In partnership with
Talk to your AI tools the way you'd talk to a colleague.
You don't send a colleague a three-word brief. You explain the context, the constraints, what you've already tried. But typing all that into ChatGPT takes forever — so you don't.
Wispr Flow lets you speak your prompts instead. Talk through your thinking naturally and get clean, paste-ready text. No filler words. No cleanup. Just detailed prompts that actually get you useful answers on the first try.
Millions of users worldwide. Works system-wide on Mac, Windows, and iPhone.
Anthropic spent two months refusing to give you its best model.
In April, the company built something called Mythos and locked it behind a restricted program. The official reason: it was too good at finding security vulnerabilities. They didn't trust the open internet with it. TechCrunch
On Tuesday, they released it to the public anyway. It's called Claude Fable 5. CNBC
Before release, Anthropic ran an external bug bounty with more than 1,000 hours of testing. Nobody found a universal jailbreak. The dangerous stuff (cybersecurity, biology, chemistry) gets blocked and answered by an older model instead. That's the safety story, and it's what every headline is covering. Yahoo FinanceTechCrunch
Here's what the headlines are missing.
The upgrade that matters for your business isn't intelligence. It's stamina.
Every model release gets sold the same way: smarter, higher benchmarks, better answers. And yes, Fable 5 beat Anthropic's previous best model by more than 10% on some benchmarks. But a 10% smarter answer to your question changes very little about your week. CNBC
What changes your week is this: the model can work on a task for days at a time. It plans its approach, checks its own progress against the goal, and refines its work as it goes. It sustains long, complex work that previous models could not. AmazonMacRumors
Read that again. The shift isn't better answers. It's that you stop asking questions and start assigning work.
Most business owners use AI like a search bar. Type question, get answer, copy, paste, repeat. That habit made sense when models lost the plot after twenty minutes. It's now the most expensive habit in your workflow.
The new habit: hand it the whole job, not the next step of the job.
Old way: "Write me an email to a late-paying client." New way: "Here are my 12 open invoices and my payment terms. Identify which clients are chronically late, draft a tailored collection email for each based on their history, and give me a follow-up schedule for the next 30 days."
Same tool. Completely different output. The first gets you a paragraph. The second gets you a system.
Three honest caveats, because you should know what you're buying into.
First, some questions won't get a Fable 5 answer at all. High-risk topics fall back to the older Opus model. Anthropic's early data says this happens in fewer than 5% of sessions, and for normal business work you'll likely never notice. TechCrunch
Second, Anthropic retains prompts and outputs for 30 days on this model, on every platform, for safety monitoring. It says it won't train on the data, but if you're pasting sensitive client information into any AI tool, that was already a conversation to have with yourself. PYMNTS.com
Third, the clock. Fable 5 is included in paid Claude plans at no extra cost only until June 22. After that it moves to usage credits until Anthropic adds capacity. That's not me manufacturing urgency. That's their published rollout plan. MacRumors
The verdict: ignore the benchmark noise. The real story is that the gap between "AI that answers" and "AI that works" just closed for normal users. If you have a paid Claude plan, you have until June 22 to test the delegation habit for free. Pick one annoying multi-step task this week and hand over the whole thing.
If the output is mediocre, you lost an afternoon. If it works, you just found out what the next two years of your business look like.
Stay sharp.
— @Chris
P.S. Reply with the one task in your business you'd hand to an AI if you trusted it to run for a full day. I'm testing the best submissions in Fable 5 and the results are going in a future issue.