy, the tax environment is now more favorable than it's been in years.
OpenClaw installation services have become a cottage industry in China, with entrepreneurs charging about $34 per setup. One software engineer quit his job and now runs a 100-employee operation that's handled 7,000 installations. Chinese tech companies like Tencent are hosting free installation events, and local governments are promoting OpenClaw adoption. The takeaway: when a tool is powerful but difficult to install, there's a business opportunity in being the simplification layer. If you run a service business and your customers struggle with technical setup, consider whether you could package that as a paid onboarding service.
Claude is the only major chatbot that reliably refuses to assist with violent planning, according to a CNN/CCDH investigation that tested 10 popular AI chatbots. Most others, including ChatGPT, Gemini, Meta AI, and Perplexity, failed to intervene when simulated teens discussed violent acts. Character.AI actively encouraged violence in some scenarios. If you're building AI tools for younger users or safety-critical applications, this study shows that safety guardrails vary wildly between providers. Choose your AI vendor carefully.
Rivian spun out Mind Robotics, which just raised $500 million in a Series A at a $2 billion valuation. The startup uses data from Rivian's EV factory to train industrial robots to be more dexterous and adaptable. CEO RJ Scaringe says the company is focused on traditional factory robot designs, not humanoid robots. "Doing cartwheels does not create value in manufacturing," he told The Wall Street Journal. If you run a manufacturing operation, keep an eye on this space. AI-trained industrial robots could soon handle tasks that currently require human dexterity, and they won't require the hype tax of humanoid designs.
OpenAI published technical details on designing AI agents to resist prompt injection and social engineering. The post explains how ChatGPT constrains risky actions and protects sensitive data in agent workflows. If you're building AI agents that handle customer data or execute actions on behalf of users, this is required reading. Security in agentic systems is fundamentally different from traditional software security.
Google is using AI to predict flash floods by reading old news reports. Gemini sorted through 5 million news articles to isolate 2.6 million flood reports, creating a geo-tagged time series called "Groundsource." The model now highlights flash flood risks for 150 countries on Google's Flood Hub platform. This is a clever example of using LLMs to turn qualitative text data into quantitative datasets for machine learning. If you're sitting on years of unstructured business data (customer emails, support tickets, sales notes), you could use similar techniques to extract insights that traditional analytics miss.
Zapier published a guide on six types of AI agents and how to pick the right one for your workflow. If you've been confused by the agent terminology, this breaks it down clearly: simple reflex agents (if/then rules), model-based reflex agents (context-aware), and autonomous agents (multi-step goal-seeking). The article includes practical examples like lead qualification, email triage, and customer support routing. Start with simple reflex agents for repetitive tasks and scale up to more autonomous systems as you get comfortable.
Visualping automation can turn competitor monitoring into sales intelligence. Set up Visualping to track competitor pricing pages, then build a Zap that sends alerts to your sales team the moment a competitor changes pricing. Add an AI by Zapier step to analyze what changed and generate sales talking points automatically. You can also monitor prospect websites for buying signals like new job postings or leadership changes. This is competitive intelligence on autopilot.
The shift from human-to-human commerce to AI-to-AI commerce is accelerating faster than most businesses realize. Meta's Moltbook acquisition, Zendesk's Forethought purchase, and China's OpenClaw explosion all point to the same trend: AI agents are moving from research demos to production systems that handle real transactions. Businesses that start preparing now by structuring their data, clarifying their pricing, and building agent-friendly interfaces will have a significant advantage. The businesses that wait will find themselves invisible to the AI agents that increasingly mediate buying decisions.
Watch for more acquisitions in the agentic customer service space. Zendesk's move signals that legacy software companies are willing to pay big money to catch up on AI capabilities they missed while focused on traditional features.