AI in Marketing

Anthropic has launched Opus 4.6, an advanced model enhancing its Claude Code capabilities. This release introduces "agent teams," a feature that allows multiple agents to divide and coordinate tasks, improving efficiency and speed. Scott White, Anthropic's Head of Product, likens this to having a skilled human team working in parallel. Currently, these agent teams are accessible in a research preview for API users and subscribers. Opus 4.6 also features an expanded context window, capable of recalling up to 1 million tokens per session, similar to the company's Sonnet models. This enhancement supports work with larger code bases and documents. Additionally, Opus 4.6 integrates Claude directly into PowerPoint, allowing users to create and edit presentations within the application, a significant improvement over the previous method of transferring files. White notes that Opus has evolved from a tool primarily for software development to one useful for a broader range of professionals, including product managers and financial analysts. This shift reflects the diverse user base that has adopted Claude Code for various tasks beyond its initial scope. The new features aim to broaden Opus's appeal and functionality, making it a versatile tool for different industries and applications. Anthropic's latest release underscores its commitment to expanding the utility and accessibility of its AI models.

Next.

OpenAI has introduced GPT-5.3 Codex, a new model that enhances its existing agentic coding tool, Codex, aimed at software developers. This model is designed to extend Codex's capabilities beyond just writing and reviewing code, enabling it to perform a wide range of tasks typically done by developers and professionals on computers. OpenAI claims that GPT-5.3 Codex can develop complex games and applications from scratch within days. The model is reportedly 25% faster than its predecessor, GPT-5.2, and is notable for being the first model that contributed to its own development, as early versions were used by OpenAI staff for debugging and performance evaluation. The release of GPT-5.3 Codex comes shortly after Anthropic, a competitor, launched its own agentic coding model. Both companies initially planned to release their tools simultaneously at 10 a.m. PST, but Anthropic advanced its release by 15 minutes, slightly preceding OpenAI. This development highlights the competitive landscape in the field of AI-driven coding tools, with both companies striving to enhance the capabilities and efficiency of software development processes. The advancements in these models suggest a shift in how software can be built, potentially broadening access to software development and altering traditional workflows.

On a different note.

There's no shortage of marketing agencies claiming to use AI. But Nytro Marketing, a B2B agency based in the Philadelphia area, is doing something that actually backs that up.
The company is currently hiring a Junior Marketer, and the first step in the process isn't a phone screen or a recruiter call. It's a conversation with a custom-built AI chatbot powered by GPT-4o.
The chatbot adapts its questions in real time based on the candidate's responses, scores everyone against the same weighted competencies, and removes the kind of unconscious bias that often creeps into early-stage screening. Candidates who prefer a human conversation can skip it entirely and speak directly to the hiring manager.
It's a small but telling example of what happens when a marketing team stops treating AI as a talking point and starts shipping it into real workflows.

Meanwhile.

Reddit is exploring AI-powered search as a significant business opportunity, aiming to merge traditional and AI search capabilities. During its fourth-quarter earnings call, Reddit highlighted the potential of generative AI search, particularly for queries with multiple perspectives. CEO Steve Huffman emphasized Reddit's strength in addressing questions without definitive answers, suggesting that AI could enhance traditional search functions. The company reported a 30% increase in weekly active search users, rising from 60 million to 80 million, and a substantial growth in AI-powered Reddit Answers users, from 1 million to 15 million over the past year. Reddit is modernizing its AI answers interface to include more media-rich responses and is piloting dynamic agents. The company plans to eliminate the distinction between logged-in and logged-out users by Q3 2026, aiming to personalize the site using AI. Reddit's content licensing business, which allows other companies to train AI models on its data, is also expanding, contributing to an 8% year-over-year increase in "other" revenues, reaching $36 million in Q4 and $140 million for 2025.

In other news.

The BBC reports that OpenAI CEO Sam Altman faced criticism for his online response to Anthropic's Super Bowl ad campaign. Anthropic's ads, which mock the introduction of ads to ChatGPT, were described by Altman as "dishonest" and "deceptive." Social media users, however, likened Altman's reaction to a "tantrum," with some calling it hypocritical. Anthropic's ads humorously depict scenarios where ads disrupt chatbot interactions, ending with the slogan "Ads are coming to AI. But not to Claude." Despite the playful tone, the decision to air these ads during the Super Bowl, a costly marketing event, indicates the seriousness of the rivalry. Altman defended OpenAI's ad strategy, emphasizing free access and contrasting it with Anthropic's premium offerings.

With that said.

Meta is testing new link posting limits on Facebook for non-verified professional accounts, allowing only two links per month unless users subscribe to Meta Verified, which costs between $14.99 and $500 monthly. The move aims to keep users engaged on the platform, aligning with Meta's business model of serving ads and collecting data. Meta Verified subscribers can post unlimited links and gain more reach. Mari Smith suggests that instead of subscribing, businesses might benefit more from spending on ads. Meanwhile, Meta's $2 billion acquisition of Manus, a leader in autonomous AI agents, aims to integrate advanced AI capabilities into its platforms, potentially enhancing customer support and other functions. Manus agents can execute tasks like market research and data analysis, distinguishing them from other AI platforms. This acquisition reflects Meta's efforts to remain competitive in the AI space, particularly against companies like DeepSeek, OpenAI, and Google.

Moving on.

Stanford and Princeton scientists have introduced MedOS, an AI-XR-cobot system designed to assist clinicians in medical environments. MedOS integrates artificial intelligence, augmented reality, and robotics to address physician burnout by reducing cognitive overload and enhancing precision. The system acts as a clinical co-pilot, using smart glasses and robotic arms to assist in real-time medical procedures. It aims to reduce errors and support clinicians by bridging digital diagnostics with physical interventions. MedOS employs a dual-system architecture to mimic human cognition, demonstrating advanced reasoning and autonomous execution of complex clinical tasks. It has been tested in surgical simulations and hospital workflows, showing promise in tasks like laparoscopic assistance and anatomical mapping. The system is modular and adaptable, capable of interpreting real-time video and assisting with robotic tool alignment. Supported by NVIDIA and other partners, MedOS is in early pilot stages at several universities and will be publicly unveiled at NVIDIA’s GPU Technology Conference. The project seeks to integrate robots into all aspects of medicine.

Shifting gears.

SaaS applications, despite their transformative impact on business operations, remain cumbersome and inefficient to use. They often require extensive implementation, dedicated administrators, and complex navigation, creating a significant operational burden. This complexity, termed the "Ops Tax," results in high costs and inefficiencies, as businesses spend more on human resources to manage these tools than on the software itself. The introduction of AI was expected to simplify these processes, but instead, it has added another layer of complexity. Each SaaS tool now comes with its own AI agent, which requires additional management and does not integrate seamlessly with others. This has led to a fragmented system where AI agents operate in silos, further complicating workflows. The potential of AI to revolutionize SaaS lies in its ability to understand and execute tasks through natural language, eliminating the need for complex interfaces. However, current implementations merely add AI features to existing systems without addressing the core issue. The future of SaaS should focus on conversational interfaces that simplify operations and reduce the reliance on specialized personnel, ultimately shifting the focus from managing tools to achieving outcomes.

Finally.

The AI in Marketing newsletter and podcast will be taking a short break and returning on February 16, as the Nytro Marketing leadership team heads into an off-site meeting. We wish you a great week ahead. Unless a major breaking story shakes up the world of AI in marketing before then, we'll see you on the other side. Stay tuned.