CONSUMER

Can Siri's AI makeover on June 8 win back trust?

Two years after Apple first unveiled its vision for a generative AI-powered Siri at WWDC 2024, the long-awaited AI overhaul is expected to finally take center stage at this year's event on June 8. 

Siri will need to not only live up to expectations of transforming into an all-powerful assistant with full awareness of your personal context, but also deliver enough new and unique capabilities to justify the wait. According to Bloomberg, Apple plans to do just that.

As previously reported, Siri will feature a rebuilt model with Google Gemini technology, a standalone app, personal context and an AI-powered web search. Other features, as detailed in the Bloomberg report, are likely to include: 

• Third-party agents: Apple has been testing ways to allow third-party agents to operate in iOS 27 through the App Store. 

• New AI app interface: Apps from Claude, Gemini, and ChatGPT will operate within a home screen that includes previous conversations, voice, text and attachment options, no matter which model you used. 

• Smarter presentation: Siri will present text cards for topic prompts, understand your schedule, and write content based on web and device content. 

• Camera: Siri will become integrated as a dedicated mode in the camera app to replace the Visual Intelligence experience and have images analyzed by AI. The Photos app will now feature tools called Reframe and Extend to help with video editing. Natural language prompt-based editing is also being tested. 

• Shortcuts: The company is planning to revamp its Shortcuts app, which lets people create automations using natural language. There's an opportunity to transform this into a platform for creating agentic workflows.

The way users will access Siri is also different. Siri will live inside Dynamic Island as an always-on agent activated by either saying "Siri" or holding down the iPhone's power button as per usual, or an entirely new way where you could swipe down from the top center of the iPhone to launch a new search or ask interface.

There is a lot riding on this launch. It is likely Tim Cook's last big one as CEO before John Ternus takes over. The fall launch will likely include Apple's first push into a new category with foldables, and, most importantly, it is finally Apple's opportunity to show that the wait was worth it and earn back some credibility in the AI space.

There's no denying that Siri has trailed behind competitors for years, and I've personally been among those advocating for third-party AI assistants accessible via the Apple interface as a workaround. The updates coming at WWDC next week look promising enough that users may finally be able to rely on the native assistant, which would be huge. It would be far more convenient and offer fewer reasons for most iPhone users to reach for the ChatGPT, Claude, or Gemini apps. The option to integrate third-party agents and services is also a smart move, keeping users within the Siri interface even when running a different frontier model to accomplish some of the coolest workflows that today's AI can pull off.

TOGETHER WITH NOOKS

Your reps have 100s of accounts. AI helps them work the right 10.

Most sales teams run their book on instinct. Best guess on which accounts to prioritize. Hours of research before a single email. Discovery calls that never make it to the AE.

Nooks is the Agent Workspace where AI and reps work together to change that.

Ask Nooks which accounts to focus on this week. It pulls from your CRM, call recordings, and intent signals to tell you exactly where to start — and why. Ask it to write outbound that's personalized to what a prospect actually said on their last call. Ask it to turn a 30-minute discovery into a one-pager your AE can close with.

We process a million calls a day. Every one of them makes the intelligence sharper.

On June 4th, our own reps show you how they run their book live — no theory, just the actual workflow.

Save your seat

PRODUCTS

New Gusto tool makes agents usable for SMBs

Enterprises keep doubling down on AI. Small businesses, however, are still trying to figure out what to make of it. 

That's what led Eddie Kim, co-founder and CTO of payroll startup Gusto, to create a solution. It was one that took him and his team just nine weeks to ideate and build. On Tuesday, the company announced Gusto Cofounder, an AI agent purpose-built to connect a business's disparate platforms and handle tasks. 

“I think this is the most exciting thing we've launched since the launch of the company, personally,” Kim told The Deep View.

Cofounder, which Kim said is built on top of foundation models from OpenAI, allows businesses to delegate tasks such as handling payroll, generating reports and reviewing expenses to an agent using only plain-language prompts. Kim called this "the work before the work," or tasks that take away from a business owner’s time for their core product or passion. 

Gusto Cofounder also solves what Kim calls the "blank canvas problem." There’s no doubt that small businesses stand to benefit from AI. In a recent Gusto survey of 1,500 small business owners, a third said AI has already made it easier for them to start their businesses. But getting started, Kim said, can be the hardest part. For instance, these operators generally don’t have the time and resources to set up their own OpenClaws, and aren’t going to get significant returns just using chatbots. 

“I think, to be perfectly honest, a lot of the technology that powers Cofounder already exists today,” said Kim. “The thing that Gusto does uniquely is that we're starting from the problems that we're already solving for small businesses.” 

Here’s what it looks like in practice: 

• Because Cofounder already knows your context, such as team, payroll cadence, benefits and compliance calendar, the platform is useful from the first prompt. 

• The platform comes with 20 pre-built automations that help small business owners figure out where to start, and proactively surfaces risks, anomalies or upcoming documents that need to be filed. 

• Cofounder also integrates with all the other places where small businesses operate, such as Slack, Google Workspace and Notion.

• Additionally, Cofounder has multiple layers of built-in security. This includes the permissioning measures already built into Gusto’s flagship payroll product, as well as a consent framework that keeps business owners in the driver's seat.

“I think the future is that this is how small businesses are run,” said Kim. “Where you have this whole part of your business that you automated that Cofounder is running for you, and that frees you up for why you started your business in the first place.”

Small and medium-sized businesses represent a massively untapped market for AI. However, these businesses largely don’t have the budget, technical expertise, or time to embed AI in their operations. And Gusto isn’t the first company to consider this market. Anthropic launched its own Claude for Small Business tool targeting the same use cases. But Gusto, already being embedded within thousands of organizations, has a significant advantage in reaching those markets more natively. It’s the same reason Apple has the potential to sweep the consumer AI market. Though these firms aren’t making models themselves, they are meeting customers where they’re at and providing a seamless interface to the people who need to use the tech every day. 

TOGETHER WITH KOG

Inference this fast should require custom silicon. It doesn't.

Agentic software engineering is a sequential loop. Inspect, plan, edit, test, revise. Each step depends on the previous one. And decode speed per request sets the loop rate.

As agents become more autonomous, the productivity frontier shifts from intelligence alone to intelligence times iteration speed.

Kog's tech preview generates 3,300 output tok/s per request on standard datacenter GPUs, full precision, by co-designing the runtime, low-level GPU code, and model architecture as a single latency-optimized pipeline.

Try our coding model by yourself and read the full engineering breakdown on the Kog Labs blog.

HARDWARE

Nvidia solves key challenge in robotics

As the AI industry looks beyond language models, Nvidia is betting big on the buzzy new technology powering physical AI: world models. 

At Nvidia GTC Taipei at Computex, the company unveiled Cosmos 3, a new generalist world foundation model that it calls a "fully open omnimodel," capable of reasoning and generation across text, video, images, ambient sound and action. This iteration of the Cosmos world model family builds on a previous generations by providing improved generalization capabilities, which is a major barrier to physical AI development and deployment. 

"We wanted to build this Cosmo 3 model to help physical AI developers to build more generalizable physical AI models," Ming-Yu Liu, Nvidia's VP of Cosmos Labs, told The Deep View. 

Cosmos 3 debuts a number of world model innovations, Liu said: 

• The model utilizes a new architecture called "mixture-of-transformers," which combines the best aspects of two types of transformers: one for reasoning and one for generation. This enables it to understand object interactions, motion, and spatiotemporal relationships before generating video or action paths. 

• Cosmos 3 also doesn’t treat just one kind of data as a first-class citizen, said Liu. Instead, being omnimodal, it reasons with and generates "image, video, sound, and action, together with text," he said. 

• Additionally, Cosmos 3 is trained on one of the largest multimodal datasets for physical AI, spanning 20 trillion tokens, 1 billion images and 400 million authentic and synthetic videos.  

The model comes in several sizes: Super, the larger model for high-quality physics and accuracy, and Nano, for more efficient, quick generation needs, both of which are available now. Edge, which offers real-time inference for edge computing, will be available soon.

The models are also open-source, which Liu said offers developers more control and usability in physical AI development, a process that can be "challenging to do with API assets only." That allows enterprises to run them locally, customize them for their needs, and better control data security. 

Because the foundation models themselves are "just a starting point for physical AI developers," the goal is to integrate these models into ecosystems to provide a foundation for solving critical problems, he said. 

Cosmos 3 is just one step in the right direction in solving one of physical AI’s most pressing challenges. "We believe that the key problem to solve in physical AI is the generalization capability of the agent," Liu said. "To be clear, [Cosmos] is not yet solving the problem, but I think this architecture provides a great foundation to solve what I think is the holy grail in robotics." 

With Cosmos, Nvidia is feeding the open model ecosystem, both for the benefit of the ecosystem and for its own benefit. Along with providing the foundation for developers to create what Liu calls robotics’ "holy grail", any opportunity to feed a market that will inevitably demand more compute is an opportunity for Nvidia to make money in the end, as well as potentially make its own chips better through extreme hardware co-design. And while the benefits would extend back to Nvidia, a rising tide lifts all boats. As the industry broadly embraces the promise of physical AI, Nvidia's sharing of its resources and innovation will help stimulate further innovation. 

LINKS

• Anthropic confidentially files to go public to the SEC

• Nvidia launched NVIDIA RTX Spark, a new superchip that for Windows PCs

• Anthropic is giving the EU cybersecurity body access to Mythos 

• Nvidia, Meta and Schlumberger are among the top companies adopting AI, report says

• Runway, Nvidia announce Cosmos Coalition to build frontier world models for physical AI

• Microsoft expected to unveil new AI models, Copilot super app, Windows improvements

• OpenAI breaks ground on 1GW Stargate Michigan AI factory with host of social promises

• Perplexity's Search as Code (SaC): A "new reference search architecture" for agents

• Hy-Memory: Tencent Hy launched “a powerful memory plugin built specifically for long-term collaborative Agents like OpenClaw” 

• Claude: Long press to send is now here for dictation in Claude

• Lovable: automatically runs a security scan before you publish.

• OpenAI Robotics: full-stack hardware, ops, systems, and ML engineers

• Paramount: AI Platforms and Orchestration Lead

• Adobe: Principal Product Manager, Research and AI - Data 

• Boeing: Senior Artificial Intelligence Operations Specialist

GAMES

A QUICK POLL BEFORE YOU GO