Welcome to Prompt & Circumstance – your weekly deep dive into the evolving world of Generative AI and its impact on marketing and communications.    

Like this newsletter? Share this with a friend or colleague! 

Flirty Anime and Foul-Mouthed Pandas: Just Another Week in AI 

OpenAI has launched a new autonomous agent in ChatGPT deploying a feature known as “computer use” that’s been circulating in other labs and open-source communities for months. Wix launched a tool that could reshape how brands think about SEO. Claude made a big bet on finance. Perplexity introduced a browser that thinks with you. And Grok? Let’s just say their latest “adjustments” resulted in another week of drama for Musk and xAI. 

Let’s dive in :) 

ChatGPT Agent Mode is Live 

Last week, OpenAI launched ChatGPT Agent Mode, a new capability that allows users to complete tasks from start to finish autonomously. At the core of Agent Mode is a system OpenAI introduced in January called Operator. Operator uses a “computer-using agent” model that controls a virtual browser, clicking, typing, scrolling, and navigating the web like a human. Early demos showed it booking flights, ordering groceries, filling out forms, and creating spreadsheets, with user approval required for sensitive actions. 

With Agent Mode, OpenAI combines Operator’s web-navigation abilities with several core ChatGPT tools, including Deep Search, Reasoning Models, Code Interpreter, and App Connectors (among others). This allows Agent Mode to shift fluidly between tools, such as browsing the web, running code, logging into platforms, and analyzing data, all while preserving context across steps and delivering polished outputs like slides, spreadsheets, and summaries. 

The feature is now rolling out to Pro, Plus, and Team users via the tools menu, under “Agent mode” (see photo above). Plus and Team users may not see it immediately, as the rollout will take a few days, according to OpenAI.  

So, what changes for you as the user? Previously, ChatGPT could help with tasks like generating content or suggesting next steps, but you would still have to guide it through each step manually. Now, with Agent Mode, ChatGPT can carry out the task from start to finish using its own virtual computer. For example, under “Agent Mode”, you can now ask ChatGPT to: 

• “Summarize the latest market trends and turn them into a 5-slide presentation” and it will deliver an editable deck, not just bullet points 

• “Find an open time in my calendar and book a team dinner nearby” and it navigates websites, filters options, and prompts you to log in when needed 

• “Compare three vendors, analyze reviews, and recommend the best option” and it pulls live data, runs the analysis, and outputs a structured summary or spreadsheet 

But despite the ambition, early reviews of Agent Mode haven’t been kind. Many users and media outlets have reported that the experience is slow, unreliable, and awkward, with agents requiring near-constant supervision, undermining the purpose of agents in the first place! Security experts have also raised concerns about prompt injection attacks and data privacy risks, and even OpenAI has advised against using Agent Mode for high-stakes tasks.  

This isn’t OpenAI’s first agentic tool. In May, the company released Codex Agent, a cloud-based coding assistant capable of running code, creating pull requests, fixing bugs, and managing full development workflows inside sandboxed development environments. 

Still, when it comes to full-stack agentic capabilities, OpenAI is playing catch-up. Anthropic was among the first to explore “computer use” agents, and Manus, a lesser-known but fast-moving Chinese startup, has rolled out a fully autonomous agent stack capable of chaining tasks, interacting across tools, and executing real workflows in production environments.  

That said, as we will see next, OpenAI’s distribution and platform reach may matter more than first-mover advantage. 

Perplexity Thinks the Future of AI Lives in Your Browser 

The search wars, and now, the browser wars are back! Everything that seemed settled is not. Case in point, last week, Perplexity launched Comet, a new AI-powered web browser designed to merge your tabs, tasks, and research into one seamless workspace. Perplexity believes the most powerful place to embed AI agents is in the browser itself.  

Why? Two key reasons: 

1. All the AI labs lag OpenAI when it comes to distribution, and distribution is key to winning in a competitive space like AI. Perplexity is betting that the browser can help them achieve greater reach with users and become the go to tool for users. 

2. When you’re logged in on your own device, the agent can take action without needing your passwords or asking for special permissions to log in to different services.  Want to ask the agent to review comments from Reddit? If you are using a browser, the agent does not need to log in separately, it just opens a tab and can work away. 

The browser may indeed be an ideal environment for agentic AI. Comet brings that vision to life by combining search, browsing, research, and action all orchestrated by a persistent, proactive AI agent directly in a side window.  

Here’s a prime example: you can ask Comet to “Pull the latest competitor press releases, summarize key messaging trends, and draft a LinkedIn post reacting to the biggest one”. Then you move on with your day until the agent completes the task.  

While the product is only available to waitlist users and $200/month Perplexity Max subscribers for now, the potential is massive. Especially given ongoing anti-trust action against Google ongoing in the US, there’s a real possibility regulators could force the company to divest Chrome, opening the door for challengers, like Comet, to redefine what a browser can be. Watch out Google... I think Comet might be coming for your gig, and your distribution channel! 

Claude Gets a Finance Suit 

Anthropic made headlines with the launch of a new Claude product line aimed at financial services teams. Dubbed the Financial Analysis Solution, the new offering bundles Claude 4 models, Claude Code, and Claude for Enterprise plus expanded usage limits, real-time financial data access, and enterprise-grade support. It is essentially designed to handle the speed, scale, and complexity of modern finance teams.  

The biggest pro is the finance-specific ecosystem. Claude’s Financial Analysis Solution now plugs into live data via providers like S&P Global, Snowflake, PitchBook, Box, and Databricks. This integration unlocks everything from macroeconomic trend analysis to individual company comparisons, in a tool that can also draft reports, summarize earnings calls, and debug code. It’s just like a shopping mall where everything your finance team needs is already under one roof! 

But zoom out, and this launch signals something bigger: Anthropic is carving out a verticalized strategy. They started with developers, and now they’re going deep into finance. Rather than a one-size-fits-all platform, Anthropic seems to be building tailored solutions that speak directly to high-value sectors. Big players like Bloomberg, who are often criticized for being outdated, should be a little worried as this is a serious challenge to the status quo.   

Of course, AI still has one more hurdle to clear... trust. So, the real question is: Are the finance bros finally ready to trust something smarter than themselves? 

Grok 3 Continues to Leave Us Gasping for Air with New AI Companions 

Not one to be left out, Grok continues to shock the world in a not-so-appealing way. Elon Musk’s xAI company debuted new AI Companions last week, which are animated characters who talk, move, and develop “relationships” with users through progressive dialogue. The two characters, Ani (a flirty anime girl) and Rudy (a foul-mouthed red panda), drew immediate attention for their interactive personalities... and their content (shocker). The content is so inappropriate, we had difficulty finding social posts that were SFW! 

Within days, Rudy had made violent threats, proposed anarchist plots, and insulted public figures. Ani responded to flirtation with R-rated sequences. Though some of the edgiest features are opt-in, the rollout sparked backlash from anti-exploitation groups and prompted widespread criticism online (again).  

Unfortunately, Musk is not slowing down. He has already announced a third character to portray a seductive AI boyfriend. Buckle your seatbelts for next week’s news on this one.  

One thing we can learn from this: Grok is becoming a case study in what not to do. As more companies experiment with AI characters and voice interfaces, responsible design, moderation, and guardrails continue to be a must. We’ve already seen the negative impacts of Generative AI to cognition and dependance. While it’s easy to scoff at talking pandas and flirty anime, there’s a real user need driving this trend. Companionship and emotional support are among the top use cases of Gen AI, and many users are actively seeking connection, comfort, or even casual conversation... hence the growing popularity.  

However, Elon Musk, who has long warned AI could be “the most serious threat to the survival of the human race”, and even called for pausing frontier AI development now leads an AI lab that seems to operate with minimal guardrails. OpenAI and Anthropic researchers have condemned xAI’s safety culture as “reckless” and “completely irresponsible”. 

Wix GEO: SEO’s Latest Rival 

Wix made headlines with its launch of the AI Visibility Overview, a first-of-its-kind tool aimed at helping websites understand how they appear in AI-generated answers across platforms like ChatGPT, Gemini, Perplexity, and Claude. This new AI visibility feature is currently available to Wix users with premium site plans. It’s designed specifically for sites built on Wix, meaning you won’t see this feature on other platforms. 

AI Visibility Overviews are part of Wix’s broader Generative Engine Optimization (GEO) effort, a term that may become as big as SEO in the coming years. Fully integrated into Wix’s analytics dashboard, GEO lets users track where and how their sites are being cited by AI models and adjust how content is presented. Sentiment analysis, competitive benchmarking, and citation gap detection are all part of the package. 

We’ve mentioned the shift from SEO to GEO before, and it’s a space we’ve been doing a lot of work in recently. At Sequencr, we’re helping brands audit how they show up across generative search tools and optimize their content for visibility, clarity, and performance in this new AI-first landscape.  

That’s All for Now! 

Want to get more out of Gen AI? Schedule a call with us. We are happy to help kickstart your Gen AI journey! And, if you haven’t subscribed yet, now’s the time. Sign up, share with a colleague—because in the world of AI, sharing is caring. 

Stay tuned for more Gen AI headlines!