OpenAI executives have discussed filing an antitrust complaint with US regulators against Microsoft, the company’s largest investor, The Wall Street Journal reported Monday, marking a dramatic escalation in tensions between the two long-term AI partners. OpenAI, which develops ChatGPT, has reportedly considered seeking a federal regulatory review of the terms of its contract with Microsoft for potential antitrust law violations, according to people familiar with the matter.
The potential antitrust complaint would likely argue that Microsoft is using its dominant position in cloud services and contractual leverage to suppress competition, according to insiders who described it as a “nuclear option,” the WSJ reports.
The move could unravel one of the most important business partnerships in the AI industry—a relationship that started with a $1 billion investment by Microsoft in 2019 and has grown to include billions more in funding, along with Microsoft’s exclusive rights to host OpenAI models on its Azure cloud platform.
Imagine a tool that takes an image of whatever appears on your computer’s screen, saves it locally, and lets you access it all like a time machine. A magical looking glass for the computing past. That’s essentially what Microsoft’s Recall is all about. Yet, when it was first introduced, it stirred up a security storm.
Microsoft pulled its release plans, fortified the security guardrails, and relaunched it a few weeks ago. This time around, Recall got a minor-but-amazingly practical upgrade. The best part? Instead of having you scrub through a long timeline of pictures, you can simply search through the entire activity history with words.
All those perks come at a steep processing price, both in terms of system resources and AI chops. So much that the only machines that support Recall need a Copilot+ branding and a processor launched within the past year, or so. The experience, however, is shockingly good.
What is Recall?
Nadeem Sarwar / Digital Trends
In Microsoft’s words, Recall lets you “quickly find and jump back into what you have seen before on your PC.” But before any of that happens, do keep in mind that it’s an opt-in process, and when you enable it, you need to biometrically verify your identity (face unlock or fingerprint scan) to access the history.
Next, your PC must meet the system requirements, too. For recall, you need a Copilot+ PC that has a dedicated AI chip offering over 40 TOPS output, 16GB of RAM, 256GB storage (of which 50GB must be free at the very minimum), active device encryption, and Windows Hello sign-in enabled.
Recall is also tied deeply to your web browsing activity, and as such, it needs a compatible browser. Broadly, it works with Chromium-based options, such as Edge, Chrome, Opera, and Firefox. But why are browsers important when Recall is essentially saving your entire screen activity?
Nadeem Sarwar / Digital Trends
First, it doesn’t log screenshots of your incognito or private mode browsing. More importantly, you can limit it from saving screenshots of certain websites, using a system of filters. Think of your banking activities or secure communications.
Likewise, you can also exclude certain apps from getting visually cataloged into a digital ledger. Those exceptions are extremely important, especially for folks who think the idea of a “snapshot everything” tool is too risky.
For the sake of privacy, I disabled Recall for WhatsApp to protect my personal chats and Adobe Reader, because I regularly handle confidential papers. You can choose to pause/resume Recall’s snapshot gathering — briefly or permanently – from a dedicated shortcut in the system tray.
How does it help me?
Nadeem Sarwar / Digital Trends
Microsoft is selling the idea of Recall as a photographic memory for your PC. The idea is almost loathsome for any normal person who despises the concept of all their computing activity being recorded in the form of crisp snapshots. AI skeptics would absolutely run away from it.
I, a man of forgetful ways and critic of poorly-designed apps, love this. My daily routine entails reading and writing words. Lots and lots of them. From long articles in Docs and notes in a tiny scratchpad to detailed pitches in email and daily chat with work colleagues, most of my waking time revolves around words.
I believe I have a decent memory of events, but not so much with exact terms like “Embargoed until May 16, 6AM PT / 9AM CET” or “Microsoft meeting with John scheduled for April 9. Does 6AM ET or 5PM ET sound good?” It’s not scientifically possible either to remember it all, even though I have a vague recollection of the main event.
Nadeem Sarwar / Digital Trends
This is where Windows Recall and its image recognition come to the rescue in a very controlled fashion. I remember having a brief chat with a source named John, but couldn’t quite remember where the transcripts were saved. I went back to Recall, looked up the name, and found just what I was looking for.
From the looks of the snapshot, it was a browser tab, but I no longer had it active. This is where Recall offered two routes to recover it. First, it offered an outbound URL button to open the webpage in a browser. Alternatively, the Click To Do system lets me copy-paste the conversation directly from the saved snapshot.
Nadeem Sarwar / Digital Trends
It even pulled information from a diagnostic test and directly guided me to Microsoft’s official Get Help dashboard with a single click. That’s impressive stuff.
Depending on the content you have selected in a Recall snapshot, you will also see contextual actions such as summarizing or rewriting the text passage. You can directly choose to open it in Notepad, or pick from any app of your choice.
Nadeem Sarwar / Digital Trends
For example, I was recently looking up a recipe and forgot to bookmark it. With Recall, I recovered it and directly moved the wall of text to Copilot, where it was neatly rewritten across headings and bullet points, like a cookbook.
These impressive capabilities are driven by Optical Character Recognition (OCR), powered by the AI accelerator chip on a Windows machine. Everything you search in Recall is processed contextually, and the results you are shown are divided across text and images.
Nadeem Sarwar / Digital Trends
The safety protocols
Microsoft has also built a set of safety features around sensitive information. For example, it won’t save details such as license, credit card, account, personal identification, tax filing, citizen registration, and license numbers.
I tried two of my banking operators and a government website where my national identity card copies are accessible, and they didn’t appear in the Recall snapshots. The list covers a healthy bunch of documents used across North America, Europe, and Asian countries, but it’s not all-encompassing.
Nadeem Sarwar / Digital Trends
For the sake of added precaution, you can open such web portals in incognito mode or private windows. Alternatively, you can simply set those websites within Recall’s filters, and their snapshots will never be saved in the first place.
Furthermore, the Recall app will give you an option to delete all images belonging to a certain website or app, in one go. Finally, there is an auto-delete function, where you can set the erase all snapshots at a cadence of 30, 60, 90, and 180 days. Of course, you can wipe the slate clean at your own discretion.
Another huge respite for me? The poor search system in apps. From Slack to Google Drive, the search system in these platforms often leaves me frustrated. Plus, if you don’t pay for a premium, the free user limits mean you lose your chat data due to the auto-delete protocols.
Nadeem Sarwar / Digital Trends
Recall offers a fantastic way to not just overcome the poor search system in virtually any app that runs on your computer, but also does a much better job at it. It fared better at helping me find the right Docs file in a list of thousands that I have hoarded over the years and refuse to delete.
Contextual understanding is a huge benefit here. For example, if you work across different apps, the search results at the top will let you narrow down the results. For example, you can only look for an item in Outlook if a match is found in the email client, among a bunch of other apps. This approach saves a lot of time.
Why are experts spooked?
Security experts are skeptical about the idea, and rightfully so. Nick Hyatt, Director of Threat Intelligence at Blackpoint Cyber, told me that Recall is a great idea, but also an awful one for a lot of reasons.
Nadeem Sarwar / Digital Trends
He pointed out how the feature logs every single aspect of your digital life, from personal communications to files and what you look up on the web. The stakes are even higher for situations such as domestic abuse.
“A categorized, searchable database of every activity conducted by a user could literally result in people dying should attempts to get help be discovered,” Hyatt pointed out. Plus, the cyber threat ecosystem is an ever-evolving field, so the risks persist.
Nadeem Sarwar / Digital Trends
“Recall will be heavily targeted by attackers because of the goldmine of user data it contains,” Nate Warfield, Director of Threat Research & Intelligence at Eclypsium, told Digital Trends. “The repercussions of being able to steal what is essentially a complete history of what a user does on their machine are near impossible to comprehend.”
Jeff Williams, co-founder and CTO at Contrast Security, expressed confidence in the kind of security protections Microsoft has put in place for Recall. It is encrypted and lives behind biometric authentication.
Even if you are running it in the background and randomly pull up the dashboard, you will only be able to see the timeline after going through the security check. “For now, I don’t see how this is much more risky than log files stored locally,” Williams pointed out, adding that only extensive penetration testing will reveal the real picture of Recall.
Nadeem Sarwar / Digital Trends
For an average user like me, I believe Recall is a deeply rewarding feature. And yes, every layer of security that is added to the system is a welcome change. The risk of infostealers and malware with OCR capabilities will persist, but going past the Recall guardrails won’t be easy, especially with safety features such as just-in-time decryption.
You can always stay a step ahead and block Recall from logging your activity across specific websites and apps to be on the safe side. You can go with the auto-delete protocol, or even batch delete all snapshots from a certain website or app from Recall, in case you didn’t already filter it out.
In the wake of my tests, I found Recall to be one of the most practically rewarding AI features, right up there with Deep Research. What it really needs is to bring down the access walls (read: price), or wait for Copilot+ PCs to get a bit cheaper. The new 12-inch Microsoft Surface Pro is a great start, but more brands need to follow suit.
After months of teasers, previews, and select rollouts, Microsoft’s Copilot Vision is now available to try for all Edge users in the U.S. The flashy new AI tool is designed to watch your screen as you browse so you can ask it various questions about what you’re doing and get useful context-appropriate responses. The main catch, however, is that it currently only works with nine websites.
For the most part, these nine websites seem like pretty random choices, too. We have Amazon, which makes sense, but also Geoguessr? I’m pretty sure the point of that site is to try and guess where you are on the map without any help. Anyway, the full site list is as follows:
CEO of Microsoft AI Mustafa Suleyman announced the release on Bluesky yesterday and shared a few of his favorite use cases.
Copilot Vision is out now, free in Edge. It can literally see what you see on screen (if you opt in). Pretty amazing! It’ll think out loud with you when you’re browsing online. No more over-explaining, copy-pasting, or struggling to put something into words.
Usually, when you want to ask Copilot a question, you have to write out the paragraphs of context yourself, and aside from being slow and annoying, this can also be pretty difficult if you’re trying to ask about something you don’t know much about.
With Copilot Vision, instead of trying to describe what you’re looking at or what you’re talking about, the AI model can see it right on your screen.
So, according to Suleyman’s examples, you can search for “breathable sheets” on Amazon and ask Copilot if any of the results are made from appropriate fabrics. Copilot can point the right ones out to you or give you examples of breathable fabric to search for.
On the Food & Wine recipe website, Copilot can help you go hands-free while you cook by answering your questions and reading out parts of the recipe to you. This works because the whole experience is designed to work through voice — you speak directly to the AI and the AI speaks back.
According to one of the videos on the Copilot Vision page, however, it looks like you can type out questions too and receive written responses.
Microsoft is taking things very slowly and carefully with this feature, almost certainly because it wants to avoid triggering another backlash like it did with Recall. The limited number of compatible sites is connected to copyright issues, and the company makes sure to stress that the feature is “opt-in,” doesn’t record your screen, is only on when you turn it on, and deletes the data as soon as you end a session.
If you’re interested in testing it out, you can set things up and see a little tutorial through the Microsoft website.
There is always going to be a big divide between macOS and Windows. Much of it has to do with the functional disparities that are deeply ingrained at an OS-level. Or if you dive into the heated community debates, you will see it broadly as a battle between seamlessness and flexibility.
Gaming remains the guiding star for Windows adherents. A handful of highly specialized niche industry tools also remain locked to the Microsoft platform. On the other hand, macOS fans swear by the fluid software, plenty of firepower options in the M-series silicon era, and fantastic hardware.
But you can’t realistically compare an “entry-level” Apple laptop at $999 with a Windows laptop half that price and non-comparable innards. To put it in simpler terms, the gulf is here to stay. But in the past year or so, a leveling tool has arrived, riding atop the AI wave. Some of it is tremendously hyped, while a few others are genuinely useful.
AI: From hype to utility
Microsoft
Microsoft refers to its generative AI vehicle as Copilot, while Apple ships it under the Apple Intelligence bundle. Apple Intelligence on macOS has been meh, at best, and severely lacking vis-a-vis the Copilot competition.
So far, people who haven’t embraced it in their workflow aren’t bothered by it, but enterprises are adopting it rapidly. Analysts, on the other hand, are calling AI efforts one of Apple’s biggest flubs so far. Internal management changes and leaks certainly paint a picture of panic.
But what matters to an average user is the experience they get today, and that’s where Microsoft has a huge upper hand. The company’s Office 365 productivity suite has already been juiced up on Copilot. Such is the enthusiasm that even the good ‘ol MS-Paint has received AI image generation tools.
Microsoft / Digital Trends
You can’t open a mainstream Microsoft software today, and not see Copilot. Beyond the expectedly obvious stuffing across the board, there is some serious utility to be drawn from the AI stack.
In Word, it lives as an ever-helpful writing assistant for creative and brainstorming processes. PowerPoint users can summon it for making cool presentations with a text prompt, while in Excel, it turns data visualization into a cakewalk. We are merely scratching the surface here
Barely a few weeks ago, Microsoft 365 Copilot landed an AI Researcher tool that can scrape the internet as well as internal documents to compile comprehensive reports. Emails, chats, internal meeting logs, calendars, transcripts, and shared documents – it can process it all to handle your queries.
Notably, Copilot now has Vision capability. It can understand what’s appearing on the screen, and if there’s a camera available, it will also make sense of the world around you. “You’ll be able to use Copilot to search, change settings, organize files and collaborate on projects without switching between files or apps,” says Microsoft.
The best macOS integration is currently tied to ChatGPT, and its best features require an internet connection. Apple’s Macs are definitely ready, and obviously overpowered, but macOS simply lacks an in-house AI package to get the best out of it. That brings us to the topic of …
A hardware level-up for Windows
Luke Larsen / Digital Trends
A recurring theme in the Windows vs Mac debate is reliability and performance fluidity. To an extent, it’s accurate, primarily because Apple has a uniquely tight control over the hardware and the software stack. I have often found MacBooks perform more smoothly compared to a Windows laptop with identical hardware specs.
On paper, that is. Copilot has offered Microsoft an opportunity to serve a Windows performance boost that can stand toe-to-toe with macOS machines. Or at least, the MacBooks. That secret sauce is the Copilot+ range of PCs. Think of it as Intel’s Evo badge, but created by Microsoft in the age of AI.
Qualcomm’s Snapdragon X series chips remained the only option for a while that could match that NPU performance criteria. For comparison, Apple’s M4 can only manage 38 TOPS at AI workflows. But it’s not just AI performance where Copilot+ machines are leveling up.
Looking at the NPU requirement from another perspective, it has forced the likes of Intel and AMD to accordingly upgrade their CPU and GPU stack for the Copilot+ platform. And to ensure that these powerful processors don’t run into a bottleneck, OEMs are putting faster storage and memory chips inside their Copilot+ PCs.
Think of it as a reversing feedback loop, but one that is only pushing the cause of Windows laptops into a new territory, and one where consistently high performance and premium experience are guaranteed. Our own tests validate those incremental hardware enhancements.
Digital Trends
The Surface Laptop (7th Edition) won high praise in Digital Trends’ review, eclipsing the comparable MacBook Air’s multi-core performance and battery life figures. That’s no small feat, especially when both laptops are evenly matched at their $999 asking price.
So far, the biggest problem with Windows machines has been fragmentation. Unlike Apple, Microsoft has taken a near-universal approach to serving the core Windows experience on all machines, irrespective of whether it has a fire-breathing processor or if it’s just a cheap school-bound laptop with a low-power AMD Ryzen APU inside.
The latter is a recipe for OS frustrations, but one that Copilot+ erases at a certain price premium.
Luke Larsen / Digital Trends
Microsoft will also face a considerable challenge in hawking Copilot to its expansive community. A quick look at Reddit and Microsoft forums suggests that Copilot is a deeply divisive topic, and not everyone is happy with the AI stuffing within Windows.
It likely won’t fix the fundamental OS gaps where Windows lags behind macOS.
What Copilot offers is an opportunity to create standout computing experiences that focus more on practical benefits than flashy presentations. AI has its own set of inherent problems, so there’s that. But the sum total of all the development so far is that Copilot is changing the Windows experience in nearly all corners.
It may look the same, but functionally, there is a lot of new stuff to explore. And a few of them are quite useful. Having a reliably powerful (and nearly price-matched) hardware platform with Copilot+ is what Microsoft needs to edge ahead of the venerable Mac aura. It’s not impossible, but definitely worth a thoughtful attempt.
As AI tools improve, we keep getting encouraged to offload more and more complex tasks to them. LLMs can write our emails for us, create presentations, design apps, generate videos, search the internet and summarize the results, and so much more. One thing they’re still really struggling with, however, is video games.
So far this year, two of the biggest names in AI (Microsoft and Anthropic) have tried to get their models to generate or play games, and the results are probably a lot more limited than many people expect.
This makes them perfect showcases of where generative AI is really at right now — in short: it can do a lot more than before, but it can’t do everything.
Microsoft generates Quake II
Generating video games has similar problems to generating videos — movement is weird and morph-y, and the AI starts to lose touch with “reality” after a set amount of time. Microsoft’s latest attempt, which anyone can try out, is an AI-generated version of Quake II.
I played it quite a few times and it’s a truly trippy experience, with weird, smudgy enemies appearing out of nowhere and the environment changing around you as you move. Multiple times when I entered a new room, the entrance would be gone when I turned back to face it — and when I looked forward again the walls would have moved.
A screenshot of the real Quake II from its Steam page.SteamA screenshot of Copilot’s generated version of Quake II.Microsoft
The experience only lasts a few minutes before it cuts out and prompts you to start a new game — but if you’re unlucky, it can stop properly responding to your inputs even before that.
It’s a great experiment, however, and one I think would be useful for more people to see. It lets you experience for yourself what gen AI is good at, and what its current limitations are. As impressive as it is that we can generate an interactive video game experience at all, it’s hard to imagine that anyone could play this tech demo and think the next Assassin’s Creed will be made by AI.
These kinds of thoughts and assumptions do exist, however, and it’s largely because people can’t escape hearing about AI right now. Even if you couldn’t care less about artificial intelligence, it will still be shoved in your face everywhere you go. The problem is, that the information the average person gets is almost entirely made up of big tech marketing and CEO comments that get picked up by news publications.
That means they hear exaggerated and conflicting claims like these:
It has the potential to solve some of the world’s biggest problems, such as climate change, poverty and disease. (Bill Gates)
Probably in 2025, we at Meta, as well as the other companies that are basically working on this, are going to have an AI that can effectively be a sort of midlevel engineer that you have at your company that can write code. (Mark Zuckerberg)
Using AI effectively is now a fundamental expectation of everyone at Shopify. It’s a tool of all trades today, and will only grow in importance. Frankly, I don’t think it’s feasible to opt out of learning the skill of applying AI in your craft. (Tobi Lutke, CEO of Shopify)
We are now confident we know how to build AGI as we have traditionally understood it. We believe that, in 2025, we may see the first AI agents “join the workforce” and materially change the output of companies. (Sam Altman, CEO of OpenAI)
AI is more dangerous than, say, mismanaged aircraft design or production maintenance or bad car production, in the sense that it is, it has the potential — however small one may regard that probability, but it is non-trivial — it has the potential of civilization destruction. (Elon Musk)
This is all pretty extreme, right? It will both save us and destroy us, it’s both a tool of all trades for professionals and a tool that will replace professionals — and apparently, we could get sci-fi-level AGI as soon as this year. When this is all people hear, they start expecting pretty amazing things from these tools and believing all office workers spend their days conversing with their computers like Star Trek characters.
However, that is not what reality looks like. Reality looks like a trippy, smudgy Quake II with incomprehensible shapes for enemies. ChatGPT-level LLMs really were an exciting breakthrough in 2022, and a ton of fun for everyone to play around with — but for the majority of uses big tech is pushing on us right now, AI just isn’t capable enough. Accuracy levels are too low, instruction-following abilities are too low, context windows are too small, and they’re just trained on internet nonsense instead of real-world knowledge.
But generating a video game is a pretty complex goal — it takes whole teams of humans years to make these things, after all. How about playing video games instead?
Well, it turns out people are experimenting with that, too. Anthropic’s newest model, Claude 3.7 Sonnet, has been playing Pokémon Red on Twitch for around two months now, and he’s doing the best job an LLM has ever done at playing Pokémon. One slight caveat, however, is that he’s still miles behind the average 10-year-old human.
One of the problems is speed — it takes Claude thousands of actions spanning multiple days to do things like make it through Viridian Forest.
Why does it take so long? It’s not because he can’t figure out how to strategically win Pokémon battles — that’s actually the part he’s best at. Navigating through the environment and avoiding trees and buildings, on the other hand — not so good. Claude has never been trained to play Pokémon, and it’s not easy for him to understand the pixel art and what it represents.
Making it through maze-type areas like Mt. Moon is particularly difficult for him, as he struggles to form a map of the area and avoid retracing his steps. One time, he got himself so stuck in a corner that he concluded the game was broken and generated a formal request to have the game reset.
These early attempts were not without moments of levity too.
On one occasion, Claude got stuck in a corner and—convinced something must be broken—typed out a formal request to reset the game. pic.twitter.com/5RIiCJdxCM
He’s also not great at remembering what his goals are, what things he’s already tried, or which places he’s already been.
There’s a pretty straightforward reason for that one — LLMs have a finite “context window” that acts as their memory. It can only hold so much information, and once Claude hits the limit, he condenses what he’s got to make room for more. So a piece of information like “Visited Viridian City, entered every building, and spoke to every NPC” might get condensed to just “Visited Viridian City” — prompting Claude to go back and check if there was more to do in the city.
To sum it up: Claude can’t figure out where he’s going, he walks into walls, mistakes random objects for NPCs, forgets where he’s been and what he’s trying to do, and every decision he makes requires paragraphs and paragraphs of reasoning. This isn’t a criticism — these are both exciting experiments that are pushing LLMs as far as they can go.
But with all the hype around AI, it feels important for people to see demos like these and make their own minds up about AI. Certain figures are trying to push the narrative that we’re about to reach the peak — that within years, AI will be beyond even the smartest humans — but I don’t think they’re being sincere, they’re just being salesmen. We’re nowhere near the peak, this whole thing is only just beginning.
Microsoft’s 50th anniversary event was quite loaded, but the company reserved most of its attention for the Copilot AI stack. The buzzy event introduced two crucial upgrades – Actions and Deep Research — which firmly push Copilot into the realm of agentic AI.
Agentic AI is essentially a fancy way of describing an AI tool that can perform multi-step web-based tasks autonomously, or semi-autonomously, on your behalf. In Copilot’s case, the fancier one is Actions. So far, AI chatbots have mostly been able to give answers based on a certain input, but haven’t been able to perform autonomous multi-stage actions.
Microsoft says Actions can book “event tickets, grab dinner reservations or send a thoughtful gift to a friend and it will check that task off your list.” The core idea is that instead of having users visit a website and use a combination of clicks and keyboard typing, they can just tell Copilot to do it as a natural language command.
For example, you can ask Copilot to find a list of nearby restaurants that open late into the night, and book a table at the one you like the most, from within the search results. Copilot will do it all by browsing the web, filling in the necessary details, and occasionally asking for input wherever necessary.
Microsoft says it has joined hands with 1-800-Flowers.com, Booking.com, Expedia, Kayak, OpenTable, Priceline, Tripadvisor, Skyscanner, Viator and Vrbo for Copilot actions to get the job done on behalf of users.
Microsoft won’t be the first to launch a browser-based AI agent for handling cores. It’s actually late to the party. OpenAI’s Operator has been available for a while now, and it can accomplish more or less the same chores as Copilot Actions.
Microsoft (MSFT) announced an array of updates to personalize and automate its Copilot AI assistant at the company’s 50th anniversary celebrations Friday.
OpenAI is delaying the rollout of ChatGPT’s latest image generation features, Images in ChatGPT, for users on its free subscription tier, CEO Sam Altman said in an X post on Wednesday.
Microsoft is late to the party, but it is finally bringing a deep research tool of its own to the Microsoft 365 Copilot platform across the web, mobile, and desktop. Unlike competitors such as Google Gemini, Perplexity, or OpenAI’s ChatGPT, all of which use the Deep Research name, Microsoft is going with the Researcher agent branding.
The overarching idea, however, isn’t too different. You tell the Copilot AI to come up with thoroughly researched material on a certain topic or create an action plan, and it will oblige by producing a detailed document that would otherwise take hours of human research and compilation. It’s all about performing complex, multi-step research on your behalf as an autonomous AI agent.
Just to avoid any confusion early on, Microsoft 365 Copilot is essentially the rebranded version of the erstwhile Microsoft 365 (Office) app. It is different from the standalone Copilot app, which is more like a general purpose AI chatbot application.
How Researcher agent works?
Underneath the Researcher agent, however, is OpenAI’s Deep Research model. But this is not a simple rip-off. Instead, the feature’s implementation in Microsoft 365 Copilot runs far deeper than the competition. That’s primarily because it can look at your own material, or a business’ internal data, as well.
Instead of pulling information solely from the internet, the Researcher agent can also take a look at internal documents such as emails, chats, internal meeting logs, calendars, transcripts, and shared documents. It can also reference data from external sources such as Salesforce, as well as other custom agents that are in use at a company.
“Researcher’s intelligence to reason and connect the dots leads to magical moments,” claims Microsoft. Researcher agent can be configured by users to reference data from the web, local files, meeting recordings, emails, chats, and sales agent, on an individual basis — all of them, or just a select few.
Why it stands out?
Microsoft
The overall idea is to create a research tool that can create detailed plans on external data available on the web, as well as the internal company material. For businesses, that’s a huge relief. Microsoft already has tens of thousands of enterprises that have created their bespoke AI agents to automate internal work using the Copilot Studio tool.
“It leverages the enterprise knowledge graph to integrate user and organizational context, including details about people, projects, products, and the unique interplay of these entities within the user’s work,” says the company. During early tests, Microsoft claims the Researcher agent saved 6-8 hours on a weekly basis for selected adopters.
Access to Researcher will first start rolling out to Microsoft 365 Copilot customers in April. It will be released as part of a new Frontier program that gives early access to new Copilot tools, somewhat like the Insider Preview program for beta testing Windows OS.