It’s nearly impossible to see any recent tech news that doesn’t somehow involve AI. According to Statista, the term AI (excluding when it is part of a company name like OpenAI) was collectively mentioned nearly 200 times by Alphabet, Meta, Microsoft, Amazon and Apple in recent earnings calls. That number is up drastically from fewer than 40 mentions in the same period last year. Whether it’s to boost street cred (Wall Street, that is—looking at you, Zuck) or serve customers’ needs, the generative AI (GAI) frenzy that started with Microsoft’s ChatGPT Copilot announcement has sparked a tech news bonanza.
Zoom has its own recent GAI news, introducing partnerships with AI companies to integrate GAI into its communications platform offerings. This week, Zoom announced a new collaboration with Anthropic, an AI safety and research company that also builds AI models, to bring Anthropic’s Claude large language model (LLM) chatbot to the Zoom Contact Center portfolio. The partnership comes just months after Zoom’s announcement that it would integrate OpenAI’s ChatGPT LLM into the AI-driven Zoom IQ “smart companion.” Zoom is also exploring other AI functionality, taking a federated approach for its family of products.
In this article, I’ll examine what a federated approach to AI means for Zoom. I’ll also dig into the productivity boost that Claude could bring to the Zoom Contact Center and the rest of the Zoom platform. Finally, and perhaps most importantly, I’ll examine how these things matter to employees—in contact centers and beyond.
Zooming in on a federated approach to AI
Zoom’s federated approach to AI means the company will incorporate a variety of AI models, including its own proprietary models, those from leading AI companies like Anthropic and OpenAI, and select customers’ models. This flexibility allows Zoom to cater to the diverse needs of its users by customizing AI models that best suit specific business needs and requirements.
With its expanded AI ecosystem partnerships, Zoom is clearly signaling it has big plans to integrate AI more broadly across the Zoom platform. The federated approach has served the company well so far, as evidenced by the company developing its own AI chatbot, “Zoe,” for Zoom Contact Center, prior to partnering with third-party LLMs. Given the high cost of LLMs, the strategy certainly makes sense for a company with a user base that runs the gamut from individual free users to enormous enterprise accounts.
It is also worth noting that Zoom put its money where its chatbot is. In a parallel development to the working partnership with Anthropic, Zoom Ventures, the investment arm of Zoom, also made a strategic investment in Anthropic. This investment solidifies the relationship between the two companies and underscores Zoom’s commitment to driving further innovation within its federated ecosystem. While Zoom did not disclose the investment amount, Zoom Ventures typically invests a minimum of $250,000 and does not account for more than 20% of a given investment round, according to the fund’s FAQ. Anthropic has also received $300 million in investments from Google.
Anthropic’s AI assistant, Claude, is an intelligent virtual assistant (i.e., a chatbot) designed to help users with various online tasks and interactions. Claude’s natural language processing (NLP) abilities allow the chatbot to understand user intent and context during any interaction. Claude can answer questions, provide detailed information and perform tasks based on user requests.
Claude’s steerability—the ability to change or modify the model—allows Zoom customers to influence the bot’s behavior and customize its responses. For users, this ensures a personalized experience relevant to the brand or product the end consumer is engaging with. For Call Center users, this means the chatbot can be trained on specific information relevant to consumers’ needs. For example, if a consumer has a broken widget, the bot can be trained to know everything about the widget and help troubleshoot until the problem is resolved. The bot can also be trained to know when a problem isn’t solvable by self-service and escalate the issue to a human agent. Additionally, the model allows users to customize and guide the AI system’s tone and personality, granting users control and responsiveness that aligns with their brand.
Enhancing productivity and employee experience through intelligent solutions
I’ve written before about how companies must cultivate and reinforce brand loyalty with each customer interaction. By employing a smart AI assistant, Zoom allows efficient non-human troubleshooting for consumers’ basic queries and problems. This frees call-center workers to focus on the harder problems and crucial moments that really matter to their customers. In doing so, it also diminishes the stress put on call center pros by reducing call volume. This enables Zoom’s enterprise clients to deliver exceptional customer service at the same time they improve their employees’ workplace experience.
Ultimately, the use of AI assistants boils down to increasing productivity. “We are as excited as anyone to see the innovations in large language models that are going to make generative AI a driving force in making our customers’ organizations more productive,” Smita Hashim, chief product officer at Zoom, said in a blog post about the recent AI announcements.
Dario Amodei, CEO and cofounder of Anthropic, echoed Hashim’s sentiment: “We look forward to working together to showcase Anthropic’s capabilities and our joint commitment to enhancing productivity through intelligent solutions.”
Both Zoom and Anthropic are invested in creating solutions for the Zoom platform that answer the diverse needs of users while also boosting productivity across the company’s applications.
Giving Claude some context
Anthropic recently announced that it had expanded Claude’s context window to roughly 75,000 words. The context window refers to the amount of text the bot can digest to help it understand a prompt, and Claude’s capability is now significantly larger than that of existing models such as GPT-4. This means that the AI assistant can quickly comprehend lengthy text, such as user manuals. This allows Claude to develop a deeper understanding of user input on the fly, leading to more contextually relevant and meaningful interactions.
Claude is designed to be self-contained and does not search the internet when formulating responses. This further speaks to the contextual relevancy and lack of “hallucinations” that users should encounter when using Claude.
Risk mitigation with federated AI
I see much value in Zoom’s approach to federated AI, which allows the company to plug in different AI models across different functions as the industry and the technology evolve. It makes sense that Zoom would first use Claude in the Contact Center, a setting where it is imperative to follow corporate rules and brand guidelines closely.
Anthropic’s Constitutional AI model is focused on safety, interpretability and steerability. The model adheres to predefined safety constraints and ethical considerations and allows for training with clear parameters around its decision-making processes. Particularly as this week’s Senate Judiciary subcommittee’s inquisition on AI oversight raises many unanswered questions, aligning AI technology with human values and societal needs is a top-of-mind challenge for the whole tech industry.
Will Claude and its kin gobble up jobs?
AI regulation is likely coming. What isn’t likely to be regulated is how AI will affect the workforce. While the notion of AI-driven automation making jobs easier is attractive, many workers worry that bots like Claude will take their jobs. PwC’s most recent annual global workforce survey showed that nearly one-third of workers fear that technology could replace their role in the next few years.
These fears are hardly ill-founded. A McKinsey study suggested that AI may replace about 15% of the global workforce between 2016 and 2030. That percentage works out to roughly 400 million people. But the same study also stated that if AI adoption becomes widespread, the amount of jobs displaced could double, to 30%. Considering the recent rush to embrace GAI, it seems reasonable to assume that we’re aiming at the upper end of the range McKinsey suggested, if not even higher.
So far, no big wave of job losses to AI has materialized. But the technology is also not yet advanced enough to produce reliable results consistently. It’s likely just a matter of time, however. IBM’s CEO recently declared that AI might replace 7,800 jobs at the company as it paused hiring for replaceable roles.
Because of these fears, companies like Zoom must find a way to manage workers’ fears as they introduce new tools that could potentially replace humans somewhere down the line. Otherwise, they risk employee disengagement. If handled well, however, managing workers’ expectations and properly training them on the latest technologies should benefit both employees and the organization.
Zoom’s recipe for AI-driven success
Anthropic’s focus on building reliable, interpretable and steerable AI systems is commendable. However, AI can still be prone to errors or biases, leading to incorrect or unreliable outcomes. With all the talk in the Senate and beyond about the flaws and dangers of AI, introducing even the safest and most transparent AI systems into the Zoom platform may face resistance or skepticism from some users. Zoom’s investment in the company shows a commitment to a high level of responsible AI, but Zoom must also carry that through its entire federated ecosystem.
I think Zoom is invested in AI for the long haul, and that its federated approach will serve the company and its diverse customer base well. Much like any other company diving deep into AI, Zoom needs to focus on how these integrations will affect the employee experience. It’s usually true that a good employee experience leads to a good customer experience. Put the two together, and more often than not you end up with a successful company.
By leveraging Claude’s capabilities, Zoom should be able to guide its call-center agents toward trustworthy resolutions at the same time it empowers end users through better self-service options. Just as every company investing heavily in AI needs to do, Zoom must prepare for concerns about job displacement, loss of the human touch or over-reliance on AI.