Inbox, meet intelligence
Journalists and knowledge workers will benefit as AI assistants shift from reactive chatbots to proactive collaborators.
“Which AI do you use?” is starting to become an icebreaker question at some cocktail hours (it could be just the ones I go to). I think a better question, though, is where do you use AI? If it’s mostly via the public chatbots, you’re leaving a lot of potential untapped. I’m more convinced of that idea the more I use Perplexity’s Comet browser. Now the company has an email assistant that brings similar intelligence to your inbox.
More on smart email assistants and how they could end up being a crucial tool for journalists in a minute, but first I’d like to let you know I’ll be joining Econify CEO Vince Wadhwani on Friday at 11 a.m. ET for fun session that’s going to explore how AI is forging new relationships between journalists and engineers. Spoiler: they’re becoming more like each other, and I’m excited to unpack what that means. You can register for the webinar here.
That same day I’ll be teaching my monthly AI beginners class, AI Quick Start. Come on by to hear more about browser agents, and how journalists and PR professionals can use them to cut down on time navigating tedious databases. Go ahead and sign up here.
A MESSAGE FROM HELP A REPORTER OUT
HARO is back.
One of the most iconic names in media sourcing has returned—stronger than ever. Help a Reporter Out (HARO), now owned and operated by Featured.com, is once again connecting journalists with credible, ready-to-talk experts across every industry.
For media executives and journalists, HARO offers a direct line to trusted sources who can add authority, insight, and perspective to your stories. For PR professionals and marketers, it’s your chance to build visibility for your brand, clients, or executives by getting quoted in top-tier publications.
HARO has modernized on the backend to maintain quality and trust while keeping its greatest strength intact: simplicity. Opportunities land straight in your inbox three times per day, making it easy to pitch or source on deadline. Thousands of journalists and sources rely on HARO to power stories every day.
Discover why HARO remains the bridge between journalists and sources at HelpaReporter.com.
Beyond chatbots: Why the future of AI is contextual and embedded
One of the core lessons in the AI trainings I do with editorial teams is to get in the habit of looping in AI on any problem you’re working on. Whether it’s strategizing, working through a story angle, or getting recommendations on sources, AI can probably get you to the finish line faster if you just ask.
The thing is, continually toggling between a chat window and the work you’re doing gets tiresome. It also typically means you need to constantly re-explain the context of what you’re doing to an oblivious chatbot. Even if you’ve trained yourself to regularly collaborate with AI, getting your chatty digital friend up to speed can often lead to so much iteration that you’re better off just doing the task yourself.
That’s why I’m biased toward any AI tool that extends beyond the chatbox. A good example of this for newsrooms is YESEO, an SEO optimizer that connects directly to Slack. Another one is GPT for Work, which lets you apply AI actions directly in individual cells within a spreadsheet. And I’m an early adopter of Perplexity’s AI-powered Comet browser, which enables you browse in tandem with a virtual assistant.
Intelligent email
Now Perplexity is launching a different assistant, this one for email, which you can connect to your Gmail or Outlook. The idea isn’t just that it performs tricks like autocomplete and creating calendar invites from messages—though, yes, it can do those things—it’s that it has access to the context of your entire inbox. With all that knowledge, the assistant will enable users to “ask anything about your inbox and discover new ways to extract value from your email data.”
It’s easy to see how that would be useful, especially for busy reporters who are being pitched all day and all night. Inbox tools for surfacing important messages have existed for well over a decade, but they (rightly) prioritize known contacts and ongoing conversations without doing much else. What would help greatly is something that can look at the morass of unsolicited emails and pick the diamonds in the rough, or possibly better, recognize underlying patterns.
It also promises to be a much smarter autocomplete. It’ll supposedly be able to mimic your style and take the reins on tedious threads, like those back-and-forth exchanges just to pin down a meeting time. Or when you need to blast out personalized requests to a dozen potential sources.
Context gone wrong
I haven’t used Perplexity’s email assistant yet, but in my view it can’t be worse than how Gemini interacts with Gmail, which functions as a counterexample of AI understanding and acting on the right context. If, on your Gmail inbox, you start typing in a new message window on the desktop and ask Gemini about it, the AI will often not understand that you’re referring to the message you’re composing instead of the one that might be in the main window. Its abilities are hit-and-miss, too—I recently had Gemini outright refuse to create a calendar invite for an event because the email said it was at capacity. Even when I insisted, it still said no.
In other words, those are examples of bad context engineering, something many are predicting will be the next frontier in AI. Yes, the models have improved greatly, and they’re all very “smart”—that’s practically a given these days. But how do you ensure the AI understands what you’re doing, and has access to the data it needs, to apply those smarts in the best possible way?
There are serious questions around privacy and security that must be addressed for that to happen, and it seems clear we’re still figuring out the answers, so I’m not endorsing that newsrooms eagerly deploy email assistants tomorrow. That said, the idea of bringing intelligence out of chatbots and dedicated tools, and layering it on top of the things we do every day, is a powerful unlock. For AI to offer real utility, it needs to stop making us go somewhere else to get it.
Learn Practical Ways of Using AI in Journalism and PR
AI isn’t a fad. For most journalists and PR professionals now, it’s simply the reality. Yet many reporters and executives are still stuck dabbling: using ChatGPT for story ideas or a few headlines and calling it a day.
It’s time to upgrade.
Coming up, The Media Copilot is offering a one-hour power session on how to actually use AI for real media work. Research smarter. Pitch sharper. Write faster—without losing your voice to the algorithm.
Here’s what you’ll walk away with:
The real state of AI in media—no hype, just how it’s changing the game.
A simple way to approach prompting to get what you actually want.
Pro-level use cases: Find better sources. Focus your stories and campaigns. Protect your data.
📅 Live: Sept. 21, 1 p.m. ET
💸 $49: The cheapest upskill you’ll make all year.
🎁 Extras: Slides, prompt templates, and a quickstart AI guide to keep you sharp.
👉 Don’t just survive AI. Use it to do better work.
The Chatbox
All the AI news that matters to media*
Big Tech returns to the media bargaining table
According to reports, Microsoft and Meta are taking somewhat different approaches to paying publishers for AI content. Microsoft’s Publisher Content Marketplace, reported by Axios, represents a systematic solution—a two-sided marketplace that would compensate outlets based on actual usage by Copilot, starting with a small group of publishers before expanding. Meta, meanwhile, is scrambling to catch up after years of retreating from news. The Wall Street Journal reports Meta is approaching Fox, News Corp, and Axel Springer for traditional licensing deals similar to what OpenAI already secured.
The contrast is revealing: Microsoft is building infrastructure for sustainable, usage-based compensation while Meta is playing defense, trying to acquire the same content deals its competitors already locked down. Microsoft’s marketplace model echoes efforts of players like ProRata: establishing a system for fair, measurable compensation. By contrast, Meta’s late-to-the-game negotiations suggest the company recognizes that quality training data has become a competitive necessity rather than a nice-to-have. (AI-assisted)
Penske’s legal gambit changes the AI game
Penske Media’s lawsuit against Google marks a shift in how publishers are fighting the AI wars, and it could end up being pivotal. Penske is discarding the losing playbook that focuses on copyright violations and trades it for an antitrust strategy that could actually work. By arguing that Google’s AI Overviews constitute “reciprocal dealing”—essentially forcing publishers to hand over content to maintain search visibility—Penske is dusting off a 1960s legal theory that once broke up conglomerates and applying it to Silicon Valley’s algorithmic monopolies.
The timing couldn’t be better: recent court rulings have already declared Google a monopolist, Pew data shows AI summaries tank click-through rates, and the copyright route has proven futile after judges tossed similar cases from Anthropic and Meta. Penske’s creative legal maneuver is a potentially more viable path to push back against tech platforms that have graduated from aggregating content to replacing it entirely. (AI-assisted)
NYT invents an AI detective
Digiday did a deep dive on The New York Times’ custom AI toolkit. The highlight is an investigative tool called Cheat Sheet, which performs semantic search on massive document dumps that would otherwise remain impenetrable. The tool enables vibes-based searching, which helped reporters parse 5 million words from leaked Zoom recordings to expose election interference—the kind of investigation that would have required an army of interns just two years ago. With 1,700 staffers trained but zero AI-generated articles published, the Times demonstrates that the most valuable AI deployment in journalism might be keeping the machines in the background, crunching data while humans do what they do best: make the calls, ask the questions, and tell the stories. (AI-assisted)
The copyright wars just went platinum
The music industry’s copyright battle with AI companies just escalated from skirmishes to total war, with the International Confederation of Music Publishers alleging that OpenAI, Suno, Udio, and Mistral have scraped the entire world’s music catalog without licensing. This marks a significant expansion of the copyright conflict beyond text and images into audio, where the economics are even more brutal: AI-generated music already comprises 28% of daily uploads on Deezer, and artists face losing a fifth of their income within four years.
For media executives watching their own licensing negotiations with AI firms, the music industry’s predicament offers a stark preview: when AI companies can ingest entire creative catalogs through web crawlers and licensed platforms like YouTube, the traditional leverage of rights holders evaporates. The fact that only one AI music generator, Eleven Music, has signed a licensing deal while the rest allegedly operate without permission suggests the AI industry’s playbook remains unchanged—scrape first, negotiate later, if at all. (AI-assisted)
AI drafts go mainstream, disclosures don’t
Business Insider’s decision to let journalists draft entire stories with ChatGPT while keeping readers in the dark, reported by Status, marks a watershed moment for newsroom AI adoption. The Axel Springer-owned outlet’s new policy, which explicitly allows AI-generated first drafts but deems disclosure “unnecessary,” turns away from transparency standards that have slowly become a de facto standard over the last two years. The news comes just weeks after the outlet was duped by a fictitious AI-powered freelancer. While CEO Barbara Peng insists AI won’t replace “talking to real people,” the policy’s mixed messaging, simultaneously allowing and discouraging AI drafts, fuels the growing tension between editorial standards and the desire for AI-powered efficiency. (AI-assisted)
*AI-assisted Chatbox items are created with AI and then carefully edited by Media Copilot editors.
Do journalists not care about protecting their information from private companies anymore? Is that a completely bygone era?