Strawberry Whine (OpenAI Edition)

Bard.

Bing.

Claude.

Pi.

Socratic.

PerplexityAI.

Replika.

ChatSonic.

Cleo.

AkashChat.

Rasa.

I’ve tested out every single AI chatbot known.

In my humble opinion…

ChatGPT is still the king on most things you would use an AI chatbot for.

(Lately, I’ve been using it to help piece together Campbell family history.)

As such, I’m one of those keeping an eye on ChatGPT’s creators, OpenAI.

In short…

OpenAI is soon to roll out a new model - Strawberry - that “thinks before it speaks.”

Now, this model has received no shortage of hype.

One ex-OpenAI employee called it “a threat to humanity”.

BUT… we’ve seen this tactic before.

In 2019, the original ChatGPT-2 was too dangerous to share.

So what does it mean when AI says it “thinks before it speaks”?

Well, it could be one way to turn a bug into a feature.

In plain speak, it’ll take 10 to 20 seconds to “process” before spitting out an answer.

But this is a bigger deal than most think.

Now, why is this important?

Because Strawberry is aiming to tackle more complex problems—like math equations or coding tasks—without immediately giving you something half-baked.

Should I get excited?

Well, the early feedback is a mixed bag.

People are already - no surprise - whining about the wait time, saying the answers aren’t exactly lightyears ahead of what GPT-4 already offers.

For 95% of use cases, people can get by just fine with GPT-4o or its smaller cousin.

Strawberry’s real value shows up in that final stretch—maybe the top 3% to 4% of questions that need deeper reasoning.

But, again, the implications are deeper.

When?

The rumor mill says two weeks. No later than Sept 24.

(I’ll believe it when I see it.)

How much will it cost?

Strawberry will be integrated into ChatGPT, but it could also be available as a standalone service.

However…

It's expected to come with increased costs due to the higher computational demands.

Some say $2,000 per month.

Now, here’s where we dive into speculative territory. Especially if they plan to charge $2K per month.

It’s Already Happening

For the record, I doubt it will be that expensive…

BUT if I’m right about what I’m going to tell you…

I DO think future versions of it will cost that much. Maybe more.

Here’s why…

After following OpenAI’s work for months, I think Strawberry is a big deal. It’s not going to be a “better ChatGPT”.

It’s going to be something else entirely. 

Put simply, I think Strawberry is going to be the start of AI Agents.

AI Agents are often defined as: “self-governing systems capable of executing complex tasks autonomously and intelligently, without constant human input.”

Why do I think that? Easy.

For starters, OpenAI has hinted at moving beyond conversational AI towards agents that can operate independently.

Sam Altman and other OpenAI leaders have discussed how future AI models will become more autonomous, capable of handling tasks over time without constant human supervision.

These agents would be capable of:

• Transferring data between applications (e.g., from documents to spreadsheets)
• Automating multi-step processes like expense report filing
• Mimicking human actions such as clicking, typing, and navigating between app

Altman has also said that this will allow for integration of AI intelligence into virtually EVERY product and service.

We’re already seeing AI models deeply integrating with various platforms, automating entire workflows.

Take Microsoft’s integration of OpenAI models into their Office suite—AI now assists users in tasks like generating reports, summarizing data, and even managing emails autonomously within apps like Word and Excel.

This is the beginning of a larger trend where AI handles not just singular tasks but entire processes across multiple tools.

Nobody is Ready

Moreover, Strawberry could take a page from AI systems already acting as middlemen between various platforms, performing integrations and optimizations.

For example, Zapier automates tasks between apps like Slack, Google Sheets, and Salesforce, without requiring human involvement once the workflows are set up.

Imagine Strawberry going beyond automation—actually reasoning and planning within those systems to make complex decisions.

One thing I know…

The way you will use AI Agents in the future will be way different than a “chat” with a bot. It will be a hands-off assistant.

More like JARVIS and less like Siri. 

No, I don’t think OpenAI is set to release IronMan’s assistant in less than two weeks.

But I DO think this is the first big step in the rise of AI Agents.

And there’s going to be a LOT of ways to invest.