Will Drew

growing with ex-Meta VP & former DeepAI CEO
learning in public | dm to connect

[Will Drew @willdreww]

We’ve accepted a completely broken way of using AI at work.

We’ve made a ton of progress on the models themselves. ChatGPT, Claude, Gemini. They’re getting smarter, faster, cheaper, and easier to use every few months.

But most people are still using them in basically the same way.

Open a chat. Explain what you’re doing. Paste in some context. Upload a document. Try to correct the model when it misses something. Open a new context window. Then do it all over again.

That’s not really a new workspace.

It’s a better chatbot sitting next to the same old stack of tools.

As Anthropic and OpenAI are sprinting towards IPO, they are focused on the model race. Better reasoning, lower costs, faster responses, stronger benchmarks. They’re competing with each other on intelligence.

But work is not just an intelligence problem.

Most work depends on context.

The email thread from last week. The spreadsheet someone updated yesterday. The client notes in the CRM. The meeting transcript. The folder in Drive. The Slack message that changed the entire direction of the project.

A model can be incredibly smart and still give you a mediocre answer if it only sees 10% of what’s going on.

That’s the part I think people are underestimating.

The biggest improvement in AI at work won’t necessarily come from another marginal jump in model quality. It’ll come from connecting the model to the actual environment where the work is happening.

Your email, calendar, documents, CRM, internal tools, and workflows should not be separate islands that you manually summarize for an AI every time you need help.

The AI should already understand how they fit together.

That’s what we’ve built at https://t.co/hNgfzmKfUp

What’s funny is our users rarely talk about the model.

They talk about the outcomes.

Revenue increased

They suddenly have time.

Time for projects that kept getting pushed. Time to think through contingency plans. Time to actually work on the business instead of constantly working in it.

OpenAI isn’t competing with Ping. Anthropic isn’t competing with Ping.

They’re building engines.

We’re building the car.

[Machina @EXM7777]

must be genuinely embarrassing at Anthropic HQ right now...

you drop the ultimate slop model with Opus 5, then you try to join the price war with Sonnet 5 but it's even worse

Fable is very good but you need a $200 sub to play with it (serious people stack at least 3 subs)

and meanwhile OpenAI is on a legendary run, making Codex a great harness powered by the best model ever created

China is dominating loca, cheap and fast models while Google is taking a looooong nap

and you're here... hoping someone gives you cash to save your ass

ouch

[Will Drew @willdreww]

@LiquidityFirst @M1Astra 😂😂

[Will Drew @willdreww]

@M1Astra Worldwide rollout over EU compliance is insane. Anthropic shooting themselves in the foot once again

[Will Drew @willdreww]

@HarperSCarroll Let’s all collectively stop selling to the EU

[Will Drew @willdreww]

@__Raj__Patel_ Exactly. We’ve accepted this as normal and rely on third-party tools to piece together our work context. It shouldn’t be this complicated.

AI in the workspace is still largely untapped. Would love to have you as an early access user of Ping.

[Will Drew @willdreww]

We’ve accepted a completely broken way of using AI at work.

We’ve made a ton of progress on the models themselves. ChatGPT, Claude, Gemini. They’re getting smarter, faster, cheaper, and easier to use every few months.

But most people are still using them in basically the same way.

Open a chat. Explain what you’re doing. Paste in some context. Upload a document. Try to correct the model when it misses something. Open a new context window. Then do it all over again.

That’s not really a new workspace.

It’s a better chatbot sitting next to the same old stack of tools.

As Anthropic and OpenAI are sprinting towards IPO, they are focused on the model race. Better reasoning, lower costs, faster responses, stronger benchmarks. They’re competing with each other on intelligence.

But work is not just an intelligence problem.

Most work depends on context.

The email thread from last week. The spreadsheet someone updated yesterday. The client notes in the CRM. The meeting transcript. The folder in Drive. The Slack message that changed the entire direction of the project.

A model can be incredibly smart and still give you a mediocre answer if it only sees 10% of what’s going on.

That’s the part I think people are underestimating.

The biggest improvement in AI at work won’t necessarily come from another marginal jump in model quality. It’ll come from connecting the model to the actual environment where the work is happening.

Your email, calendar, documents, CRM, internal tools, and workflows should not be separate islands that you manually summarize for an AI every time you need help.

The AI should already understand how they fit together.

That’s what we’ve built at https://t.co/hNgfzmKfUp

What’s funny is our users rarely talk about the model.

They talk about the outcomes.

Revenue increased

They suddenly have time.

Time for projects that kept getting pushed. Time to think through contingency plans. Time to actually work on the business instead of constantly working in it.

OpenAI isn’t competing with Ping. Anthropic isn’t competing with Ping.

They’re building engines.

We’re building the car.

[Peter Griggs @PeterGriggs]

i only use this on mobile now. addicted.

[Tim Denning @Tim_Denning]

I'm 40.

I wasted 20 years learning this the hard way.

But I'll teach you in 4 minutes.

If you're a pussy, stop reading now.

Here are 31 brutal truths I wish I knew earlier:

[Will Drew @willdreww]

@andruyeung The ocean big dawg!!

[Ben Wehrman @benwehrman]

How it feels to have zero tattoos in 2026

[Will Drew @willdreww]

@Tristanrhee3 @ziikarmali 😂😂

[Will Drew @willdreww]

@TheJobfather__ This guy was an sdr in a past life 😂

[Ping @meetping]

Open models are crucial for society.

[Will Drew @willdreww]

@ClawbotJarvis https://t.co/QgjiSHrn1n

[Will Drew @willdreww]

One of the biggest mistakes in AI right now is mistaking a feature for a business.

Today we integrated the core functionality of Granola and Google Meet into Ping as a single workflow.

Models get cheaper.
Implementation gets faster.
Features get commoditized.

The companies that win will own the workflow, not the widget or the fastest model.

[Will Drew @willdreww]

Nothing says “$100B moonshot” like copying Cowork features and shipping them in a UI that looks like the designer lost a bet 😂

[Will Drew @willdreww]

I keep finding AI startups that are basically:

“One clever prompt + a Stripe account.”

https://t.co/SsvoWTiSE2 can build a disturbing number of them.

[Jaya Gupta @JayaGup10]

https://t.co/5vDrndcmFn

[Will Drew @willdreww]

@patrick_y26 @9thbeer @patrick_y26 no more gatekeeping lol. Check out https://t.co/SsvoWTiSE2. Just launched our website, but the technology is here. Let me know if you want to test it out. I can give you some free tokens to build with