AI DUNN Right Weekly — Issue #30
- Apr 20
- 6 min read
Practical AI insights for business growth
Hey AI Innovators!
This week something shifted. We're not talking about AI anymore as a future thing... we're watching it become the central operating system for how actual businesses work. Design, drug discovery, customer calls, security, code. All of it. And the money backing it? Let's just say nobody's betting against AI anymore.
Here's what you need to know:
Canva AI 2.0 turns your design process into a conversation
OpenAI just raised $122 billion (yes, billion) in one funding round
AI agents are now making phone calls to book your appointments
Perplexity built an AI that lives on your Mac and never stops working
Google's cutting AI energy costs by 100x while actually improving accuracy
Read time: 9 minutes
Big Story
Canva AI 2.0: Design Intelligence That Actually Thinks Like a Designer
Most AI design tools work the same way: you describe something, AI generates an image, you get a result. If it's not right, you try again. And again. Until something sticks. It's functional but it's not efficient.
Canva just changed how this works. They released Canva AI 2.0 and trained it fundamentally differently than other tools. Instead of just learning from finished designs, they trained it on the entire design process... millions of designs and the exact sequence of decisions designers made to build them.
That means when you tell Canva AI "create a minimal ad with tension in the negative space," it doesn't just generate something. It reasons through what you mean the way a designer would. It combines language understanding with actual design knowledge. Then it edits at the layer level... text, images, colors, positioning. Everything stays editable. Nothing gets locked down.
For business owners this is huge. You need marketing visuals. You need them fast. You don't want to wait on a designer. You don't want to spend hours prompting in circles. Canva AI 2.0 sits right in the middle. It gives you professional-looking output on the first try because it understands what you're actually trying to communicate.
Cameron Adams, Canva's co-founder, made a point I keep thinking about: when AI makes everyone good at design, what matters most isn't the tool. It's your judgment. Your understanding of your audience. Your instinct about what will resonate. AI handles execution. You handle strategy.
My take on this? This is the template for how AI tools should work with non-technical people. Not replacing your job. Handling the parts of your job that don't need your human judgment. So you can focus on the parts that do.
What's New This Week
OpenAI Just Raised $122 Billion (In One Round)
This isn't just funding news. This is a statement. OpenAI closed a massive funding round with Amazon ($50B), Nvidia ($30B), and SoftBank ($30B) leading. The message from Wall Street is clear: AI companies aren't risky bets anymore. They're the infrastructure of the next decade. Anthropic is reportedly hitting $30B in annualized revenue. The companies building AI aren't startups anymore. They're becoming the infrastructure layer everyone else builds on top of.
AI Agents That Make Phone Calls (Ring-a-Ding)
Ring-a-Ding just launched AI agents that can make actual phone calls. Not text. Not chatbots. Real phone calls. They handle booking appointments, getting price quotes, checking store inventory. The agent records the call and summarizes it for you. Cost is $19/month. For small businesses this is massive. Imagine not having to hire someone to handle appointment calls. Your AI handles it. You get a summary at the end of the day. It's not the future anymore. It's available right now.
Perplexity's Persistent AI Agent on Mac (Never Stops Working)
Perplexity shipped an AI agent that lives on your Mac full-time. It's not a chatbot that you open and close. It's something that runs persistently... always available for work that's too messy for a normal chat. Think of it as an employee who works while you sleep. The difference between this and Claude Cowork or similar tools is that Perplexity's approach is designed for Mac-native integration and persistence. It's positioned as the thinking partner that never clocks out.
Tool of the Week
Cloudflare Mesh (Secure AI Agent Access)
Here's a problem nobody talks about until it's too late: how do you let AI agents access your internal systems without exposing sensitive data? Cloudflare just solved it with Mesh. It gives AI agents secure access to private company networks in minutes instead of days.
You're not exposing databases to the internet. You're creating a controlled tunnel that only specific agents can use.
Why this matters for your business: if you want to build AI agents that actually work with your internal data (customer databases, pricing, inventory), you need this security layer. Cloudflare Mesh is the easiest way to do it without bringing in a whole security team. Deploy it once. Your agents get access. Your data stays protected.
It's not flashy. It's not exciting to watch. But it's the infrastructure layer that makes the other AI tools actually useful for real companies.
Quick Hits Worth Your Time
HEADS UP: Claude Opus 4.7 Costs More Than You Think — Same price per token as 4.6, BUT the new tokenizer produces 1.0-1.35x more tokens for the same input. Plus the model generates more output at higher effort levels. Your actual costs will jump 30-35% or more even though per-token pricing didn't change. If you use Claude through the API, measure this on your real traffic before upgrading. Don't assume costs stay flat.
→ Google TurboQuant cuts the memory overhead in running large AI models by huge amounts. This is technical but it matters... it means AI models get faster and cheaper to run. Everyone benefits from this.
→ Researchers just published a breakthrough showing how combining neural networks with human-like symbolic reasoning can slash AI energy use by up to 100x while actually improving accuracy. The future of AI isn't just bigger models. It's smarter models.
→ Microsoft is bringing AI agents to your Windows 11 taskbar starting this week. Right-click on something. Get AI help. It's that integrated now.
→ Northwestern University engineers printed artificial neurons that can actually communicate with real brain neurons. This is wild. The merger of machines and human biology just moved from "someday" to "it's happening now."
→ Google's negotiating with the Pentagon to deploy Gemini AI in classified military settings. The line between commercial AI and government AI just blurred significantly.
Prompt of the Week
This week's prompt is about identifying where AI can actually save you time. Not just make things, but do work for you. Use this to audit your business and find the quick wins.
I run a [YOUR BUSINESS TYPE] and I spend too much time on [MAIN TASK CATEGORY]. Here's what I currently do manually: - [Task 1] - [Task 2] - [Task 3] - [Task 4] Which of these could an AI agent actually handle without me needing to check every single output? For each one that could work, tell me: 1. What the agent would do (be specific) 2. What you'd still need me to review (I want to know what I can't automate) 3. What tool would actually work for this (name a real tool if possible) 4. What ROI looks like (time saved per week) 5. The biggest risk if the AI makes a mistake Help me find the 1-2 highest-impact automations I should start with this week.
Why this works: AI agents are good at some things and terrible at others. This prompt forces you to think critically about what actually makes sense to automate versus what you should keep your hands on. Too many people automate things they shouldn't. This helps you find the few things that will actually move the needle.
My Take
I keep noticing something this week that nobody's really talking about. We're past the hype phase. We're not arguing about whether AI will matter anymore. We're watching the infrastructure get built.
OpenAI's $122B funding round isn't about betting on AI's potential. It's about betting on the fact that it's already working. Novo Nordisk is using AI for drug discovery. Banks are getting briefed on security risks. Microsoft is putting agents in Windows. Google is talking to the Pentagon.
For you, the business owner, this means something specific: the window for "figuring out AI" is closing. The companies that move now have an advantage. Not because AI is magic. But because they'll be six months ahead of everyone who waits.
The good news is that the barrier to entry just got lower. Canva AI doesn't require technical knowledge. Ring-a-Ding phone agents cost $19/month. Cloudflare Mesh makes security straightforward. You don't need a team of engineers. You need to start experimenting. Now.
The companies that win in 2026 aren't the ones with the fanciest AI. They're the ones that figure out where AI actually saves time and money... and get that running before their competitors do.
See you next week.
Jackie
Comments