Get tomorrow's brief in your inbox
Today: Google is embedding Gemini AI across Android with agentic features that shop for you and auto-fill forms, pushing dictation startup competition aside. SAP invested in workflow platform n8n at a $5.2 billion valuation to bring visual AI orchestration to enterprise. Anthropic launched legal automation tools as the AI legal services market heats up past $11 billion in valuations.
Google just announced its biggest Android AI push yet, and it's not about better search results. The company is rebuilding Android around what it calls "Gemini Intelligence," giving the AI assistant the ability to take actions across apps on your behalf. Starting this summer on Galaxy S26 and Pixel phones, you'll be able to show Gemini a photo of your grocery list and have it add those items to your Instacart cart, or ask it to pull event details from a flyer and book tickets across multiple apps.
This matters because Google is turning AI from a chatbot into an operating layer. The assistant can now browse the web for you, fill out forms using information from your photos and emails, and complete multi-step tasks that previously required jumping between five different apps. Every action requires your final approval before completing transactions, addressing the obvious concern about AI agents spending your money without permission.
For small business owners, this signals two things. First, if your customers are on Android (which is most of the world outside the US), expect more of their interactions with your business to happen through AI intermediaries. Your website needs to work for AI agents, not just humans. Second, the tools you build or buy need to integrate with these platforms. Google's bet is that personal AI wins by living inside the OS, not as a separate app. That's why it's launching ahead of Apple's expected AI reboot at WWDC in June.
Sage + AWS bring AI accounting to SMBs Sage announced an expanded partnership with Amazon Web Services to simplify AI adoption for small business accounting. The core offering is Sage AI Developer Solutions built on Amazon Bedrock AgentCore, automating accounts payable, cash flow management, and compliance reporting. The tools will be available in the AWS Marketplace later this year. This addresses a real problem: IDC predicts AI spending will grow 32% annually through 2029, but SMBs struggle with the complexity and cost of modernizing financial systems. The partnership focuses on migrating desktop accounting software to the cloud with minimal disruption. If you're still running QuickBooks Desktop or similar, this is worth watching. Cloud-based AI tools can give you real-time cash flow insights and automate reconciliation work that currently eats hours each week. Source: Small Business Trends
n8n lands $5.2B valuation with SAP investment Workflow automation platform n8n raised funding from SAP at a $5.2 billion valuation, more than doubling its value in less than a year. The bigger news: n8n is being embedded directly into SAP's Joule Studio, giving SAP developers a visual way to build AI workflows across their entire tech stack. n8n supports 1,000+ integrations and lets you orchestrate AI agents with visual branching, error handling, and human-in-the-loop checkpoints. This partnership makes n8n relevant to enterprises that need data sovereignty and sector-specific compliance. If you're not in the SAP ecosystem, n8n still offers a self-hosted option for businesses that want Zapier-style automation without sending data to third parties. The core platform is open source, and you can run it on your own infrastructure. That's useful for handling sensitive customer data or regulated workflows. Source: n8n Blog, n8n Blog
Vapi raises $50M at $500M valuation for AI voice agents AI voice startup Vapi raised $50 million at a $500 million valuation after Amazon Ring selected it over 40 competitors to handle 100% of its inbound customer support calls. The platform now processes 1-5 million calls daily across customer support, lead qualification, appointment scheduling, and outbound sales. Ring chose Vapi because it gave engineers granular control over AI agent behavior without depending on vendor customization. Vapi says it's processing over 1 billion calls to date. The platform offers both a self-serve developer tier (used by over 1 million developers) and enterprise contracts. Pricing wasn't disclosed, but enterprise customers include Intuit, New York Life, and Kavak. If you're spending on call center staff or losing leads because you can't answer the phone, this is worth evaluating. The key question is whether the AI can handle your specific workflows, not just generic support scripts. Source: TechCrunch
Google's Rambler kills dictation startup moat Google launched Rambler, an AI-powered dictation feature built into Gboard (Android's default keyboard). It removes filler words, corrects mid-sentence changes, and supports code-switching between languages. The feature uses Gemini multilingual models and runs with a mix of on-device and cloud processing. It's rolling out to Galaxy S26 and Pixel phones this summer, then expanding to all Android devices. This is bad news for dictation startups like Wispr Flow and Typeless, which have built audiences on iOS and desktop but haven't cracked Android yet. When a platform player ships a feature at the OS level to hundreds of millions of users for free, standalone apps need a very strong reason to exist. Better accuracy, deeper features, or stronger privacy guarantees are the only defensible moats left. For small business owners, this means better voice input is coming to Android without installing anything. If your team is mobile-first, this improves productivity for anyone who types on their phone. Source: TechCrunch
Dessn raises $6M for production-focused design tool Design startup Dessn raised $6 million for a tool that runs your codebase in the cloud so designers can build directly in production. It's not for ground-up ideation like v0 or Lovable. It's for teams with an existing codebase who want to iterate without developer handoff friction. Current customers include Color, Wispr, and Mercury. The tool abstracts away dependencies so you don't need local setup, and it lets designers prompt their way into new designs or use AI-generated toolbars for specific contexts. Pricing starts at $39 per user per month after a free trial. Dessn doesn't integrate with Figma by design because the founders think that pulls teams away from production. If your product team is constantly blocked waiting for developers to implement designs, this could reduce iteration time. The trade-off is that it only works if your codebase architecture is supported. Source: TechCrunch
Anthropic launches legal automation tools Anthropic launched new Claude for Legal features, including plug-ins and MCP connectors for specific areas of law. The tools automate document search and review, case law research, deposition prep, and document drafting across commercial, privacy, corporate, employment, product, and AI governance law. MCP connectors integrate Claude with platforms law firms already use, including Docusign, Box, and Thomson Reuters Westlaw. The tools are available to all paying Claude customers. This comes as competition heats up: AI law startup Harvey raised $200 million at an $11 billion valuation in March, and rival Legora raised $600 million in April. The pitch is that law firms need automation to handle rising caseloads without expanding headcount. The risk is real: dozens of lawyers have been caught submitting AI-generated documents with fake citations, and courts are already overwhelmed with AI-generated legal slop. If you're running a law firm or legal department, these tools can speed up document review and research, but you need a rigorous review process before anything goes to court. Source: TechCrunch
AutoScout24 cuts development cycles from weeks to days with Codex AutoScout24, Europe's largest online car marketplace, deployed OpenAI's Codex across 2,000 employees and cut development timelines from 2-3 weeks to 2-3 days for select projects. About 1,000 builder roles now use Codex integrated directly into their workflows. The company chose Codex after a three-month evaluation, citing measurable improvements in productivity and code quality. Key use cases included automated pull request reviews, large-scale refactoring, technical documentation, and post-incident analysis. Non-technical roles also gained the ability to prototype ideas independently, expanding innovation capacity beyond the engineering team. The company's advice: combine broad AI access (like ChatGPT for all employees) with deep workflow integration (like Codex for builders), prioritize real-world use cases over top-down mandates, and establish cross-functional champions to scale knowledge organically. If you're running a product or development team, this is a roadmap for AI adoption that balances enablement with measurable impact. Source: OpenAI
Medicare's ACCESS program creates swim lanes for AI healthcare Medicare launched ACCESS, a 10-year program that pays for health outcomes instead of clinician time, creating the first payment mechanism for AI-driven medical care. The program covers diabetes, hypertension, chronic kidney disease, obesity, depression, and anxiety. Participating organizations get predictable payments and earn the full amount only when patients meet measurable health goals. This is significant because traditional Medicare has no way to pay for an AI agent that monitors patients between visits, coordinates housing referrals, or ensures medication pickup. ACCESS creates that mechanism for the first time. The first cohort includes 150 participants, ranging from AI doctor startups to virtual nutrition therapy providers to wearable makers like Whoop. Pair Team, a healthcare company focused on patients with unstable housing and food insecurity, deployed a voice AI agent called Flora that handles intake, coordinates referrals, and does check-ins 24/7. The company says one in four hospital visits and one in two ER visits don't happen when a patient is in its care. If you're building healthcare tech or running a medical practice, this program opens a new funding path for AI tools that improve outcomes without requiring more doctor hours. Source: TechCrunch
HBR: Traditional marketing doesn't work on AI shopping agents Harvard Business Review published research showing that traditional marketing tactics don't work on AI shopping agents. This matters because a growing share of shoppers are not human. OpenAI is pushing ChatGPT deeper into product discovery, Google launched a universal commerce protocol (UCP) to let AI agents transact across retailers, and Amazon released tools that let its agents shop other retailers' sites on customers' behalf. If your marketing strategy assumes humans are reading your product descriptions and comparing prices manually, that assumption is breaking. AI agents prioritize structured data, clear specs, and direct answers over brand storytelling and SEO tricks. You need to optimize for machine readability: schema markup, API access, and clear pricing without gimmicks. The agents that shop for your customers don't care about your brand voice. They care about whether your product matches the query and whether the data is easy to parse. Source: Harvard Business Review
Threads tests Meta AI integration like Grok Threads is testing a Meta AI integration in five countries (Malaysia, Saudi Arabia, Mexico, Argentina, Singapore) that works similarly to X's Grok. Users with public accounts can mention Meta AI in a post or reply to get real-time context about trends, breaking stories, or recommendations. The feature responds in the language used in the post. This positions Threads as a place to get information without leaving the app, competing directly with X's Grok integration. The risk: giving an AI chatbot this level of visibility can backfire, as seen when Grok generated posts praising Hitler on X. Meta says it has stronger safeguards than Grok, but this remains to be tested. If you're using Threads for business, expect more AI-generated replies in your mentions. You can mute the @meta.ai account if you don't want to see them. Source: TechCrunch
Alibaba's core profit plunges 84% despite AI growth Alibaba reported an 84% drop in core profitability (adjusted EBITA) in the March quarter, falling to $750.9 million, despite strong growth in AI and cloud. The company is investing heavily in semiconductors for AI, data centers, and its Qwen model family, which paid off in cloud growth. However, adjusted EBITA in its China e-commerce group dropped 40% due to investments in quick commerce (delivery under an hour). Quick commerce revenue grew 57% year-over-year, but investors are grappling with whether the payoff justifies the cost. This shows the tension every business faces with AI investment: growth in AI-driven segments doesn't automatically translate to profitability if you're spending faster than revenue grows. For SMBs, the lesson is to measure AI investments by actual ROI (time saved, revenue increased, costs reduced), not by strategic buzzwords. Source: CNBC
Create custom Android widgets with natural language Google's "Create My Widget" feature lets you describe what you want and generates a custom widget for your home screen. Example prompts: "Suggest three high-protein meal prep recipes every week" or "Show wind speed and precipitation for cycling." The feature launches this summer on Galaxy S26 and Pixel phones first, then expands to other Android devices and Wear OS watches. If your business needs custom dashboards for specific tasks (sales metrics, inventory levels, appointment reminders), this is a free way to prototype without hiring a developer. The widgets follow Google's Material 3 design language, so they'll look native. Source: TechCrunch
Use Codex to clean financial models before reviews OpenAI published a guide showing how finance teams use Codex to review workbook structure, formulas, hardcodes, broken links, circular references, and output tabs before high-stakes presentations. Codex can make safe cleanup changes automatically and flag assumptions that require human review. It returns a cleaned workbook plus a severity-ranked QA memo. If your team builds financial models in Excel or Google Sheets, this can catch errors before they reach leadership or clients. The key is telling Codex which tabs matter most and which business assumptions should not change without review. Suggested plug-ins: Spreadsheets, Google Drive, SharePoint, Box. Source: OpenAI
Google's Android Show previewed the future: AI that lives inside your OS, not as a separate app. The race is on to control the layer where users interact with AI, whether that's Android, iOS, or your desktop. For small businesses, this means two things. First, your customers' interactions with your business will increasingly happen through AI intermediaries. Second, the tools you use daily are being rebuilt around AI agents. Expect more automation, more voice interfaces, and more features that require you to trust the AI to act on your behalf. The winners will be businesses that adapt their workflows to work with AI, not against it.