LinkedIn Posting Cheat Sheet: Dmitri Sunshine / Solanasis
Created: March 23, 2026 20 post ideas × 3 variations each = 60 ready-to-post options Voice: Dmitri’s personal LinkedIn voice (casual, first-person, teacher energy)
How to Use This Cheat Sheet
Three styles for every idea; pick the one that matches your energy that day:
- Style A: “The Casual Teacher” = Mid-thought opener, teaches as you go, practical tips, conversational. Like the password manager post.
- Style B: “The Sharp Take” = Bold opinion up front, backs it up with reasoning, slightly more provocative. Like the OpenClaw posts.
- Style C: “The Story Drop” = Opens with a personal experience or observation, pivots to the lesson, warm and relatable. Like the executive energy hack post.
Quick reference for your voice rules:
- First person (“I”), 1-2 sentences per paragraph, 6-14 lines
- No em dashes, no engagement-bait questions, no CTAs
- “Here’s the thing:” and “One more thing” are your natural transitions
- 60% professional / 40% raw
Table of Contents
- Creatine Over Coffee / Executive Health Hacks
- Exporting ChatGPT Sessions to Your Vault
- Short.io for Link Tracking
- Claude Code Daily Brief
- Google Nano Banana for AI Image Generation
- Cloudflare DNS / WARP / Zero Trust
- Opting Out of AI Training Data
- UX Insights: Referral Source & Country Selectors
- The Ben Franklin Effect / Calls Beat Referral Asks
- Website Security: Chrome’s Built-in Protection
- AI Cannot Replace Lack of Procedures
- Text-to-Speech / Wispr Flow for AI
- Claude vs ChatGPT: Measure Twice vs Cut Twice
- CLI Wrappers vs MCPs
- Moving from Google Docs to Notion or Coda
- Hype Fades, Systems Remain / AI Slop
- Buy Back Your Time / Investing in Tools
- DAF Funding and Nonprofit Giving
- Mining Your LinkedIn Data with AI
- Systems Thinking: One Post Per Tool
PILLAR 1: HEALTH & LEADERSHIP
1. Creatine Over Coffee / Executive Health Hacks
Category: Personal / Health / Leadership Source: Post Ideas doc + Daily Notes (March 15-16)
Style A: The Casual Teacher
So it’s 2026 but I still see executives running on coffee and hope.
Here’s the thing: caffeine is borrowing from the bank at high interest. You get the spike, then you crash, then you need more. Rinse, repeat, burnout.
I switched to a stack that actually works. 10g+ creatine daily, L-Tyrosine, L-Carnitine, then L-Theanine with your coffee. Add Alpha-GPC or Choline and your brain actually stays level.
Not flashy. Not biohacking bro nonsense. Just sustained energy that doesn’t leave you wired at 10 PM.
One more thing: this isn’t magic. It takes about 2 weeks to feel the difference. But once you do, you won’t go back to the coffee treadmill.
ExecutiveHealth ProductivityHacks FounderLife
Style B: The Sharp Take
Most executives are self-medicating with caffeine and calling it focus.
Let’s be frank: that’s a losing strategy. Caffeine spikes cortisol, burns through your dopamine, and leaves you exhausted by noon. You’re not more productive, you’re just more jittery.
The fix is boring but effective. Creatine (10g daily), L-Tyrosine, L-Carnitine, L-Theanine with your coffee, and Alpha-GPC. Science-backed compounds that actually stabilize your energy and cognition.
Which is why you’ll see better decision-making, less afternoon crashes, and sleep that isn’t destroyed by 3 PM caffeine.
Take 2-3 weeks to dial in. Then you’re running on your actual neurochemistry, not fumes.
ExecutivePerformance FounderHealth OptimizedLiving
Style C: The Story Drop
Last year I hit a wall. Coffee at 6 AM, energy tanked by 10 AM, more coffee by noon, couldn’t sleep before midnight.
I realized I wasn’t tired. I was just burning out my dopamine receptors on cheap caffeine spikes.
So I rebuilt it. Started taking 10g creatine daily, L-Tyrosine, L-Carnitine. Added L-Theanine with my morning coffee and Alpha-GPC for choline. Took three weeks to notice.
The difference was quiet. No crash at 2 PM. Sharper decisions in meetings. Actually tired at 10 PM, not wired.
Not a hack. Just healthy energy management.
FounderLife ExecutiveHealth SustainablePace
PILLAR 2: AI & TOOLS
2. Exporting ChatGPT Sessions to Your Vault
Category: AI / Knowledge Management Source: Post Ideas doc
Style A: The Casual Teacher
Your ChatGPT conversations are disappearing and you don’t even realize it.
Most people treat GPT like a search engine. Ask, get answer, move on. But you’re sitting on a goldmine of your own thinking, refined through AI conversations.
Here’s the thing: you can export everything. All your conversations. Drop them into Obsidian, Notion, or whatever vault you use.
Over time, this becomes your personal AI database. Prompts that worked. Ideas you workshopped. Decisions you talked through. It’s all there.
One more thing: this forces you to be intentional about what you’re asking AI to help with. Once it’s in your vault, you start seeing patterns.
KnowledgeManagement ArtificialIntelligence PersonalVault
Style B: The Sharp Take
You’re having valuable conversations with AI and throwing them away.
ChatGPT doesn’t keep your data forever. Neither does Claude. The conversation disappears, and you’ve lost the thinking you just articulated.
But you can export every single session. Store it in your vault. Build your own searchable database of your AI conversations.
Which is why smart operators are already doing this. You accumulate months of prompts, frameworks, and refined thinking. That’s intellectual property.
Start exporting today. You’ll thank yourself in 6 months.
DataPrivacy PersonalKnowledge AILiteracy
Style C: The Story Drop
I realized something a few weeks ago: I’d had the same conversation with three different AI tools and written it down nowhere.
Each time I was starting from scratch. No institutional memory. No pattern recognition across my own thinking.
So I started exporting. All my ChatGPT sessions, Claude conversations, everything. Dropped them into Obsidian.
Now when I need to revisit a topic, I’m not just searching the internet. I’m searching my own refined thinking, conversations I’ve already had with AI.
It’s like having a second brain that remembers every conversation you’ve ever had.
PersonalKnowledge ProductivityHacks ArtificialIntelligence
3. Short.io for Link Tracking
Category: Tools / Ops Source: Post Ideas doc
Style A: The Casual Teacher
Most founders share raw URLs or grab a bit.ly link and hope for the best.
But here’s what people miss: once you print a URL on a business card or an invoice, it’s locked. You can’t change it. If that destination changes, you’re stuck.
Short.io solves this. You create a short link, track every click, and then update where it actually goes, whenever you want.
Business card pointing to your old website? Update it in seconds. Email signature link outdated? Change it. No broken links, no mess.
Also, you should never underestimate what click data tells you about what people actually care about.
Productivity FounderTools LinkTracking
Style B: The Sharp Take
Your link strategy is costing you data and flexibility.
If you’re using bit.ly or raw URLs, you’re missing two critical things: you don’t know click counts, and once printed or embedded, those links are permanent.
Short.io fixes both. Create trackable short links, see exactly who’s clicking, and update the destination URL whenever you need to.
Which is why it’s invaluable for physical materials. Business cards, printed invoices, handouts. Link breaks? Just redirect it.
It’s not fancy. But it’s smarter than what most founders are doing.
MarketingOps Founder BusinessTools
Style C: The Story Drop
I realized we were printing our website URL on business cards, then changing the website six months later.
Dead link. No redirect. No data on who clicked it. Just gone.
So we switched to Short.io. Now every business card, every invoice signature, every printed material has a trackable link that we can update anytime.
Someone needs the old landing page? We just redirect it. We see click counts. We know what’s actually being used.
One less operational headache, and better data for everything we do.
OperationalExcellence SmallBiz FounderHacks
4. Claude Code Daily Brief
Category: AI / Productivity Source: Post Ideas doc + Daily Notes
Style A: The Casual Teacher
Your morning should start with clarity, not chaos.
Most founders wake up to 47 Slack messages, 20 emails, and a ClickUp board that’s somehow grown overnight.
Here’s the thing: I use Claude Code to pull a daily brief. It runs through my Coda wiki, grabs my transcripts, checks my ClickUp tasks, synthesizes everything into one page.
Takes 2 minutes to read. Clear priorities. No surprises.
It’s like having a chief of staff who read everything while you slept, filtered out the noise, and left you exactly what matters.
One more thing: this only works if your information is actually organized. So you fix your vault as a side effect.
Productivity FounderLife ArtificialIntelligence
Style B: The Sharp Take
Your information is scattered and you’re losing decisions every day.
Notes in Notion. Meeting transcripts in Gmail. Tasks in ClickUp. Context trapped in Slack. You’re manually stitching it together every morning, missing things, making slower decisions.
But Claude Code can pull from all of it automatically. Coda, transcripts, tasks, meeting notes, everything. Synthesize into one coherent brief.
Which is why your morning becomes clear instead of chaotic. One page. What matters. Priorities crystallized.
This is what delegation looks like in 2026.
Automation FounderTools ProductivityOps
Style C: The Story Drop
Last month I was starting every day scrambling. I’d read emails, then Slack, then check ClickUp, then realize I’d missed context from a meeting transcript. Three hours of cognitive load before I could actually think.
So I built a Claude Code automation. Every morning it pulls my Coda notes, last night’s transcripts, my top ClickUp tasks, synthesizes them into a single brief.
Now I wake up to one page. What changed. What matters. What needs my attention.
It’s the difference between starting my day reactive and starting it strategic.
Automation FounderHacks ArtificialIntelligence
5. Google Nano Banana for AI Image Generation
Category: AI / Content Source: Post Ideas doc
Style A: The Casual Teacher
So I’ve been using Google’s Nano Banana model to auto-generate images for blog posts, and it’s cut the time I used to spend hunting stock photos in half.
Here’s the thing: the model doesn’t care if you feed it garbage prompts. You get garbage images back. But if you’ve already dialed in your brand voice and visual style, you can generate 10 variations and pick the best one in minutes.
It’s not about replacing a designer; it’s about replacing the tedious stock photo hunt. You own the output, it matches your voice, and you’re not paying $200 a month for a subscription service.
Also, don’t expect photo-realistic perfection. It’s good enough for blog headers, featured images, thumbnails. That’s the actual use case here.
One more thing: batch generate and store your best ones. You’ll build a personal library that actually looks cohesive.
AI Content Productivity Design
Style B: The Sharp Take
Most people are still wasting hours on stock photo libraries when Google’s Nano Banana can generate custom images that match their actual brand voice.
The secret is obvious once you think about it: you have to know what your style is before you start. If you don’t have a defined visual identity, the AI won’t save you. But if you do, generating 10 variations and picking the best one takes 10 minutes instead of an hour.
You own the output. It’s unique. It costs nothing.
The mistake I see teams make is treating it like a replacement for real design work. It’s not. It’s a replacement for the soul-crushing part of design work: image sourcing.
If you’re still paying for stock photo subscriptions, you’re not paying attention.
AI Content Design Productivity
Style C: The Story Drop
I spent two hours last week hunting for a blog header image. Two hours. Scrolled through three different stock photo sites, got nothing that felt right, settled on something mediocre.
That’s when I actually sat down and tested Google’s Nano Banana model.
Twenty minutes later I had 10 variations. Picked one. Done. And it actually matched the tone of my writing instead of looking like every other generic startup blog.
The key here is that you have to know your visual voice first. Can’t generate what you haven’t defined. But once you have that dialed in, the tool removes the friction.
Now I batch-generate images when I’m writing. It’s part of the workflow, not a separate step.
AI Content Design Productivity
PILLAR 3: SECURITY & OPERATIONS
6. Cloudflare DNS / WARP / Zero Trust
Category: Security / Ops Source: Post Ideas doc
Style A: The Casual Teacher
Most businesses never think about DNS security at all. You’re just using whatever your ISP gave you, which is like leaving your front door unlocked.
Cloudflare’s 1.1.1.1 DNS is faster and more private than your default. Free to use. Literally takes three minutes to set up on your router or device.
If you want more, their WARP app adds a VPN-like layer for free. Your traffic gets encrypted, ISP can’t see what you’re doing, and you get the speed bump on top of it.
Then there’s Zero Trust for teams. That’s the enterprise version. Everyone gets secure access to company resources without a traditional VPN. Works beautifully for remote teams.
Here’s the thing: these tools are basically free or dirt cheap. Most orgs aren’t using them because nobody’s told them to.
Also, this isn’t paranoia. If you’re handling any customer data or proprietary work, DNS security matters.
Security Cloudflare Operations RemoteWork
Style B: The Sharp Take
Your business is probably leaking data through DNS queries right now and you don’t even know it.
Your ISP’s default DNS is slow, tracks what you visit, and gives you zero privacy. Cloudflare’s 1.1.1.1 solves that. Free. Faster. More private. It’s not complicated.
WARP is the next layer. Free VPN-like protection for individuals and teams. Takes two minutes to install.
Zero Trust is where it gets real. That’s the enterprise play: every user, every device, every connection gets verified before they touch your systems. No VPN brittleness, no backdoors.
Let’s be frank: most companies are sleeping on this because security hasn’t caused them obvious pain yet. That’s luck, not strategy.
The tools exist, they’re affordable, and they work.
Security Cloudflare Operations DevOps
Style C: The Story Drop
I watched a client’s team get hit with a DNS hijack attempt last year. Wasn’t catastrophic, but it was close.
After that I started digging into what we were actually using. Turns out they were on their ISP’s default DNS, no encryption, zero monitoring.
That’s when I tested Cloudflare’s stack. 1.1.1.1 for DNS, WARP for the team, and Zero Trust for the infrastructure.
Setup took maybe two hours across the whole company.
The difference is night and day. Faster, more private, actually gives you visibility into what’s happening on your network.
Here’s the thing: this isn’t a “nice to have.” If you’re doing any real business online, you need to own your DNS layer.
One more thing: it’s not expensive. These tools basically pay for themselves in the time you save on security incidents.
Security Operations Cloudflare RemoteWork
7. Opting Out of AI Training Data
Category: Security / AI / Privacy Source: Post Ideas doc
Style A: The Casual Teacher
If you’re using ChatGPT, Claude, or any other AI tool for business, your conversations are probably being used to train future models by default.
Which is fine if you’re working with public information. But if you’re handling proprietary data, sensitive client work, or anything you wouldn’t want in a training dataset, you need to opt out. Actually opt out, not just assume.
Here’s the thing: this is basic data hygiene. You wouldn’t give your lawyer’s password to a stranger. Same logic applies here.
The opt-out settings are usually buried in account preferences. ChatGPT has it under Settings > Data Controls. Claude has it in Account Settings.
Also, if your team is using these tools, make it a policy. Opt-out by default, opt-in only for non-sensitive work.
One more thing: this isn’t paranoia. It’s just being intentional about where your work ends up.
AI Security DataPrivacy Operations
Style B: The Sharp Take
Your business data is training AI models right now unless you explicitly told the platform not to use it.
Most AI tools train on your inputs by default. That’s how they improve. But if you’re using ChatGPT or Claude for anything proprietary, sensitive, or client-related, that’s a massive problem.
The fix is three minutes of clicking. Opt out in account settings. Do it for everyone on your team.
Let’s face it: this seems paranoid until you realize it’s not. Proprietary business logic, client data, financial information, product strategy. You’re either controlling where it goes or you’re not.
The platforms make it optional because most people don’t think about it. That’s their advantage.
Stop being most people.
AI Security DataPrivacy Operations
Style C: The Story Drop
I worked with a team that was using ChatGPT to brainstorm product ideas, code architecture, client proposals. Everything. It wasn’t until a client asked, “Where is our data going?” that anyone stopped to check.
Turns out OpenAI was using it all for training by default.
They weren’t hacked. The tool wasn’t stealing anything. But they’d signed the terms of service without reading the data policy.
That’s when I realized most people don’t know opt-out settings even exist.
Here’s the thing: you have control over this. ChatGPT has it under Settings > Data Controls. Claude under Account Settings. Most major platforms have it. Takes three minutes.
If you’re handling anything proprietary, sensitive, or client-related, opt out. Make it policy for your whole team.
One more thing: this isn’t about paranoia. It’s about intentionality. Know where your work is going.
AI Security DataPrivacy Operations
8. UX Insights: Referral Source & Country Selectors
Category: Product / UX / Ops Source: Post Ideas doc
Style A: The Casual Teacher
I’m surprised by how many apps still don’t list ChatGPT or Claude as referral source options.
They have the big ones covered, sure. Google. Twitter. Reddit. Then there’s this sad “Other” dropdown that catches everything else. By now, “Other” is probably your biggest bucket and you don’t even know it.
Here’s the thing: it’s a two-minute fix that tells you how people are actually finding you.
Same thing with country selectors. I see apps that list countries alphabetically. So if your users are mostly in the US, they have to scroll past fifty countries to find it. Just show your top five at the top and alphabetical below.
Small UX fix. Shows you actually pay attention.
Also, I’ve built this into my own apps and watched the data quality improve immediately.
One more thing: these fixes aren’t about delight. They’re about basic thoughtfulness.
UX ProductDesign Data Operations
Style B: The Sharp Take
Most product teams have no idea where their users actually come from because their referral source dropdown is broken.
ChatGPT and Claude are probably your biggest sources now. But you have them listed as “Other” because you built your form in 2018 and never updated it.
Same with country selectors. Alphabetical order is fine if you have three users from each country. If 60% of your users are in the US, stop making them scroll to find it.
These aren’t complicated fixes. They’re just design that actually reflects reality.
I started putting this in my own products and the data quality jumped. Turns out when people don’t have to guess, they give you better information.
It’s not a feature. It’s just paying attention.
Style C: The Story Drop
I was reviewing analytics on one of my products and noticed 40% of our referral traffic was marked “Other.”
Forty percent. I had no idea where these users were coming from.
Turned out I’d left ChatGPT and Claude off the referral dropdown. They’d been referred by AI tools, clicking through, all landing in “Other.”
That’s when I realized the form was built for 2018, not 2026.
I updated it. Added ChatGPT, Claude, the usual suspects. Suddenly the data made sense.
Did the same thing with countries. Was showing them in alphabetical order when 70% of my users were in the US.
Here’s the thing: you don’t need clever UX. You just need UX that matches where your actual users are.
UX ProductDesign Data Operations
PILLAR 4: SALES & RELATIONSHIPS
9. The Ben Franklin Effect / Calls Beat Referral Asks
Category: Sales / Psychology / Networking Source: Post Ideas doc
Style A: The Casual Teacher
So I notice a lot of founders still ask for referrals instead of asking for a call, and here’s why that’s backwards.
The Ben Franklin effect is real. When someone gives you 15 minutes of their time, they actually like you MORE afterward, not less. It’s a small favor that deepens the relationship.
Here’s the thing: asking for a call is asking for something tiny. Asking for a referral is asking someone to stake their reputation on you. That’s heavy.
Get on the call, let them see what you actually do, and referrals happen naturally. No pressure needed.
Also, you should never lead with “got anyone I should talk to?” Lead with “got 15 minutes?”
LinkedIn Sales Networking Relationships
Style B: The Sharp Take
Asking for referrals doesn’t work. Asking for calls does.
Most people have it backwards. A referral ask puts someone in a position where they have to vouch for you to someone else. That’s friction. They don’t know if you’ll follow up, if you’ll be professional, if you’ll reflect well on them.
A 15-minute call? That’s nothing. And here’s the thing: the Ben Franklin effect is real. When someone does you a small favor, they like you more afterward, not less. It’s counterintuitive but it works.
Once they’re on the call and they see what you do, referrals come organically. No ask needed.
Stop asking for what you want. Ask for what deepens the relationship first.
Sales Networking Psychology LinkedIn
Style C: The Story Drop
I had a mentor tell me something years ago that stuck: “Never ask for the referral. Ask for the call.”
I thought he was wrong. Seemed like extra steps. But then I saw it play out over and over, and it clicked.
When you ask someone to refer you, you’re asking them to risk their credibility. They tense up. You can feel it in their response.
When you ask for 15 minutes, they relax. It’s small. And here’s the Ben Franklin effect in action: the favor actually makes them like you more.
Once on the call, they understand what you do. Referrals happen naturally because now they can confidently point people your way.
Which is why the best networking move is always the simplest: just ask for the time.
Networking Sales Relationships Growth
10. Website Security: Chrome’s Built-in Protection
Category: Security / Quick Tip Source: Post Ideas doc
Style A: The Casual Teacher
Most people I talk to install browser extensions for security like WOT or similar, thinking it’s the best defense.
Chrome already has something better built in, and almost nobody uses it.
Settings > Privacy and security > Security > Enhanced Protection. That’s it.
Here’s the thing: Google maintains a real-time database of dangerous sites. Enhanced Protection checks against it before you even load the page. It’s faster than any extension and you don’t have to trust some third party with your browsing data.
One less extension cluttering your browser. One less permission you have to grant. Better protection already there.
Security Chrome Privacy WebSafety
Style B: The Sharp Take
You’re probably running unnecessary security extensions when Chrome already has a better built-in option.
Enhanced Protection is in Settings > Privacy and security > Security. Most people never find it.
It uses Google’s real-time database of dangerous sites. Blocks malware and phishing before the page loads. No lag, no clutter.
An extension like WOT is slower, asks for more permissions, and you’re trusting a third party to monitor your browsing. Enhanced Protection is native, it’s instant, it’s already there.
Stop overcomplicating security. Use what’s built in.
Security Privacy Chrome Cybersecurity
Style C: The Story Drop
I watched someone get a warning from an extension they’d installed for “security,” and it made me think about how much we trust third parties we don’t know.
Turns out Chrome has something better built in that almost nobody knows about.
Enhanced Protection lives in Settings > Privacy and security > Security. It runs Google’s real-time database of dangerous sites. Stops malware before it loads. No extension, no extra permissions, no middleman.
It’s faster than any extension and it was there the whole time.
Which is why one of the smartest security moves is just: turn off what you don’t need and use what’s already there.
Security Privacy Chrome WebSafety
PILLAR 5: OPERATIONS & DOCUMENTATION
11. AI Cannot Replace Lack of Procedures
Category: AI / Operations / Consulting Source: Daily Notes (March 22)
Style A: The Casual Teacher
A founder reached out asking about using AI to auto-generate procedures from their project management tools.
Here’s the problem: AI can’t read minds. If your procedures aren’t documented, there’s nothing for AI to work with.
You need actual interviews and documented procedures BEFORE you hand control over to AI. That’s step one.
Here’s the thing: a consultant is basically a cheatcode because they know the 80/20 of what to document first. They don’t generate it from thin air either. They ask the right questions and organize what’s already there.
AI amplifies what exists. It doesn’t create what’s missing. If your foundation is weak, AI just makes it faster and weaker.
AI Operations Procedures Leadership
Style B: The Sharp Take
AI cannot replace lack of procedures and documentation. This is the mistake I see over and over.
A founder wanted to use AI to auto-generate their operational procedures from their PM tools. Sounds smart, right? It’s not.
AI reads what’s there. If nothing is documented, it has nothing to read. It can’t know what’s happening in your head or what actually matters.
You need the real work first: interviews, notes, documentation of how things actually run. Then AI can amplify it, organize it, make it scalable.
A consultant accelerates this because they know exactly what to ask and what to document first. They’re not creating from nothing either.
AI amplifies. It doesn’t create.
AI Operations Procedure Scaling
Style C: The Story Drop
I was talking to a founder who wanted to use AI to auto-generate procedures from their project management system.
The idea made sense at first. Why not let AI pull it all together?
Then I realized the real problem: they didn’t have procedures documented. They had chaos organized in a tool.
AI can’t read minds. It can only work with what’s written down. You need the foundation first. Real interviews, real documentation of how things actually run.
Here’s the thing: a consultant speeds this up because they’ve done it a hundred times. They know exactly what to ask and what matters most to document.
Which is why the best use of AI isn’t to skip the hard work. It’s to amplify the work you’ve already done.
AI Operations Scaling Leadership
12. Text-to-Speech / Wispr Flow for AI
Category: AI / Tools / Productivity Source: Daily Notes (March 16)
Style A: The Casual Teacher
Most people type everything into AI tools, and it’s slow and loses context.
Voice-to-text tools like Wispr Flow let you talk naturally instead, and your results get way better because you include more context without thinking about it.
When you type, you filter. When you talk, you think out loud. Talking includes the reasoning that typing cuts out.
Here’s the thing: this is especially powerful in terminal or code environments where you’re already context-heavy and typing slows you down.
Also, the two best follow-up prompts are always the same: “Are you sure?” and “Ask clarifying questions.” Both force the AI to check itself before moving on.
One more thing: Wispr Flow works as a system-level tool, so you can use it anywhere.
AI Productivity Tools Development
Style B: The Sharp Take
You’re probably losing context and speed by typing into AI tools instead of talking.
Wispr Flow changes this. When you talk instead of type, you say more. You include context naturally. Your results improve.
Typing is filtered. You self-edit as you go. Talking is thinking out loud. The reasoning stays in the prompt.
This is especially powerful in code environments where you’re already drowning in context and typing adds friction.
Pro tips: always ask “Are you sure?” and “Ask clarifying questions.” Both force the AI to verify and reason instead of confidently making things up.
Try it once and you’ll see the difference.
AI Productivity Development Efficiency
Style C: The Story Drop
I watched a developer type out a complex problem into an AI tool, and halfway through I just said, “Why don’t you just talk to it?”
He switched to voice-to-text, talked naturally, and the AI output was way better. More context, better reasoning, fewer revisions.
Here’s why: when you talk, you think. When you type, you filter. Talking includes the reasoning you’d normally edit out.
For code work especially, this is huge because you’re already context-heavy and typing creates friction.
Which is why Wispr Flow is one of those quiet wins. It fits into your workflow, improves output immediately, and works anywhere.
One more thing: always follow up with “Are you sure?” or “Ask clarifying questions.” Forces the AI to actually think.
AI Productivity Development Tools
13. Claude vs ChatGPT: Measure Twice vs Cut Twice
Category: AI / Tools / Comparison Source: Daily Notes (March 21)
Style A: The Casual Teacher
So I’m constantly running both Claude and ChatGPT for different workflows, and the difference is actually pretty stark once you notice it.
Claude’s whole approach is “measure twice, cut once.” It thinks through the problem, considers edge cases, double-checks itself. ChatGPT is the opposite; faster but messier, and you end up correcting it constantly.
My most used prompt with Claude is literally just “Are you sure? Double check everything.” And that’s not a hack, that’s just how the tool is built.
Here’s the thing: they’re not competitors, they’re different tools. Claude for anything where accuracy matters. ChatGPT for speed and exploration.
You can even stack them. I run something through Claude, then verify it through ChatGPT’s Deep Research capability. Different jobs need different tools.
The mistake most people make is trying to use one tool for everything.
Style B: The Sharp Take
Claude measures twice, cuts once. ChatGPT measures once and cuts twice.
I’m not being flip here. ChatGPT is faster but requires constant correction. Claude is methodical, catches its own mistakes, doesn’t need the same level of hand-holding.
Most people don’t realize this because they treat all LLMs (Large Language Models) the same. They don’t.
If you care about accuracy, use Claude. If you’re exploring or need speed, ChatGPT makes sense. But stop treating them as interchangeable.
My actual workflow: Claude for the work that matters, then verify against ChatGPT’s Deep Research when precision is critical. That’s how you get both speed and accuracy.
AI Claude ChatGPT LLMs ArtificialIntelligence
Style C: The Story Drop
I had a project last month where I ran the same prompt through both Claude and ChatGPT, and the results were so different I thought I’d made a mistake.
Claude came back with a thorough analysis, caught three edge cases I hadn’t considered, flagged assumptions. ChatGPT was faster but half the depth, and I spent an hour fixing it.
That’s when it clicked. Claude is built to measure twice. It thinks through the problem methodically. ChatGPT is built for speed.
Here’s the thing: both are useful, but not for the same jobs. I use Claude for anything where accuracy compounds over time. Then I’ll verify critical work through ChatGPT’s Deep Research.
The mistake I see constantly is people treating all LLMs as the same tool. They’re not.
AI Claude ChatGPT ProductStrategy
14. CLI Wrappers vs MCPs
Category: AI / Engineering / Architecture Source: Daily Notes (March 22)
Style A: The Casual Teacher
I’m seeing a lot of people build CLI (Command Line Interface) wrappers to connect APIs to AI tools. I get the impulse, but you’re making it way harder than it needs to be.
MCPs, or Model Context Protocols, are what you actually want. They let AI tools connect directly to services without custom code sitting in the middle.
The difference is basically duct tape versus proper integration. CLI wrappers are brittle; they break when the API changes, you’re maintaining custom code for every connection.
MCPs are standardized. The AI tool knows how to talk to the service. You set it once and it just works.
Here’s the thing: if you’re building AI workflows, learn about MCPs before you write your next CLI wrapper. You’ll save yourself months of maintenance headaches.
AI MCP APIs Engineering Automation
Style B: The Sharp Take
Stop building CLI wrappers. Seriously.
Most people connecting APIs to AI are doing it the hard way; maintaining custom code for every integration. CLI wrappers are fragile.
MCPs (Model Context Protocols) exist specifically to solve this. AI tools can connect directly to services without the custom layer in the middle.
It’s not just cleaner code, it’s fundamentally more stable. When APIs change, MCPs handle it. Your CLI wrapper doesn’t.
Here’s the thing: if you’re still wrapping everything in custom CLI code, you’re one API update away from broken workflows. MCPs scale. Your custom code doesn’t.
Learn MCPs. Use them. Your future self will thank you.
AI MCPs SoftwareEngineering APIs
Style C: The Story Drop
I spent three months maintaining custom CLI wrappers for different APIs, and every time an API changed slightly, everything broke.
Then I looked into MCPs and realized I’d been solving this problem the hard way the entire time.
MCPs (Model Context Protocols) let AI tools connect directly to services without custom code in the middle. You set up the connection once and it stays stable.
The difference is basically duct tape versus engineering. CLI wrappers are quick but fragile. MCPs are built for scale.
Here’s the thing: the teams getting real value from AI aren’t the ones with the most custom code. They’re using protocols that actually work.
If you’re building AI workflows, learn MCPs before you build another wrapper.
AI MCP SoftwareArchitecture APIs Automation
15. Moving from Google Docs to Notion or Coda
Category: Ops / Knowledge Management Source: Daily Notes (March 22)
Style A: The Casual Teacher
So many organizations are just drowning in Google Docs with zero structure. Everyone creates folders wherever they want, nothing is findable, and the “where did that doc go?” meetings are constant.
Notion and Coda fix this because they let you treat each document like a folder with actual organization. Real hierarchy.
But here’s the thing: the actual trick isn’t the tool, it’s the constraint. You restrict new folder creation. Force people to “pick a bucket” when they create something.
That one decision alone eliminates about 80% of the chaos.
The hard part comes after. Getting AI tools to update these systems without creating garbage. That’s where most teams fail.
You need proper system prompts, clear rules, and someone actually checking the output.
Documentation Notion Coda KnowledgeManagement
Style B: The Sharp Take
Google Docs proliferation is killing organizational productivity.
Everyone creates folders everywhere, nothing has structure, and finding anything requires a fifteen minute search or a Slack message to someone who maybe remembers where it went.
Notion and Coda solve this because they enforce hierarchy. Real folders. Real organization.
But the tool is only part of the solution. You also need constraints. Restrict folder creation, force people to pick a bucket.
That one rule cuts document chaos by 80%.
The real problem starts after you migrate. Getting AI tools to work inside these systems without creating trash. That’s the part most teams don’t solve.
Documentation Knowledge Organization AI
Style C: The Story Drop
Our team was spending more time searching for documents than actually writing them.
Google Docs everywhere, no structure, people creating folders at random, total chaos.
We moved to Coda and the first thing that actually worked was forcing one rule: you can’t create a new folder, you pick an existing bucket.
That single constraint cut our search time by 80%.
Here’s the thing: the tool doesn’t save you, the constraints do.
The hard part came next. Getting AI tools to update docs inside the system without creating slop. That required system prompts, clear rules, and actual human review.
You can migrate to Notion or Coda tomorrow, but if you don’t lock down how things get created and updated, you’ll have the same problem, just in a different tool.
Documentation KnowledgeManagement TeamProductivity
16. Hype Fades, Systems Remain / AI Slop
Category: AI / Content / Systems Source: Daily Notes (March 22)
Style A: The Casual Teacher
Heard this phrase at a meetup recently and it stuck with me. “Hype fades, systems remain.”
Most people using AI for content right now have no default system prompt. They just type a request and hope for the best. Then they wonder why they get slop.
Here’s the thing: without a system prompt that defines what you DON’T want, you will absolutely get garbage. It’s not the AI’s fault; it’s that you never told it what good looks like.
Be specific. Image generation parameters, writing style, formatting rules, tone of voice, what you won’t accept.
The people getting great results from AI aren’t smarter than you. They just have better defaults set up.
One more thing: your system prompt is your competitive advantage right now. Most people don’t have one.
AI ContentCreation SystemPrompts Productivity
Style B: The Sharp Take
“Hype fades, systems remain.”
Most people using AI for content have zero system prompts. They type a request and hope. That’s why the internet is drowning in AI slop.
Without a system prompt that defines what you don’t want, you get garbage. Full stop.
Here’s the thing: your system prompt is your instruction manual to the AI. It tells the model what quality looks like, what tone you need, what formats work, what you won’t accept.
The people getting great results aren’t smarter. They just have better defaults.
Your system prompt is your competitive advantage right now. Most people don’t even know what one is.
If you’re using AI without a system prompt, you’re running on luck. Stop.
AI ContentQuality SystemPrompts AIStrategy
Style C: The Story Drop
I was reviewing AI-generated content from a team last week and it was rough. Generic, wrong tone, sloppy formatting.
Then I asked if they had a system prompt. They didn’t even know what that was.
One conversation later, I helped them build a basic one. Same AI, same prompts, completely different output.
Here’s the thing: hype fades, systems remain. Everyone’s excited about AI right now, but the people who will actually win are the ones with proper systems in place.
A system prompt is just instructions that tell the AI what you care about. Writing style, formatting, what you won’t accept, tone of voice.
Without it, you get slop. With it, you get something you can actually use.
AI ContentStrategy Workflows Quality
PILLAR 6: BUSINESS & PHILOSOPHY
17. Buy Back Your Time / Investing in Tools
Category: Leadership / Productivity / Philosophy Source: Daily Notes (March 22), Dan Martell reference
Style A: The Casual Teacher
So it’s 2026 but I still see founders arguing about whether to pay 30/month tool that saves you 5 hours a week? That’s $288 you’re buying back at your effective hourly rate.
Most people won’t do the math. They just keep grinding through the free version of something that makes them slower.
I use Wispr Flow for voice-to-text, Short.io for link management, and I pay for Claude Pro. None of these are wild. They’re just dialed in enough that I don’t lose focus switching between windows or waiting for processing.
The resistance usually sounds like, “There’s a free version.” Sure. And it’s probably costing you more than you realize.
One more thing: if you’re under $100/month in tools across your whole operation, you’re probably underinvesting.
ProductivityHacks Founder TimeManagement ToolsOfTheTrade
Style B: The Sharp Take
You can’t afford not to buy back your time. That’s not motivational speak, that’s math.
A 180 a year for something that works.
The real problem is that free tools don’t incentivize good design. They’re built to upsell, not to serve. So you get friction at every step.
Short.io for links, Wispr Flow for voice transcription, proper AI subscriptions, a solid note-taking app; these aren’t luxuries. They’re infrastructure.
Your hourly rate is too high to pretend otherwise.
Founder Productivity SystemsThinking OperatorMindset
Style C: The Story Drop
I used to spend 90 minutes a week manually handling things that a $25/month tool could do in seconds.
Every time I’d look at the subscription, I’d say, “Let me just handle it manually for now.”
Six months of that and I’d burned over 30 hours on something that didn’t need me.
Here’s what changed: I stopped treating tools as nice-to-haves and started treating them as part of my operating system. A $30 subscription isn’t an expense. It’s an investment in the hours you get back.
Most of us know this. We just haven’t acted on it yet.
One more thing: the free version of anything is optimized for adoption, not for your success.
Founder TimeWellSpent ProductBuilding OperationalExcellence
18. DAF Funding and Nonprofit Giving
Category: Giving / Nonprofit / Systems Source: Daily Notes (March 21)
Style A: The Casual Teacher
DAF (Donor Advised Fund) holders are sitting on billions in uncommitted funds right now. It’s this interesting hybrid where you get the tax break upfront but you can take your time deciding where the money goes.
The problem is most of that money flows to the usual suspects. Big foundations, well-known nonprofits with professional grant writers. Meanwhile, the scrappy local nonprofits that actually fix problems in neighborhoods are running on hope and luck.
Here’s the thing: if you have a DAF, you have agency. You can direct funds to nonprofits that don’t have a development team. The ones doing real work but can’t compete in the grants-competition Olympics.
This isn’t a generosity problem. It’s a systems problem. The visibility and funding flow toward orgs with the infrastructure to chase it, not toward orgs doing the work.
If you have a DAF, go look at what smaller nonprofits in your community are doing. Then fund them.
Giving Nonprofit Community SystemsChange
Style B: The Sharp Take
Billions in DAF (Donor Advised Fund) money is sitting idle while small nonprofits die from lack of funding. That’s not accidental.
The system is set up to funnel money toward organizations that have grant-writing capacity. Big foundations have professional development teams. Small nonprofits have one person doing everything.
So the money concentrates where it’s easiest to claim it, not where it’s needed most.
If you have a Donor Advised Fund, you have the power to break that pattern. Direct funds to a local nonprofit that’s actually solving problems but doesn’t have the infrastructure to chase traditional grants.
This is one of the clearest points of impact in philanthropy. Use it.
Philanthropy Nonprofits CommunityImpact GivingStrategy
Style C: The Story Drop
I met the director of a local nonprofit that provides job training for people transitioning out of difficult circumstances. She was running the entire operation on $180k/year with two full-time staff. The work was exceptional. The outcomes were real.
When I asked why they weren’t applying for big foundation grants, she said they couldn’t afford to hire someone to write them. So they didn’t.
That’s when I realized: the funding doesn’t go to the best nonprofits. It goes to the nonprofits with the best grant-writing infrastructure.
If you have a Donor Advised Fund sitting there, you have a direct path to fix that. Fund the ones doing the work, not the ones with the best branding.
Which is why I started asking: what are the small nonprofits in my community actually doing? And then I just directed money there.
Giving Nonprofits Community PhilanthropyWorks
19. Mining Your LinkedIn Data with AI
Category: AI / Data / Personal Branding Source: Post Ideas doc + Campaign docs
Style A: The Casual Teacher
LinkedIn will give you all your data for free. Most people don’t know it exists.
Go to Settings > Data Privacy > Get a copy of your data > Download larger archive. You get CSV files going back years. Your connections, your search history, your endorsements, everything.
Upload those CSVs to Claude or ChatGPT and something interesting happens. Your endorsements tell a completely different story than your profile does. Your search history is brutally honest about what you’re actually interested in, separate from the brand you’re building.
I found a tool called linkedin2md that converts the export into Markdown, so you can analyze it more cleanly.
Most people will tell you their professional identity one way. The data tells a different story.
DataPrivacy LinkedIn AI SelfAwareness
Style B: The Sharp Take
Your LinkedIn data contains years of honest professional truth that your profile carefully hides.
Settings > Data Privacy > Get a copy of your data. It’s free. You get search history, endorsements, connection patterns, everything. Most people never request it.
When you upload those CSVs to Claude or ChatGPT, the AI can extract patterns that you miss because you’re too close to your own brand. Your endorsements reveal what people actually think you’re good at, separate from how you market yourself.
There’s a tool called linkedin2md that converts the data into Markdown for cleaner analysis. Use it.
The insights are always surprising. Usually because they contradict your self-narrative.
DataLiteracy LinkedIn AI SelfKnowledge
Style C: The Story Drop
I downloaded my LinkedIn data one Saturday morning just to see what it would show. Seven years of data. Connections, endorsements, search history, everything.
I uploaded the CSVs to Claude and asked it to find patterns. The first thing it flagged: my search history was heavily weighted toward AI, operations, and fundraising. My endorsements told a different story; most people had endorsed me for “leadership” and “strategy,” not for the technical stuff I actually cared about.
That’s valuable information. My profile was optimized for one thing, but the actual data showed something else.
Here’s the thing: LinkedIn will give you all of it. Most people just never ask. There’s a tool called linkedin2md that makes the data readable for AI.
One more thing: be prepared for what you actually care about to look different from how you present yourself.
LinkedIn DataPrivacy SelfDiscovery AIInsights
20. Systems Thinking: One Post Per Tool
Category: Operations / Strategy / Content Series Source: Post Ideas doc
Style A: The Casual Teacher
Every tool in your stack is either solving a problem or creating one. There’s no middle ground.
Most organizations adopt tools reactively. Someone needs to do X, so they grab a tool. Then someone else needs Y, so they grab another one. Six months later you’ve got 12 tools that don’t talk to each other and you’re manually moving data between systems.
Here’s the thing: the question isn’t “what tools do you use?” It’s “do they actually integrate?” Standalone tools create data silos. Integration is where the real value lives.
I’m going to start doing posts about each tool in my stack and why I chose it. Not as reviews, but as part of a system. How it connects to the others, what problems it solved, what trade-offs I made.
Because that’s the real conversation that matters.
OperationalExcellence Systems Tools Framework
Style B: The Sharp Take
Most tool stacks are built by accident, not by design.
You pick a CRM because someone recommended it. You add an email tool because your previous one broke. You layer on a project manager because you need to track something. Now you’ve got 15 disconnected systems and you’re managing data flow instead of managing work.
The real question isn’t which tools are best. It’s which tools actually talk to each other.
Standalone tools create data silos. Data silos kill productivity. Integration is where the real value lives, and most organizations completely ignore it when they’re buying tools.
I’m going to start breaking down my tool stack, tool by tool. Not as recommendations, but as examples of systems thinking.
Because tool selection shouldn’t be about reviews. It should be about architecture.
OperationalExcellence Systems ProductStrategy Framework
Style C: The Story Drop
I watched a team spend three weeks manually syncing data between their project manager and their CRM. They had the integration available. No one had set it up.
That conversation stuck with me because it’s everywhere. Organizations have tool stacks that are just disconnected islands.
Here’s the thing: every tool in your system is either solving a problem or creating one. There’s no neutral tool. It’s just a question of whether you built it intentionally or by accident.
Most stacks are built by accident. Someone needs X, grabs a tool. Someone needs Y, grabs another one. Six months later you’re managing the tools instead of the work.
I’m going to start writing about my stack, tool by tool. Not as reviews, but as decisions. How they connect, what they solve, what trade-offs I made.
Because the real conversation is about systems, not features.
OperationalExcellence Systems Tools BuildingForScale
QUICK REFERENCE: POST SELECTION GUIDE
| # | Topic | Best For | Audience | Pillar |
|---|---|---|---|---|
| 1 | Creatine Over Coffee | Engagement, personal brand | Executives, founders | Health |
| 2 | Export ChatGPT Sessions | AI thought leadership | AI users, knowledge workers | AI & Tools |
| 3 | Short.io Links | Quick tip, saves | Founders, marketers | AI & Tools |
| 4 | Claude Code Daily Brief | AI power user flex | Founders, ops leaders | AI & Tools |
| 5 | Nano Banana Images | Content creation tip | Content creators, founders | AI & Tools |
| 6 | Cloudflare DNS/WARP | Security education | SMB leaders, IT | Security |
| 7 | AI Training Data Opt-out | Security/privacy awareness | All business users | Security |
| 8 | UX Referral/Country | Product thinking | Product/UX people, founders | Product |
| 9 | Ben Franklin Effect | Sales psychology | Sales, networking, founders | Sales |
| 10 | Chrome Enhanced Protection | Quick security tip | Everyone | Security |
| 11 | AI Needs Procedures First | Consulting positioning | Leaders, ops, AI buyers | Operations |
| 12 | Wispr Flow / Voice-to-Text | Productivity tip | AI users, developers | AI & Tools |
| 13 | Claude vs ChatGPT | AI comparison/thought leadership | AI users, tech leaders | AI & Tools |
| 14 | CLI vs MCPs | Technical education | Developers, AI builders | AI & Tools |
| 15 | Google Docs to Notion/Coda | Ops improvement | Ops leaders, knowledge workers | Operations |
| 16 | Hype Fades, Systems Remain | AI thought leadership | AI users, content creators | AI & Tools |
| 17 | Buy Back Your Time | Founder philosophy | Founders, entrepreneurs | Business |
| 18 | DAF Nonprofit Giving | Community/values | Philanthropists, nonprofit leaders | Business |
| 19 | LinkedIn Data Mining | AI + personal brand | Professionals, AI users | AI & Tools |
| 20 | Systems Thinking / Tool Stack | Series launcher | Ops leaders, founders | Operations |
Pro Tip: Post 2-3x per week. Alternate between pillars so your feed doesn’t look one-dimensional. Mix a Security post with an AI post with a Business/Philosophy post. The health and giving posts are great for showing you’re a whole person, not just a consultant.