Prompting, for dummies

The newest models are asking you to pay them to put in more effort. Oh, and we launched something brand new at FryAI! 😁

[Sponsor FryAI]

This link leads to one of the most interesting things I’ve seen in AI recently.

What’s cookin’? Anthropic released Claude Opus 5, a new AI model built to handle coding, research, business tasks, and everyday knowledge work more efficiently. The big pitch is performance per dollar: Anthropic says Opus 5 gets close to its more powerful Fable 5 model on many tasks, but costs about half as much to run. It also improves on Opus 4.8 in software engineering, automation, computer use, science tasks, and visual outputs. One key feature is adjustable “effort” settings, which let users choose whether they want the model to think harder for better answers or conserve tokens for faster, cheaper results.

🤔 Hunter’s take: What’s interesting here is the “effort” control. AI models are starting to look less like one-size-fits-all products and more like adjustable tools: cheap and fast when the task is simple, slower and more expensive when you need deeper reasoning. That matters because it gives users more control over the tradeoff between quality, speed, and cost. The future of AI may not just be “which model is smartest?” It may be “how much intelligence do I want to spend on this task?”

Most CX platforms do not own the voice. They orchestrate a workflow, then call a third party for speech and transcription. Every hop adds latency, cost, and another vendor to manage.

ElevenAgents is the opposite. They make the voice models the market builds on, and ElevenAgents puts full orchestration on top. Voice, transcription, text-based chat, and reasoning run in one vertically integrated pipeline, so responses come back in <400 milliseconds and sound human, not synthetic.

Plus, you keep full control. Plug in any LLM, integrate tools, webhooks, and MCP servers, and ground responses in your knowledge base. Get an agent live in minutes, then A/B test with Experiments, enforce Guardrails, and version every change.

The payoff: more human conversations, lower latency, and far less time stitching infrastructure together. You build on the models you already trust. Pricing is transparent and flat at $0.08 per minute.

Start building free

What’s cookin’? Nvidia, Microsoft, SpaceX, Palantir, and dozens of other companies have launched the Open Secure AI Alliance, a new effort to build and share open AI tools for cybersecurity. The group’s goal is to help companies find, fix, and publicly disclose software vulnerabilities using AI systems that can be downloaded, inspected, modified, and run on their own infrastructure. That matters because many of these powerful tools are closed, meaning users can only access them through the company that built them. The alliance is arguing that cyber defenders need more resources and control, especially when attacks are moving fast.

🤔 Hunter’s take: The surprising lesson here is that the answer to AI getting out of control may not be locking everything down. It may be opening up more of the safety layer. If only a few closed companies control the strongest AI defense tools, everyone else is stuck waiting for their rules, updates, and permissions. But open safety tools let researchers and defenders inspect what’s happening, adapt quickly, and build protections that actually fit the threat in front of them.

…

👉 You may have realized by now that AI is only as good as your prompt.

That’s why we offer professional AI Prompt Packs: ready-to-use prompts that help you stop staring at the blinking cursor and start getting real value from your favorite chatbot.

Hi — I’m Hunter, a PhD candidate whose work has appeared in major academic journals and popular tech outlets. I founded FryAI to make staying ahead of AI clear, accessible, and fun.

How close are we to the point where AI surpasses human intelligence across nearly every domain and begins advancing beyond our control or prediction?