Clifton Mhlanga

Currently—Market Expansion & AE Dealhub | Founder | Building 👀| Check out my Newsletter Before the Numbers

[Clifton Mhlanga @Cliff305]

https://t.co/jRya01diQ0

[Clifton Mhlanga @Cliff305]

If you’re launching something this week, I’d probably study what Meta just did with Muse.

Not because everything Meta touches is suddenly genius. Please.

But the rollout has been very, very good.

First, they gave Muse a face.

That sounds like a tiny design decision until you actually use it. The character takes some of the coldness out of AI. You’re not just operating another piece of software. It feels a little more personal.

And Jessica Hische deserves credit here. The icon and logotype do a lot of work in making the whole thing feel approachable rather than technical.

Then there’s the app itself.

It’s almost suspiciously simple.

No settings maze. No enterprise dashboard trying to prove how powerful it is. You open it and pretty much understand what you’re meant to do.

And then there’s the cadence.

Muse launches.

A week later, Meta Connect gives everyone new things to talk about and new functions to try.

Then the beta opens so normal users can get early access to what comes next.

That’s the bit I really like.

Most launches are basically: big day, big spike, everyone posts screenshots, then we all move on to whatever launched on Tuesday.

Meta gave Muse multiple moments.

Each one gives people another reason to open it again, try something new, talk about it, and slowly build the habit.

Obviously the product still has to earn that habit from here. Marketing can get someone through the door. It cannot make them care six months later.

But as a rollout?

Very, very good.

Credit to @alexandr_wang , @_chriscox , the @Meta team, @finkd and everyone else behind the launch.

Whoever choreographed the sequence knew what they were doing.

[Clifton Mhlanga @Cliff305]

The iPhone Duo is incredible hardware. I’m just not convinced it solves a modern problem.

Foldables have had 7 years. They’re still 1.6% of smartphone shipments.

A bigger screen made sense when productivity meant more apps, tabs and windows.

In an agentic world, does it still?

[Clifton Mhlanga @Cliff305]

@OpenAI dropped Astra and computer use feels like it just levelled up.

So I’m curious.

Who is actually using computer use right now?

Not for demos.

For real work.

What tasks are you giving it?

And where does it still feel like:

“Please do not touch my laptop.”

[Clifton Mhlanga @Cliff305]

https://t.co/j2o3m9vurL

[Clifton Mhlanga @Cliff305]

Inspired by @MKBHD @Apple AI race.
The dangerous thing about Apple is that it is still winning.

Sounds ridiculous.

I know.

But their more to the story read it here. https://t.co/vwnnRiIJxA

[Clifton Mhlanga @Cliff305]

HubSpot Is Changing What Customers Pay For. https://t.co/hynLxWzYvp

[Clifton Mhlanga @Cliff305]

@Zoom is acquiring Common Room, which is………. very apropos.

A while back, I wrote about Zoom and what I called anchor drift, when a company moves away from the thing that made it undeniable.

Zoom’s Anchor Drift: How They Lost the Room https://t.co/DZGtX9ETam

[Clifton Mhlanga @Cliff305]

@clairevo If you need a guest, haha, I'm here. How I use AI in the GTM org

[Clifton Mhlanga @Cliff305]

Last week, hundreds of GTM operators showed up at @Anthropic event to learn how to run revenue workflows inside Claude.

Not random prompting.

Actual GTM work.

Today, I’m showing the exact setup I’ve been running:

How to set up Claude to support your revenue and GTM systems.

[Clifton Mhlanga @Cliff305]

@chalaska Right here 😆

[Clifton Mhlanga @Cliff305]

@yanndine GTM

[Clifton Mhlanga @Cliff305]

@CiprianiRanieri Cadence

[Clifton Mhlanga @Cliff305]

We all know what to do.

Train.
Eat well.
Stay consistent.

None of that is a mystery anymore.

The hard part isn’t knowledge.
It’s life.

Kids get sick.
Work runs late.
Plans change.
Energy drops.

And then you miss a session…
or eat something you didn’t plan…
and suddenly it feels like you’ve failed.

But here’s the truth:

You didn’t fail.
Life happened.

And maybe fitness was never supposed to be rigid in the first place.

Maybe the problem isn’t discipline.
Maybe the problem is the plan.

Because life isn’t predictable.
So why are fitness plans?

That’s the idea behind Cadence.

Not punishment.
Not perfection.

Adaptation.

Because progress shouldn’t depend on perfect weeks.

[Clifton Mhlanga @Cliff305]

Introducing Cadence.

A fitness coach designed around your life, not the other way around.

Because life doesn’t follow a perfect routine.
Kids get sick. Meetings run late. Weekends happen.

And most fitness apps?
They expect you to adapt to them.

Cadence adapts to you.

Missed a session?
It adjusts your week.

Unexpected dinner out?
It guides your next move.

Low energy today?
It shifts intensity, without breaking your progress.

No guilt.
No starting over.
Just momentum.

Cadence is built for real life.
And real results.

Coming soon.

[Clifton Mhlanga @Cliff305]

I built an AI prospecting playbook that runs daily and identifies accounts in active buying moments.

Not a prompt
A playbook, a folder of files that tells @AnthropicAI Claude exactly how to research, score, and prioritise accounts.

Here's how I structured it and what I learned.

Treat your file structure like a product.
Most people write one long prompt and hope for the best.

That's like giving someone a 20-page brief and asking them to remember everything at once 😅 .

Instead, I built a folder 👇

prospecting/
├── accounts/
│ └── target-accounts.xlsx
├── skills/
│ └── signal-scout/
│ ├── SKILL.md ← the playbook
│ ├── config.json ← my product, territory, competitors
│ └── references/
│ ├── signals.md
│ ├── public-company-signals.md
│ ├── review-mining.md
│ └── adaptation-guide.md
└── output/

SKILL.md is the playbook. It tells Claude what to do and in what order. Research the account.

The references folder holds the detailed knowledge.
Claude only loads what it needs when it needs it. That's the trick. Not everything at once.

----------------------------------------------------
You can't design what "good" looks like on paper.
You have to iterate.
This is the thing nobody tells you about building #AIplaybooks.
What I did was run it on accounts I already knew and had rich context on.
Then I compared the output against my own research.
Did it find what I would have found?
It missed things.
It didn't check how pricing had evolved over 12 months, just what it looked like today. I added a check for pricing model evolution. Compare now versus a year ago.

The delta tells the story.

It found leadership hires but didn't research what they were hired to do. A "new GM of Enterprise" is interesting. A "new GM hired to redesign pricing and packaging" is a buying signal.

So I added mandate depth research to ensure it matches what we solve for https://t.co/kIhnpM0Wol

Each gap became a fix.
Each fix made the next run better.
That's the entire game.

You don't design a perfect playbook. You build a decent one, run it, find the gaps, fix them, and run it again.
----------------------------------------------------
What it produces now 👇
Every morning, it deep-dives into 5 accounts.
Each account gets a dossier: a 12-month narrative, signal scores, business impact analysis, suggested contact, and an insight-led outreach angle.
----------------------------------------------------
If you're building your own:
Start with your file structure. That's the foundation.
One playbook file that orchestrates.
A reference folder for detailed knowledge.
An output folder for results.
Then run it on something you already know the answer to.

Compare. Find the gaps. Fix them. Run it again.
The AI does the research. You design the framework it follows.
That's the unlock.

#GTM #Sales #AI #Prospecting

[Clifton Mhlanga @Cliff305]

I built an AI prospecting playbook that runs daily and identifies accounts in active buying moments.

Not a prompt
A playbook, a folder of files that tells @AnthropicAI Claude exactly how to research, score, and prioritise accounts.

Here's how I structured it and what I learned.

Treat your file structure like a product.
Most people write one long prompt and hope for the best.

That's like giving someone a 20-page brief and asking them to remember everything at once 😅 .

Instead, I built a folder 👇

prospecting/
├── accounts/
│ └── target-accounts.xlsx
├── skills/
│ └── signal-scout/
│ ├── SKILL.md ��� the playbook
│ ├── config.json ← my product, territory, competitors
│ └── references/
│ ├── signals.md
│ ├── public-company-signals.md
│ ├── review-mining.md
│ └── adaptation-guide.md
└── output/

SKILL.md is the playbook. It tells Claude what to do and in what order. Research the account.

The references folder holds the detailed knowledge.
Claude only loads what it needs when it needs it. That's the trick. Not everything at once.

-------------------------------------------------------
You can't design what "good" looks like on paper.
You have to iterate.
This is the thing nobody tells you about building #AIplaybooks.
What I did was run it on accounts I already knew and had rich context on.
Then I compared the output against my own research.
Did it find what I would have found?
It missed things.
It didn't check how pricing had evolved over 12 months, just what it looked like today. I added a check for pricing model evolution. Compare now versus a year ago.

The delta tells the story.

It found leadership hires but didn't research what they were hired to do. A "new GM of Enterprise" is interesting. A "new GM hired to redesign pricing and packaging" is a buying signal.

So I added mandate depth research to ensure it matches what we solve for https://t.co/kIhnpM0Wol

Each gap became a fix.
Each fix made the next run better.
That's the entire game.

You don't design a perfect playbook. You build a decent one, run it, find the gaps, fix them, and run it again.
------------------------------------------------------------
What it produces now 👇
Every morning, it deep-dives into 5 accounts.
Each account gets a dossier: a 12-month narrative, signal scores, business impact analysis, suggested contact, and an insight-led outreach angle.
--------------------------------------------------------
If you're building your own:
Start with your file structure. That's the foundation.
One playbook file that orchestrates.
A reference folder for detailed knowledge.
An output folder for results.
Then run it on something you already know the answer to.

Compare. Find the gaps. Fix them. Run it again.
The AI does the research. You design the framework it follows.
That's the unlock.

#GTM #Sales #AI #Prospecting

[Clifton Mhlanga @Cliff305]

@yanndine GPT