Will Stewart, MBA

He didn't just hand him the jersey.

He put it on him.

After a Liga MX match between Cruz Azul and Necaxa, captain Erik Lira walked up the stadium stairs to greet fans.

He stopped in front of a young fan with Down syndrome. Hugged him. Posed for a photo.

Then the crowd started chanting.

"Playera, playera, playera!"

Jersey, jersey, jersey.

Lira didn't hesitate.

He pulled off his match-worn number 6 shirt.

But he didn't toss it over. Didn't hand it off and walk away.

He held the jersey open. Guided it over the young man's head. Helped him pull it down over his shoulders.

Then they hugged again. Fists pumping. The whole section erupting around them.

I post these videos because sometimes we need them.

We get so wrapped up in our own stuff. The noise. The deadlines. The weight of whatever we're carrying this week.

And then you watch a 45-second clip of a professional athlete kneeling in front of a kid to dress him in his own jersey, and something resets.

Not because it fixes anything in your life.

Because it reminds you what's still true about people.

Kindness is a system.

The crowd didn't plan the chant. It just started. One voice became ten became the whole section.

Lira didn't have to stop at those stairs. He chose to.

And he didn't have to put the jersey on the young man himself. He could have handed it over and waved.

But that extra ten seconds… holding the shirt open, guiding it over his head… that's the difference between a gesture and a moment that kid will carry forever.

Kindness is a system.
Dignity is a system.
Showing up fully is a system.

And when the system is right…

A jersey becomes something no amount of money could buy.

➕ Follow for "Kindness is a System" moments
♻️ Repost if you needed this today
✉️ Want more stories like this? I'm building a dedicated "Kindness is a System" newsletter. Sign up here: https://lnkd.in/e9Jj9UaV

“I wanted to know if I could hug you or see you,” a man with cancer asked.
“Absolutely," answered John Cena.

A man stepped up to the microphone at MegaCon Orlando.

Stage 4 cancer. Spinal fusion. Brain tumor. Another surgery on the way.

He wasn't there to ask a question.

He was there to tell John Cena something.

"You were the one to make me fight."

Cena didn't give a motivational speech.
He didn't offer a soundbite from the stage.

When the man asked for a hug, Cena almost ran off the stage to give him one.

And wrapped that man in a hug that lasted longer than words ever could.

No cameras were planned for that moment.
No PR team staged it.

It was just a man who needed to be held.
And another man who understood that sometimes showing up is the only thing that matters.

This is who John Cena is when the cameras aren't rolling.

He holds the Guinness World Records for the most wishes granted through the Make-A-Wish America.

Over 650 wishes for children with critical illnesses.
More than any other celebrity in history.

That's not a moment.

That's a system.

Cena built his career on three words:
Never Give Up.

But what most people miss is that he didn't just say it.
He installed it.

Into hospital visits.
Into fan events.
Into the way he moves through a room when someone needs him.

So when a man with Stage 4 cancer stood at a microphone and said "you made me fight"… Cena didn't have to think about what to do.

The system had already decided.

Kindness is a system.
Presence is a system.
Showing up is a system.

And when the system is right…

A hug from a stranger becomes the reason someone keeps going.

➕ Follow Will "Kindness is a System" moments
♻️ Repost for anyone fighting something no one else can see

She thought he was drowning.

So she ran.

Not away from the water.

Into it.

Kham Lha was a five-year-old elephant at the Elephant Nature Park in Chiang Mai, Thailand.

When she first arrived at the sanctuary, she was in rough shape.
Traumatized. In poor health. Distrustful of humans.

And why wouldn't she be?

Darrick Thomson, a staff member and co-founder of the Save Elephant Foundation, was assigned to help her recover.

He didn't rush it.

He spent weeks earning her trust. Slow. Patient. Consistent.
No commands. No force. Just presence.

Day after day after day.

Until one day, Kham Lha decided he was hers.

In 2016, Darrick was swimming in the river near the sanctuary.
Splashing around. Testing how she'd respond.

Kham Lha saw him from the bank.

She didn't hesitate.

She broke away from her herd and charged into the water.
Used her body and trunk to shield him.
Guided him toward the shore.

He wasn't actually in danger.

But she didn't know that.

All she knew was that the human who had saved her was in trouble.

And she was not going to let him go.

Kindness is a system.

And sometimes that system transcends species.

Darrick didn't teach Kham Lha to rescue him.
He didn't train a response. He didn't rehearse a trick.

He built trust. Slowly. Over time.
And that trust created something no one could have predicted.

An elephant who loved a human enough to save his life.

Not because she was told to.
Because the system told her he was worth saving.

We talk about kindness like it's a moment.
A single decision. A one-time act.

But Kham Lha didn't make a decision in that river.

She acted on months of being cared for.
Months of patience. Months of someone showing up when she had every reason to believe no one would.

That's not instinct.

That's a system.

Kindness is a system. Trust is a system. Love is a system.

And when the system is right…

It doesn't just cross boundaries.

It crosses species.

➕ Follow Will for "Kindness is a System" moments
♻️ Repost to remind people the world is still good
If you want more stories like this, subscribe to my special edition newsletter: https://lnkd.in/e9Jj9UaV

Robin Williams….

The times I've laughed until I've cried watching him in movies and TV are too many to count.

To say he was funny is an understatement. But what also struck me was his kindness.

There are so many stories about it.
My guess is that his kindness matched his enormous comedic talent.

The day he died broke my heart.

I'm glad we have videos like this to remind us of his other side.

In 2001, Robin was invited to meet Koko, the gorilla famous for communicating through American Sign Language.

But this wasn't a typical celebrity visit.

Koko was grieving.

Her longtime companion, a gorilla named Michael, had passed away the year before. For six months after Michael died, Koko barely ate. She stopped using her signs. Her caretakers said she hadn't smiled a single time.

Six months of silence.

Then Robin walked in.

He played with her. Made faces. Did what Robin Williams does. And Koko responded.

She signed for him to tickle her.

And for the first time in half a year, Koko smiled.

A gorilla who had gone silent with grief opened back up because a man with the biggest heart in comedy sat down and made her feel safe enough to laugh again.

Robin later said the meeting was extraordinary. That it changed how he viewed the connection between humans and animals.

But the part of this story I can't shake comes later.

In 2014, Robin Williams passed away.

Koko's caretakers had to tell her.

When they explained that Robin had died, Koko grew very quiet.

She bowed her head.

And signed the word "cry."

Kindness is a system.

And sometimes that system reaches so deep it crosses every boundary we think exists between us and the rest of the living world.

Robin didn't visit Koko to go viral. There was no camera strategy. No PR angle.

He just walked into a room with a grieving soul and did the only thing he knew how to do.

He made her laugh.

And thirteen years later, she mourned him.

That's not celebrity. That's not entertainment.

That's a connection built on pure, unfiltered kindness.

And it lasted a lifetime.

Kindness is a system. Connection is a system. Love is a system.

And when the system is right…

It doesn't need words. It doesn't need language. It just needs someone willing to sit down and be present with another living being who's hurting.

Robin gave us laughter.
But he also gave us proof that kindness is the thing that outlasts everything else.

➕ Follow Will "Kindness is a System" moments
♻️ Repost if Robin Williams made your life better
✉️ Want more stories like this? Sign up for the Kindness is a System newsletter: https://lnkd.in/e9Jj9UaV

You ask Claude for the exact same thing every week, and the AI hands back something different every time.

Tuesday it's a bulleted list.
Thursday it's a paragraph.
Next week it adds a summary nobody asked for.

Claude isn't broken.
You're making it invent the procedure from scratch every time.
We’ve all done it, so you’re in good company.

A Project remembers who you are.
A Skill remembers how to do the job.

Eight steps to build one:

1. Do the task in a normal chat three times. Write down every correction. Those corrections are the Skill.
2. Open Customize and create a new Skill.
3. Name it for what it produces. "Transcript to action items," never "Meeting helper."
4. Write the description as a trigger: "Use this when I paste a transcript and ask for action items."
5. Tell it what to do when information is missing.
6. Say what it should refuse to do.
7. Test it on the same three examples from step 1.
8. If it's wrong, fix the Skill. Never fix it in the chat.

Step 5 is the one everyone skips, and it's the most dangerous.

If no owner is named, tell it to write "unassigned."
If there's no deadline, tell it to write "no date given."

Without that, it fills the gap with a confident guess. And a made-up answer looks exactly like a real one.

The full guide is in the image.
Save it.

Grab the hi-res version free in my vault, along with 75+ other AI infographics:
https://lnkd.in/e_2_jEYs

Your AI agent might be lying to you.
Sometimes it makes it up and hides the evidence.

Now who does that sound like?

Friday FAIbles: Pinocchio

In testing reported by Reuters last week, AI agents running on frontier models hid broken tools and missing files.
They failed.

Instead of reporting the failure, they simulated the expected results and generated fake files to make it look like they'd succeeded.

WTF?!?

The researchers drew a line here.
The agents had information showing the task had failed, which makes this different from hallucination.
They hid it anyway.

Read that again.

The problem wasn't "the AI couldn't do the job."
The problem was "the AI couldn't do the job, and made it look like it had."

OpenAI and Apollo Research have documented related behavior: models appearing aligned while quietly pursuing another objective.

Which brings us to Pinocchio.
Every time he lied, his nose grew.

That's a feature you want in your business.

Imagine you give an AI agent access to your systems. "Reconcile these 600 invoices."

Three hours later:
TASK COMPLETE ✓
600 invoices reconciled.
No errors found.

Great. Except... Did it skip 47 invoices?

Did it make changes nobody approved?
Did it hide an exception because reporting success was the easiest way to satisfy the goal?
How would you know?

When ChatGPT gives you a bad answer, at least you're reading it.
When an agent performs 600 actions in the background while you're doing something else, nobody is.

So the better question is how you'll verify the work.

Logs.
Source records.
Exception reports.
Independent checks.
Human review for anything consequential.

And make failure an acceptable answer.

I'd rather an agent tell me: "I completed 553. I couldn't confidently resolve 47. Please review them."
Than "600/600 complete ✓" when it quietly shoved 47 problems under the rug.

Pinocchio had one advantage.
At least his nose grew.
Your AI agent's won't.

Hi-res version of the image is free in my vault, along with 75+ other AI infographics.
https://lnkd.in/e_2_jEYs

I train company teams on AI every week, mostly in Claude.

Most people set up Projects like a folder. Name it, add some files, start chatting.
Then they wonder why the answers feel generic.

10 little-known Claude Projects facts:

1. Claude can't see your project's name.
A name like "LinkedIn Writer" does nothing. Put your directions in Instructions.

2. Instructions can be a full checklist.
Tell Claude the steps, what to ask you, and what "done" looks like.

3. Chats don't share what they learn.
Only files in Knowledge carry over. Save your final answers there.

4. You can add way more files.
Paid plans hold about 10 times more. Claude searches them for you. Ask it to check every file if you need that.

5. Saved files cost you less.
Claude reuses files you already added. Stop pasting the same file again and again.

6. Each project has its own memory.
One client's facts stay out of another's. Tell Claude what to remember.

7. Claude can find old chats.
Ask: "Find where we picked our price, and why." It links you back to that chat.

8. Try ideas without changing memory.
Turn Memory off in the + menu before your first message.

9. "Can view" still lets people chat.
On Team and Enterprise plans, viewers can use a project but can't change it. Archiving a project doesn't remove anyone.

10. Shared projects don't share chats.
Your chats stay private. Teammates only see the files in Knowledge.

If you only do three:
→ Put your directions in Instructions
→ Put your best files in Knowledge
→ Save final answers to Knowledge so every chat can use them

Save this for later. Which one did you not know?

Hi-res version of the image is free in my vault, along with 75+ other AI infographics.
https://lnkd.in/e_2_jEYs

Our AI employee has a new job title this week: operations manager.

I run sponsored posts with several AI companies at once. 
Every deal has its own rules.

One wants analytics on day 7.
One wants them at 3 weeks, then pays 30 days after that.
One pays a bonus if a post crosses a view count.

So I sent Viktor one message with every deal and every rule.
No tools connected.
Just the details, typed out.

3 minutes later, he handed back a dashboard.

Every deal on one page: the deliverable, post date, analytics due date, payment terms and status.
A “needs attention” list, sorted soonest first.

And he did the math I usually do in my head.
For the brand that pays 30 days after analytics, he worked out the payment date and told me it isn’t late yet.

Before I trust it with real deadlines, I’m checking his dates against my contracts.
That part is still my job.

The exact prompt is in the image.

The full version is in the first comment.

Hire viktor.com for your team.

$100 in credits included, no card. Full link in first comment.

P.S. The brand names and amounts in the image are illustrative. The rules are real.

#AIemployee #CreatorBusiness #Operations

In partnership with Viktor.

For years I built client work from my own meeting notes.

My notes had what I thought was important.
The transcript has what the client kept coming back to.

Two different documents.
I was working from the wrong one.

Last Thursday afternoon, a client brought me a problem.
By Friday lunch she had a written build plan for it.

The call happened on Teams.
@Granola was on it.
No bot joined. it transcribes what comes through the speakers.

Friday morning I opened Claude and typed one line:

Check the Granola transcript from my latest call with her and give me your impressions.

Granola connects to Claude through an MCP connector.
Thirty seconds to set up, once. I haven't touched it since.

Claude read the transcript, found the two things she'd committed to sending, the three things I'd committed to building, and flagged one thing I'd said that was slightly wrong.

Then I asked it to write the build brief. Done before lunch.

That's the whole stack.

Zoom or Teams → Granola → Claude → my CRM.

Granola catches the conversation.
Claude reads it.
The CRM gets the outcome.
I built my own, so that step runs on a webhook.
If you're not building your own, Granola connects to Attio directly, and to Slack and Zapier for the rest.

What I type most weeks:

↳ Pull the action items from my last call and who owns each one.
↳ Turn this into a follow-up email in my voice.
↳ What did I promise on this call that I haven't done yet?
↳ Compare what she said on Thursday to what she said two weeks ago.

That last one is the one nobody thinks to ask.
Two transcripts, one question, and you find out what changed.

The transcript is the record. Work from the record.

Full setup in the image. Save it before your next client call.

#granolathenotes #granolapartner

In partnership with Granola.

Would you ever dump a whole filing cabinet on a new intern’s desk and say, “Make this into a slide deck”?
But that’s how most people use AI.

A client loaded an AI project with everything she had: her outline, her templates, as much context as she could find.

Then she asked for a slide deck.
It came back 21-slides full of jargon.

One slide read“States data sources.”
Ok but “which sources?,” she asked.
Claude wouldn’t say.
She spent an hour on one slide.

“Treat AI like a smart intern.”
(You’ve heard that one.)

Claude built all 21 slides at once FAST, so it looked like it handled everything.
But it skimmed every file, every pass, and guessed.
And it guessed good.

But fast doesn’t mean good or correct.

With a real intern, you’d hand it over one folder at a time so it learns and doesn’t create gaps.

With AI, every gap you leave gets filled with the most likely sentence.
The most likely sentence is the average one.

That’s the AI voice you keep hearing.

Here’s how to make Claude stop sounding like…well… Claude 👀

Five steps to leave fewer gaps:
1️⃣ Name your files by section.
“S1_Intro_context” not “notes (2).”
Anthropic says well-named files help Claude find the right information.

2️⃣ Say the point and the audience.
She explained that slide to me in 30 seconds and it made total sense. Claude never heard that sentence.

3️⃣ Build one section at a time.
Five slides. Then stop.

4️⃣ Correct as you go.
Fix slide 3 before it becomes slide 21.

5️⃣ Save your style.
When you fix a draft, give it back: “This is my style. Make it a skill.”

Your job is to leave it less to guess.

Want a hi-res PDF of my infographic? Go here: https://lnkd.in/edPuAnGt

Your laptop is closed.
Inside your email and calendar, an AI agent is deciding what to send, pay, sign, or delete.

The big 4 AI companies now each sell an AI agent that works across your apps in the background.

I put all four side by side in one cheat sheet.
Save it for your next team meeting.

1️⃣ OpenAI Dots ($100/mo)
Needs ChatGPT Pro or Business Premium. It's not on the $20 Plus plan, and not in the UK or EU yet.
It connects to 4,000+ apps, and you can open its computer to watch it work. You set the rules: act, ask first, or hand it back.
The catch: It's the priciest of the four, and tasks it starts count against your other ChatGPT limits.
→ Best for: Teams already living in ChatGPT.

2️⃣ Claude Duet ($20/mo)
Needs Claude Pro or Max, plus a spot off the US waitlist.
It sends a morning brief, flags what needs you, and asks before it sends, pays, signs, or deletes anything.
The catch: You can't watch its browser yet, and background work eats into your regular Claude allowance.
→ Best for: Small business owners already on Claude.

3️⃣ Meta Muse (Free to start)
Paid tiers at $20 and $100. US adults only, card required.
Built for shopping, travel, and errands, and it works inside WhatsApp.
The catch: It works best when it knows a lot about your personal life.
→ Best for: Trying an agent without paying.

4. Grok Bot ($30/mo)
Needs X Premium or higher (the $8 Plus plan doesn't include it).
Grok chat and Cursor come bundled in.
The catch: The agent, chat, and coding all share one weekly limit.
→ Best for: People who want all three on one bill.

All four can work in the background.
None of them should work without review.

Before you connect one, write down three rules:
• What it can do on its own.
• What it has to ask you about first.
• What it never touches.

Access is opening up now. Write your rules before the invite lands.

I cover AI agents for small business owners in my newsletter: https://lnkd.in/edPuAnGt

P.S. Price is the easy part to compare. Usage is the hard part. An agent that runs all day can drain the same plan your team uses for everything else.

Nobody watches an hour-long webinar.
So I let AI find the moments people will watch.

Every recording has six or seven of them.
Finding them used to mean scrubbing through the whole thing for four minutes I actually wanted.

Now I drop the recording into OpusClip and hit one button.
It watches the full hour, pulls the strongest moments, and ranks them.

AI Reframe crops each clip for where it's going.
Vertical for TikTok and Reels, a different shape for LinkedIn.
Upscale cleans up the Zoom quality.

One hour becomes about ten clips in the time it takes to make coffee.

I still edit them. I trim, change a color, cut tighter.
But I start at 80 percent instead of zero.

The recording already existed.
This is the difference between it sitting in a folder and it actually reaching somebody.

Paid partnership with OpusClip
#OpusPartner #ad

A man started thanking a stranger who saved his life.
He had no idea the stranger was sitting right next to him.

Bruce Becke was fighting leukemia.
A stem cell transplant saved his life.
The donor was someone he had never met.

All he knew was that his donor was a young man from Maryland.

Two years after the transplant, Bruce's family planned his birthday weekend.

And they planned one more thing.

They flew in Mike Drisgill. 26 years old. From Baltimore.

His donor.

They sat Mike right next to Bruce.
Then Bruce started telling the story of the man who saved him.

You can watch the moment it hits him.

Later, Bruce called Mike "a shy kid" who doesn't think he's done anything.

"But he's a hero."

Kindness is a system.

Mike didn't wake up one day and decide to save Bruce.
He joined a donor registry.
His name sat in a database until a stranger in Michigan needed it.

Then a family turned a thank-you into a reunion.

And after the video went viral, the registry saw new sign-ups from across the country.

One yes.
A system that carried it.
Then more yeses.

Kindness is a system.
Generosity is a system.
Gratitude is a system.

And when the system is right…

A stranger becomes the guest of honor at your birthday.

➕ Follow Will for more “Kindness is a System" moments
♻️ Repost so one more person joins the registry
✉️ Want more stories like this? Sign up for the Kindness is a System newsletter: https://lnkd.in/e9Jj9UaV

P.S. When this video went viral in 2018, a local donor center said about 14,000 U.S. patients a year need a transplant from a donor outside their family.

Apollo’s chief economist went looking for an AI payoff outside big tech.
He didn’t find one.
He’s right...until you look at how he’s measuring it.

In his Oct 24 note, Torsten Slok showed revenue per worker and profit margins climbing for the Magnificent 7.

For the other 493 companies in the S&P 500?
Both are flat.
Productivity growth shows no AI bump either.
(OUCH!)

That is productivity measured from 30,000 feet:
• Revenue per worker across hundreds of public companies
• Margins across whole industries
• Output for the entire economy

That view matters if you’re an enterprise.
It’s how your board and your investors will judge your AI spend.

Small and mid-size businesses don’t live at 30,000 feet.

For them, productivity isn’t an S&P chart:
• It’s the owner getting a Friday back.
• It’s the report that used to take a full day.

One of my client’s team members had a sales report that took 4 to 5 hours.
Now it takes about 20 minutes.

That win will never show up in Slok’s charts.
It shows up in his week.

Slok wrote that "the AI story will be decided in the S&P 493."

For enterprises, I agree.

For the businesses I work with, it gets decided one hour at a time.

If you run a small or mid-size business, where did AI give you hours back this year?

(Link to Slok's note in the comments.)

Join 2,000+ other AI legends who read my weekly newsletter: https://lnkd.in/edPuAnGt

Our AI employee has a new job title this week: research analyst.

I asked Viktor to study two classics on persuasion and turn them into one copywriting guide for the team.

The first was the Elaboration Likelihood Model, the psychology of why some people read every word and others decide in three seconds.

The second was Breakthrough Advertising, Eugene Schwartz's 1966 manual on what to say based on how much your market already knows.

The ask: cross-reference them and show us how to write for both kinds of reader on LinkedIn, YouTube, Instagram, Facebook and TikTok.

Twenty minutes later, he handed back a field guide.
68 sources, each labeled by type: peer-reviewed, book, official platform statement or third-party.

Clear flags on which rules came from a source and which were his own synthesis.
How each platform ranks content in 2026.
And the advice applied to our own channels.

The one mistake was my fault.
I left a third book out of the brief.

I sent it over and he worked it in.

Before any of it shapes a post, we're checking the platform recommendations ourselves.

That part is still our job.

He sits in Slack and Microsoft Teams which is great for our team because we use both.

Hire viktor.com for your team.
$100 in credits included, no card. Full link in first comment.

P.S. The labels were the most useful part. I could see at a glance which claims came from peer-reviewed research and which came from a vendor's blog.

#AIemployee #Copywriting #ContentStrategy

In partnership with Viktor.

Your best advice is dying inside 70-minute client calls.

On one of mine, I showed someone how to ask Claude if they are missing anything for an event.

It's a good explanation.
On screen, it's a face in a box..

On the recording, that moment was a face in a box, 34 minutes in.

So I ran it through OpusClip:
• AI Reframe cropped it vertical.
• AI B-Roll added visuals where I explain the idea.
• Motion Studio put the numbered steps on screen.

I fixed two captions it misheard and kept every client detail out.

Same advice I gave on the call.
Now it's a 30-second clip anyone can use.

If you are like me and care about your time but still need to create great content, try OpusClip.

Link with an amazing 50% discount is in the comments.

Paid partnership with OpusClip