Salesforce is going all in on AI agents. 

(Whoa!) 

Claude is going after enterprise customers, but likely missing the mark. 

(Yikes.) 

And Apple finally released its much-anticipated iPhone 16, with an on-device language model leading the way for Apple Intelligence. 

(Finally?!) 

It’s been a wild week. 

AI in the business world is moving faster than hot rolls in a buffet line. 

But we got you. 

Almost every Monday, we bring you the AI news that matters. 

Don’t waste your time wondering what’s going on. 

Let us do the worry for you and hit you with straight facts. 

Here’s what you need to know. 👇

After seemingly waiting like a decade for Apple to bring AI to all of its devices, we’re getting a bit closer. 

At its “Glow Time” event, Apple unveiled details behind the upcoming iPhone 16, its first entire iPhone lineup built for the company’s AI, called Apple Intelligence. 

(Eye roll. Lolz.) 

At the time of this typing, all the details are being gobbled up, as the announcement just wrapped minutes ago. 

Here’s the high-level: 

- iPhone 16 models will include a new small language model to incorporate edge AI, which is being powered by ChatGPT.

- All iPhones in the 16 lineup will include full support for Apple Intelligence

- New camera controls and visual intelligence for AI assistant

- New and improved Siri

We broke down the event in more detail in today’s AI in 5.

What it means: 

Apple's late to the AI game, but they're betting big that this will drive upgrades. 

It’ll be interesting to watch how Apple balances privacy with AI capabilities - it could set new industry standards. 

Don't rush to upgrade your company's iPhones yet, but start planning for how on-device AI could change your mobile strategy in the next 12-18 months.

Having a small/large language model on-device is now the defecto expectation moving forward. And that means billions of people (literally) are now going to be interacting with GenAI for the first time ever. 

In the same way that Apple’s offerings over the past decade have influenced the products and services enterprise companies offer, expect the Apple Intelligence rollout to be no different. 

Love em or hate em, Apple sets the bar on how society interacts with tech. And they’re legit going all in on AI. 

We can (almost) finally talk to our photos. 

Not in a weird way. 

Google's Ask Photos feature is getting slowly rolled out, Google now letting users chat with their photo library. 

You can ask things like "Where did we camp last time we went to Yosemite?" or "Show me pictures of my son as he's learned to walk over the last couple of months." 

It uses Google's Gemini AI models to understand and respond to your queries.

What it means: 

This is a major step in the shift to the concept of multimodality in language models. 

For the past year or two, we’ve largely thought of LLMs as input text, get text back. 

This lone feature from Google, though, can be a game-changer for personal data management. 

(Once/if they finally roll it out to all, that is. Lolz) 

If you're in any business involving large amounts of visual data - real estate, e-commerce, or content creation - start thinking about how this technology could streamline your workflows. 

For everyone else, consider how AI-powered search could be applied to your company's visual archives.

We see this especially useful for smaller and medium-sized businesses that don’t have a large data or AI team in-house. Being able to upload your company’s visuals into Google and have that as a working knowledge library could be a huge boon. 

(Again, if it all works as advertised.)

Everyone’s a coder now.

After a few weeks of Cursor AI dominating the AI-powered development spotlight, Replit has responded. 

Replit's new AI agent can build entire applications autonomously. It learns and adapts over time, improving its performance based on feedback. 

The standout feature is a bounty service where users can request software projects, providing real-world training data for the AI.

What it means: 

This could change software. For reals. 

We’ve talked about this before, but we see a future where everyday, non-technical people are building applications for their own use. 

Like no matter the problem, you’ll probably be prompting and “coding” a bespoke solution for yourself instantly. 

Think personalized software to solve small problems, all built and deployed with a simple text prompt. 

Replit’s launch here, coupled with Cursor’s recent success, also puts more pressure on the big players like Github Copilot, Tabnine and others to continue upping their games. 

For established companies, it's time to rethink your dev processes. 

Could AI agents handle your routine coding tasks, freeing up developers for more complex problems? 

Prolly. 

Start experimenting now, but be prepared for a learning curve as you figure out how to best collaborate with AI coders.

From concept to $1 billion raised in three months? 

Sheeesh. 

SSI, founded by former OpenAI co-founder Ilya Sutskever, just raised $1 billion to develop safe future AI systems that exceed human capabilities. 

They're focusing on two things: hiring top talent and securing compute power. 

With a $5 billion valuation after just three months, they're not messing around, y’all. 

What it means: 

Keep an eye on SSI.

No one is quite sure what to expect. 

Whatever they cook up, though, the world’s gonna pay attention. 

Sutskever is largely considered one of the brightest minds in AI so we expect massive developments out of SSI. 

AI safety is no longer just an ethical concern - it's a major business consideration. 

If you're developing AI products, expect increased scrutiny and potentially new regulations. 

(Prolly not so much in the U.S., but we expect regulations elsewhere to potentially be following any guidelines SSI might release.) 

Anthropic has officially launched Claude Enterprise, a business-focused tool designed with enhanced security and administrative controls. 

It features a native GitHub integration (fully rolling out later this fall) to boost programming tasks. 

Claude Enterprise comes with more robust security measures, including audit logs and role-based access with fine-grained permissions.

What it means: 

This is cool on the surface, but don’t jump in just yet if you’re a big business. 

We already blew the whistle on some concerns with Claude Enterprise last week when we covered it in-depth. 

Right now, it's best suited for development-heavy companies due to the GitHub integration. 

The major drawback? 

Claude is the only major large language model without access to real-time information. This is a deal-breaker for most businesses that need up-to-date data.

And the pricing model is apparently bespoke. Which we’re not fans of. 

LLMs are already largely considered a ‘black box’ of sorts, so their pricing should be much more defined and transparent.  

Not customized based on how your company will use it. 

If you're considering Claude Enterprise, first go listen/watch or read our full story on it.

Is one of the largest companies in the world going into startup mode? 

Seems like it. 

Salesforce CEO Marc Benioff is calling their move into AI agents a "hard pivot” from their traditional CRM dominance. 

They're launching Agent Force, a platform for creating autonomous AI agents for businesses. Benioff's even talking about "starting over as an AI agent startup" after 25 years in the CRM business.

What it means: 

If you're using Salesforce, big changes are coming. 

Start planning now for how AI agents could transform your customer service and sales processes.

 If you're not a Salesforce user, this move signals that AI is about to shake up CRM across the board.

(Hint: we’re gonna be talking a lot about agents and agentic workflows in the near future.) 

 It's time to audit your customer touch points and identify where AI agents could add value. 

But don't just chase the shiny new tech - make sure any AI integration aligns with your overall customer experience strategy.

This can’t be serious, right? 

Prolly not, but that didn’t stop everyone from talking about a recent report from The Information talking about potential skyrocketing OpenAI subscription costs. 

OpenAI is reportedly considering subscriptions up to $2000 per month for access to next-gen models like "Strawberry" and "Orion." 

(Shoutout to our audience for letting us know we were mispronouncing this! Lol) 

While not confirmed, it signals a potential shift in AI pricing and accessibility.

Like… big shift. 

What it means: 

If true, this creates a new tier of “ultra premium AI." 

For most businesses, this won't be necessary - focus on maximizing value from current models before chasing the bleeding edge. 

However, if you're in a field where AI gives a strong competitive advantage (like finance, R&D, consulting, market research, biotech, etc.), start budgeting now for potentially higher AI costs. 

Whether we see a 10X or 100X increase, it might be worth it. 

And like we talked about on today’s show, OpenAI is still the maker of the most powerful model in the world in GPT-4o. 

Wild part? 

The base GPT-4 model is now 1.5 years old. (Yes, it’s obviously received massive updates in GPT-4 Turbo and GPT-4o.) 

Whether it’s in 2 quarters or 2 years — if we see a future model from OpenAI (or others) that approaches AGI and human reasoning levels, a $200 or $2,000 price tag may not sound all that crazy. 

(Though we’d always expect a paid version to be more affordable to all.)

Refer Friends to Win a Year of ChatGPT!

(Or your fave LLM like Claude, Gemini, Copilot, etc)