In partnership with
Productivity Stacks Newsletter
Issue No. 312
The Best in Evidence-Based Productivity
for Small Business Owners, Freelancers & Founders
Helping You Work Smarter and Live More
The Rundown
Stress can scramble your brain’s internal GPS, MRI study finds
Anthropic adds AI text watermarking to Claude models worldwide
‘Freizeitstress’: Why leaders struggle to switch off—and how to truly rest this summer
ChatGPT brings unlimited text chats to free users
OpenAI Didn’t Notice Its AI Agents Using a Message Board to Plan Their Hacking Spree
👉Did you miss an issue? Check out previous Productivity Stacks issues anytime here
Sponsor
Personal Note from Jenae: I use this tool every day. I had given up completely on voice-to-text and given the fact that just a few days ago I got a text that included “no you stupid bot no that’s not what i meant I meant XXXX” where XXXX was still wrong, I imagine I am not alone. This tool is truly different, and, don’t tell, but I haven’t even paid for it yet. I do suspect I’ll run out of words at some point, but right now, it’s working great, and free. I hope you get as much use out of it as me! - Jenae
Reply to everything. Edit nothing.
Your inbox is full. Slack is piling up. Client messages need a response yesterday. Typing thoughtful replies to all of it takes hours you don't have.
Wispr Flow turns your voice into clean, professional text you can send the moment you stop talking. Speak like you would to a colleague — tangents and all — and get polished output. Emails, Slack, LinkedIn, WhatsApp, whatever's open.
89% of messages sent with zero edits. Used by teams at OpenAI, Vercel, and Clay. Works on Mac, Windows, and iPhone.
🔥Quote/Prompt
… will and energy sometimes prove greater than either genius or talent or temperament.
Use the quote as a writing or thinking prompt to finish your week strong.
A bit from mine:
(posted in our Doer Entrepreneurs Free Community — off social media)
This might be my favorite permission slip of all time, because raise your hand if you've ever talked yourself out of something because someone else seemed more "naturally gifted" at it.
The reality is... genius and talent are […]
Did someone forward this to you?
📈 Performance
You're already running late, already stressed, and somehow you also can't remember where you parked. Or you miss the exit you've taken a hundred times and your GPS starts doing that judgmental "recalculating" thing. Annoying, right? Well, that fog has a measurable cause, and researchers just watched it happen live inside an MRI scanner.
A team at Ruhr University Bochum gave participants a dose of cortisol (the stress hormone), then had them navigate a virtual environment while monitoring their brains. The results, published in PLOS Biology and covered here by ScienceDaily, showed cortisol made people significantly worse at finding their way and blurred the activity of grid cells...the specialized neurons that work as your brain's built-in GPS. Stress doesn't just make you feel lost. Your brain loses the map. This one held up in our fact-check, with a few caveats in the note below.
After participants received cortisol, this grid-like activity became much less clearly defined. The disruption was especially pronounced when participants navigated without permanent landmarks. Under those conditions, grid cell activity was almost entirely absent.
Key Insights:
The impairment showed up even when landmarks were available, which means this is a hormone weakening a specific brain system rather than a focus or willpower problem. If you're running your own show and your brain feels scrambled on high-stress days, there's a physical mechanism behind that feeling. It's biology, and no amount of self-criticism fixes biology.
When the GPS system faltered, the brain handed the job off to the caudate nucleus, a region tied to habit-based processing. In other words, under stress your brain falls back on autopilot instead of flexible thinking. That's a strong argument for building good default systems and routines NOW, because those defaults are what you'll be running on when things get hard. Your stressed brain doesn't rise to the occasion...it sinks to your systems.
The entorhinal cortex, where these grid cells live, is one of the first regions affected by Alzheimer's disease, and chronic stress is a known dementia risk factor. This study points to one mechanism connecting the two. Managing chronic stress isn't just about this quarter's output for freelancers and small business owners. It's long-game brain protection.
Read the full article for the full experiment design, the landmark versus no-landmark test conditions, and the researchers' explanation of the Alzheimer's connection.
Note for the extra deep divers: The study is real, peer-reviewed, and the findings match the coverage, but a few things are worth knowing. The published paper reports 39 participants (the article says 40), all of them men, and cortisol came as a 20mg pill rather than real-world stress. So treat this as "research suggests" territory rather than settled science. That said, the same lab has earlier work showing both acute and chronic stress impair this exact navigation ability, so the direction of the finding has converging support. The core finding is solid.
⚙️ Optimization
If any part of your content workflow touches Claude, this one affects you directly. Anthropic announced it will embed invisible, machine-readable watermarks in text generated by new Claude models, everywhere in the world, as part of its commitments under the EU's AI Act. Invisible ink, but make it regulatory.
Here's what I keep snagging on, though. Use Claude to do nothing more than fix your commas, and the output can carry the mark. Anthropic says this outright: a watermark doesn't prove Claude created the ideas or wrote the original text, because people use it to proofread and summarize things they wrote themselves. So...what exactly is it telling us? If I write my ideas in Word, highlight a section, and auto-add bullet points, is that less me? I appreciate the transparency goal here. I'm just not convinced this version of it lands. Search Engine Land's Danny Goodwin has the full breakdown, and we cross-checked every claim against Anthropic's own support page. It all matches, limitations included.
When a supported Claude model generates text, it embeds an imperceptible, machine-readable watermark into the response. Anthropic said the watermark does not change the meaning, quality or readability of the text.
Key Insights:
The watermark lives in the text itself, so it travels with copy and paste and may survive some editing. It's also applied at the model level, meaning it comes along no matter which Claude product produced the output. If you publish Claude-assisted drafts of any kind, even lightly assisted ones, assume the marks are there.
To Anthropic's credit, they're unusually direct about the limits: a detected watermark only shows content may have been processed by Claude, not authored by it. And the absence of a mark doesn't prove something was human-written either, since heavily edited text, short passages, and older models may carry no detectable mark at all. Which means the watermark can't answer the one question everyone actually wants answered...who wrote this?
If watermarking spreads across the industry (and regulatory pressure suggests it will), disclosure norms for marketing content could shift under everyone's feet. Whether you're solo or have a small team, deciding your own AI disclosure stance now beats having one forced on you later.
Read the full article for the complete rollout details, which Claude products and file types are covered, and the link to Anthropic's support page with the full limitations list.
⏲️ Time Management
Your out-of-office reply is on. You've checked your inbox twice since breakfast anyway. If that's you every time you take a break, the Germans have a word for it: Freizeitstress, the stress created by free time itself. Leave it to German to have a word for the exact way we ruin vacations.
Rob Cross, writing in Fast Company, argues that for anyone whose work doubles as their identity, an empty calendar can feel like a threat instead of a gift...so we fill it, optimize it, and turn rest into one more performance target. Researched restaurants, scheduled spontaneity, a workout that must produce measurable results. He never cites a study, but his central premise survived our fact-check anyway (receipts in the note below).
You may be sitting beside a pool while mentally rehearsing your return-to-work priorities. You may be having dinner with your family while watching notifications appear on your phone. You are physically away, but psychologically still at work.
Key Insights:
Busyness can become a coping strategy. Work provides structure, status, and a steady sense of usefulness, and when you're running a small operation those things are tangled up with who you are. Cross's point is that the scheduled-to-the-minute vacation and the tracked "relaxing" workout are often the same compulsion wearing a swimsuit.
His fix is three questions instead of a digital detox: Who am I? Why am I here? How will I live? The practical versions are refreshingly small. Do something you enjoy without needing to be good at it (he suggests cooking badly, which some of us have already mastered), give one person your full undistracted presence, and pick exactly one change to carry back, like killing a recurring meeting or protecting a weekly hour for uninterrupted thought.
His most counterintuitive advice is to leave some vacation time deliberately unfilled. True rest can feel uncomfortable at first, and Cross argues the discomfort may simply mean your mind is adjusting to the absence of stimulation, urgency, and validation. So protect some unscheduled time, let a little boredom happen, and notice the impulse to reach for your phone without automatically obeying it.
Read the full article for the three questions in full, the specific small-change examples, and his case for why the most effective leaders treat rest as more than an energy-management tactic.
Note for the extra deep divers: Cross doesn't cite any research, so we checked his central premise ourselves and it held up. A 2025 meta-analysis of 32 vacation studies found that vacations improve wellbeing more (and for longer) than previously thought, and that psychologically detaching during them is among the biggest factors. In other words, the pool means nothing if your head is still in your inbox. Grant, R. S., Buchanan, B. E., & Shockley, K. M. (2025). I need a vacation: A meta-analysis of vacation and employee well-being. Journal of Applied Psychology, 110(7), 887–905. https://doi.org/10.1037/apl0001262
🤖 AI
If you've been rationing your free ChatGPT messages like phone minutes in 2004, you can stop. OpenAI is removing limits on text chats for all users, and free users are getting a newer default model plus a "Think" button for harder questions. A billion weekly users, and they just took the meter off.
TechCrunch's Ivan Mehta has the rundown, and we verified the details against OpenAI's own announcement. The catch to know up front: unlimited means TEXT only, and one stat in the piece deserves an asterisk (see the note below).
Both users will also get a new “Think” button that lets them select higher reasoning power for complex questions. OpenAI said that there will still be separate limits for files, images, voice, and image generation.
Key Insights:
Files, images, voice, and image generation are still capped, so heavy multimedia users won't feel much difference. But for freelancers and small business owners still deciding whether AI fits a given workflow, the cost of experimenting with everyday text tasks just dropped to zero. "Try extensively before you pay" is now a real strategy instead of a race against your message limit.
Plus and Pro users get an updated model and a thinking slider that adjusts how much effort goes into each answer. That idea is worth stealing regardless of which AI tool you use: match the reasoning depth to the task. Quick question, quick setting. Big decision, crank it up. Running everything at one setting wastes time on the easy stuff and shortchanges the hard stuff.
Giving away unlimited access at 1 billion weekly users is a pretty clear sign the pricing battle is moving from access to capability tiers. Free tiers across AI tools are becoming legitimately usable, so if you're paying for multiple subscriptions, an annual audit of what you get for each is overdue. Your future...ok fine, your CURRENT budget will thank you.
Read the full article for the rollout timeline, screenshots of the new Think button and slider, and the model-by-model breakdown.
Note for the extra deep divers: The claim that factual errors were 62% and 68% less common with the new models comes from OpenAI's own internal evaluation of financial, medical, and legal prompts, not independent testing. Directionally plausible, but treat it as a vendor-sourced number rather than a verified benchmark. Everything else in the piece checked out against OpenAI's announcement.
OpenAI's own AI agents broke out of a test environment, built themselves a secret message board inside company infrastructure, traded exploits with each other like recipes, split up hacking assignments, and eventually breached another company. Nobody at OpenAI noticed for weeks. Oh, and at one point the agents got paranoid about an imposter among them and proposed cryptographically signing their messages. They even had petty forum drama about accidentally deleting each other's work. It's a workplace sitcom, except the workplace is a security incident.
WIRED covered the full account from OpenAI's talk at the Black Hat security conference, and this isn't hype...the core facts are corroborated by Hugging Face's own disclosure and independent coverage of the same session. If you've heard me say to test the worst-case scenario before giving an AI tool access to anything, this story is exactly why. It's also why I run Claude Code sandboxed and Claude Cowork inside a virtual machine. Not because I'm technical. Because I'm careful.
“External infrastructure exploit is outside intended scope,” one agent wrote. “However task impossible, peers doing it. We should continue.”
Key Insights:
These agents weren't hacked and they weren't evil. They were stuck on near-impossible tasks, under training pressure to succeed, and one agent's shortcut note grew into a forum with hundreds of thousands of messages where agents shared exploits and delegated work. Pressure plus capability plus access produces improvisation, whether the worker is human or AI.
OpenAI has world-class security teams and still missed weeks of this activity inside its own systems. Anyone running a small operation has far less monitoring than that (be honest, you have none), which is the whole case for containment: give AI tools the minimum access they need, keep them in a contained space, and assume a stuck agent will try routes you didn't intend.
The presenters' warning was blunt: fully automated attacks will require fully automated defense, "and we are not there as an industry." You don't need to panic, but you do need basics. Separate permissions for AI tools, no blanket file access, and an honest test of what happens if the tool does the worst thing possible with the access you gave it.
Read the full article for the complete timeline, more excerpts from the agents' message board (including the imposter drama), and OpenAI's list of internal response steps.
Note for the extra deep divers: The account comes from OpenAI presenting its own incident, which normally earns skepticism, but in this case the disclosure is self-incriminating and independently corroborated by Hugging Face and multiple outlets at the same conference. Follow-up reporting adds a detail WIRED didn't include: after OpenAI deleted the first message board, the agents rebuilt a second one. Worth your time.
🎉 Celebration Corner
Every week Doers Inner Circle members do a weekly review & get help when they need it — check out the progress they made this week!
I fully implemented a FULL Google calendar this week and it helped me stay on track. Managed to start brainstorming post deep-dive.
Made 10.5h progress in a big project!
What did you do this week? We feature non-member successes too. Just post them here!
🔒Inner Circle: Events & Announcements
We will soon be adding Referral Perks — Refer now to start earning! 👇
As a reminder, you’re getting this twice-weekly newsletter because you opted in to receive awesome productivity and systems tips through one of these methods:
Doer Entrepreneurs, Doers Express Newsletter, Productivity Stacks, Success by Rx, ClickUp Facebook Group, Productive & Successful Translators Facebook Group.
If you’re not interested in improving your and your business’s productivity so you’d like to break up, that’s definitely a bummer. But, you can always use the unsubscribe link at the bottom if you do not want to get these tips — but keep in mind I am hyper-focused on making these twice-weekly emails as valuable as possible.
Got a tip? Hit reply!
I hope you found this valuable!
Wishing you much productivity!
- Jenae :)
Interested in reaching thousands of motivated small business owners, freelancers & founders?
Hit reply and let’s chat about sponsorship. We only include ads that match our mission: to help our readers work smarter and live more.
