Get tomorrow's brief in your inbox
Today: Canva AI 2.0 transforms design workflows with agentic tools that call functions and create multi-layer edits from conversation. Anthropic released Claude Opus 4.7 with better coding ability but intentionally reduced cyber capabilities compared to the restricted Mythos model. Google's Gemini now pulls from your Photos library to generate personalized images with Nano Banana.
Canva and Adobe are racing to own the prompt-first design era, and your workflow is the prize. This week both platforms announced major AI overhauls that replace tool-clicking with conversation. Canva AI 2.0 launched with Planning Mode, an agentic assistant that analyzes your project, asks clarifying questions, and calls the right tools to build editable designs with layers. Adobe followed with Firefly AI, which does the same across its app suite. The stakes are high. Canva's 265 million users generate billions in revenue from teams that "start and end their day in Canva," per co-founder Cliff Obrecht. Now those users can describe a multi-channel campaign and get drafts ready to refine and ship without opening individual tools.
If you run a small business or solo creative shop, this changes how you budget time and money. Canva AI 2.0 includes memory features that learn your brand style and apply it automatically. It connects to Slack, Gmail, Google Drive, Calendar, and Zoom to pull context from your communications and files. You can schedule repeatable tasks to run in the background. The free tier exists, but serious features will likely push you toward paid plans. Canva says its image generation is now 5x faster and 30x cheaper, and video is 7x faster and 17x cheaper. That matters if you're producing content at scale. The research preview launches this week to the first million users, with broader rollout coming soon. PayPal's integration with Canva also went live this week, embedding payment links and QR codes directly into designs so you can sell without leaving the platform.
The shift to conversational design tools means less time learning software and more time iterating on output. If your team still clicks through menus to produce marketing assets, expect that workflow to feel outdated within months.
Canva AI 2.0 - Planning Mode orchestrates Canva's tool suite from a single chat interface. Describe what you need, the assistant asks clarifying questions, generates editable designs with layers, and applies your brand style automatically. New integrations with Slack, Gmail, Google Drive, Calendar, and Zoom let the AI pull context from your workspace. Scheduling feature lets you automate repeatable tasks. Performance improvements: image generation 5x faster and 30x cheaper, video 7x faster and 17x cheaper. Research preview launches to first million users this week. Free tier exists but premium features require paid plans. Read more
Google Gemini + Photos integration - Gemini now connects to your Google Photos library to generate personalized images with Nano Banana. If you've labeled people in Photos, you can prompt "create a claymation image of me and my family enjoying our favorite activity" and Gemini auto-selects reference photos to generate custom images. Paid subscribers get access in the next few days. Gemini doesn't train directly on your Photos library but uses limited prompt and response data to improve functionality. Details here
Roblox Assistant upgrades - Roblox revamped its AI assistant with Planning Mode that analyzes code, asks clarifying questions, and creates editable action plans before executing changes. New Mesh Generation tool creates fully textured 3D objects directly in the game world, eliminating placeholder assets. Procedural Model Generation lets you create editable 3D models with code. The assistant playtests games automatically, captures screenshots, reads logs, and identifies bugs to fix them in a self-correcting loop. If you're building games on Roblox or considering the platform for a business project, these tools cut development time significantly.
Anthropic Claude Opus 4.7 - New model launched with improvements in coding, instruction-following, and creativity for documents and slides. Better than Opus 4.6 across most benchmarks but intentionally worse than the restricted Mythos Preview model on cyber capabilities. Anthropic is testing new cyber safeguards on this model before potentially releasing Mythos-class models broadly. Security professionals can apply to a Cyber Verification Program to access full capabilities. Pricing unchanged: $5 per million input tokens, $25 per million output tokens. Available across Claude products, API, and cloud providers (Microsoft, Google, Amazon). More details
OpenAI Codex update - Codex app for macOS and Windows now includes computer use, in-app browsing, image generation, memory, and plugins. Designed to accelerate developer workflows. If you're writing code daily, this update positions Codex as a more complete development environment beyond just code generation.
Physical Intelligence π0.7 - New robotics model demonstrates compositional generalization, combining skills learned in different contexts to solve problems it never trained on. The model coached an air fryer task with only two loosely related training examples, synthesizing fragments into a functional understanding. With step-by-step verbal instructions, success rate jumped from 5% to 95%. Not ready for autonomous multi-step tasks yet, but shows early signs of a general-purpose robot brain. Matters for businesses considering warehouse automation or manufacturing robotics in the next 2-3 years.
AI augmentation beats automation for long-term growth. Harvard Business Review research shows companies choosing AI augmentation over headcount reduction and pure automation grow top-line revenue faster. The fork in the road: use AI to cut costs by replacing workers, or use AI to amplify what your team can do and expand capability. SMBs that implement AI to help employees do more valuable work, not eliminate jobs, see better customer outcomes and innovation. If you're planning AI adoption, frame it as "what can my team do now that they couldn't before" rather than "who can I replace."
PayPal + Canva integration monetizes designs instantly. PayPal Payment Links now embed in Canva, letting you turn any visual into a revenue stream without leaving the design tool. Generate payment links or QR codes, create a PayPal-hosted payment page with product images and pricing, accept payments in multiple currencies across 200 markets. Available globally through Canva Marketplace. If you sell digital products, coaching, or services and create your own marketing materials, you can now design and sell in one workflow. Learning curve exists for those unfamiliar with digital tools, and you'll need to evaluate if this fits your broader sales strategy beyond social commerce.
Canva AI 2.0 performance gains - Image generation 5x faster and 30x cheaper, video 7x faster and 17x cheaper. Research preview launching to first million users this week, broader rollout in coming weeks. Free tier exists but premium features require paid plans.
Google Gemini Photos integration - Rolling out to paid subscribers over the next few days. Requires opting into Personal Intelligence feature to connect Google apps.
Anthropic Claude Opus 4.7 - Same pricing as Opus 4.6: $5 per million input tokens, $25 per million output tokens. Generally available across all Anthropic products, API, and cloud providers (Microsoft, Google, Amazon).
Factory raises $150M at $1.5B valuation for AI coding agents targeting enterprise engineering teams. Led by Khosla Ventures, with customers including Morgan Stanley, Ernst & Young, and Palo Alto Networks. Differentiator: switching between foundation models like Claude and DeepSeek. The AI coding market remains hot with multiple billion-dollar players competing. For SMBs, this signals continued investment in developer tools, which means more options and lower prices over time.
Anthropic CPO Mike Krieger left Figma's board the same day reports surfaced that Claude Opus 4.7 will include design tools potentially competing with Figma. SEC filing confirmed departure on April 14. Figma stock up 5% since disclosure, suggesting investors aren't panicking yet. Another data point for the "SaaSpocalypse" thesis that AI labs will dominate software businesses. If you use Figma, watch for Anthropic's design tools announcement. Competition here benefits SMBs with better features and pricing pressure.
InsightFinder raises $15M for AI agent observability. Series B led by Yu Galaxy to help companies diagnose where AI models fail by monitoring data, models, and infrastructure together. Example: credit card company's fraud-detection model drift traced to outdated cache in server nodes, not the model itself. If you're deploying AI agents internally, expect more tools like this to help you debug complex failures across the stack.
European AI chip startups raising nine-figure rounds as alternatives to Nvidia GPUs for AI inference. Dutch startup Euclyd seeking $118M, UK's Optalysys planning $100M+ raise. Euclyd claims 100x higher power efficiency for inference compared to Nvidia's Vera Rubin chips. European startups targeting sovereign compute needs and lower energy costs for running AI models at scale. For SMBs, this competition drives down cloud inference costs over the next 2-3 years.
Test Canva AI 2.0's memory features with your brand assets. Upload your logo, color palette, and sample designs during the research preview and let it learn your style. Next time you need social posts or a flyer, prompt with just the concept and see if it applies your brand automatically. If it works, you've saved 30 minutes per asset.
Audit your SaaS subscriptions for AI-native alternatives. Inc. reports subscription creep kills startups. With Canva handling design, Gemini generating images, and Claude writing code, you may be paying for tools you no longer need. Review your bank statement, identify software with new AI competitors, and test free tiers. One or two cancellations could save $500/month.
The design tool wars are heating up with both Canva and Adobe betting conversational AI is the new interface. SMBs win when tools compete on speed, cost, and ease of use. Watch for OpenAI to respond with design features in ChatGPT or Codex. On the model side, Anthropic's intentional capability reduction in Opus 4.7 compared to Mythos Preview sets a new pattern: releasing less capable models publicly to test safeguards before unleashing frontier capabilities. Expect more labs to adopt staged releases with private access for security uses. For businesses, this means the most powerful models will stay behind verification programs while general-purpose models remain broadly available. The AI chip funding wave in Europe signals lower inference costs ahead as competition to Nvidia heats up.