Previously week, we’ve seen some big AI information tales break…
Right here’s what it’s good to learn about what is going on on:
Perplexity Launches AI-Powered Purchasing Expertise
Perplexity, the AI-powered search engine, simply launched an enormous new procuring function that permits you to analysis and purchase merchandise proper inside Perplexity.
The function known as Purchase with Professional, which lets Perplexity’s paid customers in america full purchases straight by way of the platform.
Right here’s the way it works:
If you ask Perplexity a query associated to procuring, you continue to get pure language responses like regular.
However you additionally could now see product playing cards that present essentially the most related objects accessible for buy and their product particulars.
Perplexity says these playing cards “aren’t sponsored—they’re unbiased suggestions, tailor-made to your search by our AI.”
If you see product you want, you should utilize one-click checkout proper in Perplexity to purchase it (should you’ve saved your transport and billing information with them) by clicking “Purchase with Professional.”
Proper now, you get free transport on all Purchase with Professional orders courtesy of Perplexity.
Microsoft Positions Copilot because the “UI for AI” at Ignite 2024
At its Ignite 2024 occasion, Microsoft doubled down on its AI technique, positioning Copilot because the default interface for AI interactions and hyping up the flexibility to create AI brokers in Copilot.
In his keynote on the occasion, CEO Satya Nadella stated Copilot is the “UI for AI.” To that finish, he made some main AI bulletins, together with:
Copilot Actions, that are customizable immediate templates that automate repetitive duties
New purpose-built AI brokers in Copilot that tackle specialised roles, together with an agent for HR and IT questions, one for assembly collaboration, and a challenge supervisor agent
Copilot Studio is getting enhanced with autonomous capabilities, permitting brokers to take actions within the background with out human prompting
For Groups customers, Copilot will quickly analyze screen-shared content material throughout conferences, offering insights from each the dialog and visible content material
And PowerPoint is getting new AI options, together with a Narrative Builder that may create shows from referenced paperwork, and the flexibility to translate total shows into 40 completely different languages whereas sustaining the design
The keynote concluded with Microsoft’s dedication to AI training, noting that they’ve helped prepare over 23 million folks in AI and digital expertise over the previous 12 months, with plans to succeed in thousands and thousands extra.
Google Gemini Will get Reminiscence and App Management
Google is rolling out two vital updates to Gemini.
First, a brand new reminiscence function permits the AI to recollect private data for extra contextual responses, in keeping with TechCrunch. (For example, Gemini would possibly bear in mind your favourite meals or particulars about your work, so it might reference that data when related and useful.)
The function, accessible solely to Google One AI Premium subscribers at $20 per 30 days, might be turned off at any time and at present works solely in English on the internet shopper.
Second, a report from The Verge signifies that Google could also be prepping Gemini to take motion inside apps.
Hidden in Android 16’s developer preview is a brand new “app capabilities” API that would give Gemini the flexibility to take direct actions inside apps.
This might imply, for instance, ordering meals by way of DoorDash with out ever opening the app—you’d simply inform Gemini what you need.
Second, hidden code in Android 16 suggests Gemini could quickly have the ability to take direct actions inside apps.
Meta Makes Main Enterprise AI Play
Meta has poached Salesforce’s CEO of AI, Clara Shih, to guide a brand new enterprise AI group.
With Meta’s Llama fashions seeing over 600 million downloads and Meta AI claiming 500 million month-to-month lively customers, the corporate seems poised for a major enterprise push.
Shih stated in a put up on X: “Our imaginative and prescient for this new product group is to make cutting-edge AI accessible to each enterprise, empowering all to search out success and personal their future within the AI period.”
U.S. Requires “Manhattan Challenge” Fashion AI Initiative
A serious U.S. congressional fee is asking for what could possibly be one of the formidable authorities AI tasks in historical past: a Manhattan Challenge-style initiative to develop synthetic common intelligence.
The U.S.-China Financial and Safety Evaluation Fee, a bipartisan group established by Congress in 2000, says America wants such a large-scale public-private partnership to remain aggressive with China within the race to develop AI programs that match or exceed human intelligence.
The historic parallel they’re drawing right here is important.
The unique Manhattan Challenge was a large collaboration between the U.S. authorities and personal sector throughout World Warfare Two that led to the event of the primary atomic bombs.
Whereas the fee is advocating for the same scale of effort, they have not but specified precisely how a lot cash needs to be invested or how the partnership could be structured.
One particular suggestion did emerge from Jacob Helberg, a commissioner and senior advisor to Palantir’s CEO.
He proposed streamlining the allowing course of for knowledge facilities, noting that power infrastructure is at present a serious bottleneck in coaching massive AI fashions.
Amazon Doubles Down on Anthropic with $4 Billion Funding
Amazon simply doubled down on its relationship with Anthropic by investing one other $4 billion within the firm (bringing its whole funding to $8 billion)—and what’s curious is that it’s doing so whereas additionally actively making strikes to cut back its reliance on the corporate.
Anthropic introduced on November 22 that that they had acquired one other $4 billion from Amazon and that AWS could be the corporate’s main cloud and coaching accomplice.
A key a part of this expanded partnership includes shut collaboration on AWS Trainium {hardware}.
AWS Trainium is a purpose-built machine studying accelerator that permits high-performance mannequin coaching.
Anthropic’s engineers are working straight with Amazon’s Annapurna Labs workforce “on the event and optimization of future generations of Trainium accelerators, advancing the capabilities of specialised machine studying {hardware}.”
Regardless of this deep collaboration, nevertheless, Amazon additionally seems to be attempting to actively scale back its reliance on Anthropic.
One other report from The Info reveals that Amazon has developed a brand new AI mannequin that may course of photos and video with a purpose to make it much less depending on Anthropic.
Says The Info:
“In growing the brand new mannequin, Amazon is exhibiting that it nonetheless hopes its internally developed AI can acquire traction amongst its cloud clients, making it much less depending on AI from Anthropic, whose Claude chatbot is now a well-liked providing on Amazon Net Companies.”
AI Agent Startup Raises Large $56M Seed Spherical
A brand new AI startup with an unconventional identify—/dev/brokers—simply secured a whopping $56 million seed spherical at a $500 million valuation. The corporate is constructing what it calls an working system for AI brokers, autonomous packages that may deal with complicated duties with out human supervision.
Led by former Stripe CTO David Singleton, and backed by a formidable workforce of ex-Android and Meta Oculus veterans, the corporate attracts an formidable parallel: Simply as Android created the inspiration for the cell revolution, /dev/brokers goals to construct the important platform for AI brokers to succeed in their full potential.
The startup has attracted high-profile buyers together with Andrej Karpathy and Scale AI’s Alexander Wang. Nevertheless, particulars in regards to the precise product stay scarce—their web site is remarkably minimal, paying homage to different current AI startups which have raised huge rounds with little public data.
Salesforce Brings AI Brokers to Slack
Salesforce is now including its AI brokers to Slack to behave as your digital coworkers proper inside Slack workspaces.
As a substitute of getting AI assistants work in isolation, Salesforce’s Agentforce platform offers them entry to a corporation’s Slack conversations and enterprise knowledge, permitting them to grasp context and take extra related actions straight within the circulate of labor.
What makes this notably fascinating is how Salesforce is positioning these AI brokers.
They are not simply passive assistants—they will actively counsel and execute actions on behalf of staff throughout completely different departments.
For example, HR brokers can deal with onboarding and advantages questions, IT brokers can resolve assist desk tickets, and gross sales brokers can put together government briefings and create proposals.
The corporate has already lined up some spectacular early adopters.
Accenture studies they’ve developed IT help brokers that deal with first-tier questions in Slack channels, whereas Field CEO Aaron Levie describes it as the conclusion of a decades-old dream in computing—having a single conversational interface the place people and AI work collectively seamlessly.
Salesforce says Agentforce will change into accessible by way of Slack, although particular pricing and availability particulars have not been introduced but.
Apple Preps AI-Powered Siri Improve
Apple is on the brink of launch an AI-powered model of Siri that it hopes might help it catch as much as OpenAI, Google, Microsoft, and others within the AI arms race.
Based on Bloomberg, Apple is growing what staff internally name “LLM Siri,” a serious improve that goals to rework the 13-year-old digital assistant into a way more conversational AI powered by massive language fashions.
The improve guarantees a number of key enhancements: extra human-like interactions, higher dealing with of complicated duties, expanded management of third-party apps, and integration with Apple Intelligence options like textual content era and summarization.
The brand new Siri can be constructed on Apple’s personal AI fashions and is at present being examined as a separate app on iPhones, iPads, and Macs.
The challenge additionally represents Apple’s bid to meet up with opponents like ChatGPT and Google’s Gemini.
Whereas Apple not too long ago launched its Apple Intelligence platform, the corporate continues to be trailing behind in lots of AI options supplied by different tech giants.
However, as Apple Intelligence taught us: Simply because we’re listening to about this doesn’t imply you’re getting it any time quickly.
The corporate plans to formally announce the updates as early as 2025 as a part of iOS 19 and macOS 16.
However customers will not really get to make use of these options till spring 2026, a few 12 months and a half from now.
Suno v4 Raises the Bar for AI Music
Fashionable AI music generator Suno simply launched a serious replace that’s getting a whole lot of consideration.
The brand new model of Suno, referred to as Suno v4, guarantees higher total audio high quality and extra subtle track buildings.
Moreover, the system can now apparently deal with extra complicated musical preparations and produce sharper, clearer lyrics.
The corporate has additionally launched some fascinating new options.
There is a “Remaster” operate that may improve present tracks to the brand new v4 high quality degree.
They’ve added a Cowl Artwork generator to create visuals that match your music’s fashion.
And there are two notably fascinating additions: a “Covers” function that may reimagine songs in numerous kinds, and “Personas” that allow customers preserve a constant musical id throughout a number of creations.
The replace is at present in beta and accessible to Suno’s Professional and Premier subscribers.
Eleven Labs Doubles Down on Voice AI
AI voice cloning firm ElevenLabs simply dropped two large updates: a platform for constructing conversational brokers and a product that turns written content material into podcasts.
First, ElevenLabs is rolling out instruments that allow builders create customizable conversational AI brokers on their platform.
These brokers might be fine-tuned throughout a number of variables—from tone of voice to response size—and may work with numerous language fashions together with OpenAI’s fashions, Google Gemini, and Claude from Anthropic.
Builders may even combine their very own customized language fashions and data bases.
What makes this notably fascinating is that ElevenLabs is leveraging their experience in voice know-how to create extra natural-sounding AI conversations.
The platform consists of options for dealing with real-world challenges like buyer interruptions and knowledge assortment throughout conversations.
Second, ElevenLabs has additionally launched GenFM, a artistic new service that transforms written content material into AI-hosted podcast discussions.
Obtainable by way of their ElevenReader app, it might take any PDF, article, or e book and switch it right into a podcast that includes two AI co-hosts discussing the content material in any of 32 languages.
Runway’s “Frames” Tackles AI Picture Consistency
Runway simply unveiled a brand new AI picture mannequin that claims to make main advances in controlling kinds throughout AI-generated imagery, which is a major limitation in at the moment’s AI picture era instruments.
The mannequin, which known as Frames, purportedly is ready to preserve constant visible kinds throughout a number of generations, which means you may generate many photos in the identical visible fashion so all of them look constant.
Runway has organized the mannequin’s capabilities across the idea of “worlds”—coherent visible kinds that may be constantly utilized throughout a number of generations.
Writes Runway:
“With Frames, you may start to architect worlds that symbolize very particular factors of view and aesthetic traits. The mannequin permits you to design with precision the look, really feel and environment of the world you need to create.”
They provide some examples of worlds they’ve created with the brand new mannequin, together with one which duplicates the aesthetic of Seventies album artwork and one other that duplicates the visible fashion of taking images on a disposable digital camera.
The know-how is being steadily rolled out by way of Runway’s Gen-3 Alpha platform and API.
