Get tomorrow's brief in your inbox
Today: Salesforce's AI-powered Agentforce IT Service has signed 180+ organizations in four months, promising 24/7 support without ticket portals. OpenAI's robotics chief quit over the Pentagon deal, while ChatGPT uninstalls surged 295% and Claude climbed to #1 in app stores. Anthropic and OpenAI are now offering six months free to open source maintainers in a battle for developer mindshare.
If you run a small business, IT support has always been a choice between expensive and inadequate. Salesforce just launched Agentforce IT Service, and 180+ organizations signed up in four months because it does something genuinely useful: it replaces the "submit a ticket and wait" model with AI agents that resolve issues immediately, 24/7, right inside Slack or Microsoft Teams.
This matters for small businesses because you don't need a full IT team anymore. Password resets, software access requests, routine troubleshooting - all handled by autonomous agents while your people focus on actual work. CoolSys, a heating and cooling company, is deploying it specifically to "free teams to focus on higher-value work," according to their CIO. Sunrun's VP of Enterprise Technology says consolidating on the Salesforce platform delivers "faster, simpler, and more scalable service" with actual business value, not just cost savings.
The key advantage: you can go from purchase to production in weeks instead of months. The conversational CMDB is built into Salesforce's platform, so there's no painful legacy system integration. If you're already using Salesforce for CRM, this is a natural extension. If you're not, consider whether the unified platform approach justifies the investment compared to duct-taping together multiple point solutions that never quite work together.
Google released Nano Banana 2, an upgrade to its text rendering tech that automatically translates and localizes text inside images. If you run a local business targeting tourists or sell products internationally, this is practical: the demo app (Global Ad Localizer) takes one advertisement and generates versions in multiple languages with culturally adapted visuals, not just translated words.
For a café creating multilingual menus or an e-commerce site adapting product descriptions, this removes the need for expensive translation services and separate design work per language. The downside: you still need quality checks. AI-generated translations miss nuances, and visual adaptations might not match your brand. Treat this as a first draft tool, not a replacement for human review when brand consistency matters.
Anthropic and OpenAI are both offering six months of free access ($200/month value) to maintainers of popular open source projects. Anthropic wants 5,000+ GitHub stars or 1M+ NPM downloads. OpenAI is less specific but asks for "GitHub stars, monthly downloads, or why the project is important to the ecosystem."
If you maintain or contribute to open source infrastructure your business depends on, this is free access to top-tier AI coding assistants. Apply for both. Use Claude Max for thoughtful architectural work and complex refactoring. Use ChatGPT Pro with Codex for rapid prototyping and security scanning. The six-month window is long enough to prove ROI before deciding whether to pay.
OpenAI's robotics lead, Caitlin Kalinowski (previously ran Meta's AR glasses team), quit over the Pentagon contract, saying "surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got." This came after Anthropic refused a similar deal and got labeled a "supply chain risk" by the Defense Department.
For small business owners, this matters because consumer trust is currency. ChatGPT uninstalls jumped 295% after the Pentagon announcement, and Claude is now the #1 free app in the U.S. App Store. If you're building products with AI, your vendor's reputation affects your customers' perception of your business. The technical capabilities matter less than whether your customers believe you're using AI responsibly.
OpenAI says they have "red lines" around domestic surveillance and autonomous weapons, but Kalinowski's point was about governance: "the announcement was rushed without the guardrails defined." This is a governance problem, not a technology problem. If your business uses AI tools in customer-facing ways, be prepared to explain which vendors you use and why.
Grammarly's Expert Review feature claims to give writing feedback "from the perspective" of famous authors and tech journalists, but none of these people are involved or gave permission. Historian C.E. Aubin told Wired: "These are not expert reviews, because there are no 'experts' involved in producing them."
If you use Grammarly (and many small businesses do for client communications), understand what this feature actually does: it analyzes publicly available writing by these people and mimics their style. It's pattern matching, not expert advice. For routine business writing, that's fine. For anything where credibility matters (proposals, thought leadership, client-facing content), don't claim expertise you don't have just because an AI tool suggested phrasing "like" a well-known writer.
OpenAI postponed the launch of "adult mode" (age-gated access to erotica and adult content) for the second time, originally promised in December, then pushed to Q1 2026, now delayed indefinitely. The company says they're focusing on "work that is a higher priority for more users right now," like making ChatGPT "more proactive."
For business users, this signals where OpenAI is prioritizing development resources: core intelligence and personality improvements over niche features. If you're paying $200/month for ChatGPT Pro, expect continued investment in the features that affect most users (reasoning, coding, multimodal capabilities) rather than edge cases.
Free AI for your open source projects: If you maintain any open source code with decent traction, apply for both Anthropic's Claude Max and OpenAI's ChatGPT Pro free access programs. Even if you're not a core maintainer, if you contribute regularly, it's worth asking. Worst case: they say no. Best case: $2,400 in free AI tools over six months.
Audit your AI vendor exposure: Make a list of which AI tools you use in customer-facing products or services. Check recent news for each vendor. If any of them are in controversy (Pentagon deals, data privacy issues, leadership departures), prepare a one-paragraph explanation for why you chose them and what safeguards you have in place. Your customers will ask eventually.
This week exposed the gap between AI capabilities and AI governance. The Anthropic-Pentagon standoff and OpenAI's rushed deal reveal that even the leading AI companies don't have consensus on red lines, much less enforcement mechanisms. For small businesses, the practical takeaway is simple: technical features matter less than vendor stability and public trust. When choosing AI tools, ask not just "what can it do?" but "will my customers trust that I'm using this responsibly?" The companies that get governance right will win the long game, even if they ship features slower.
Watch for: more public pressure on AI safety frameworks (the Pro-Human Declaration mentioned this week is one example), continued volatility in consumer sentiment toward major AI vendors, and potential regulatory action that could reshape which AI tools are viable for business use.