Zoom has announced it will embed a new AI Assistant – named “Claude” – into its CCaaS platform.
The move stems from the vendor’s new partnership with Anthropic, a developer of AI systems and language models.
Claude is Anthropic’s competitor to ChatGPT, which Zoom will soon deploy across its portfolio, staring with the Zoom Contact Center.
Indeed, the CCaaS platform will soon have two new Claude-driven agent-assist applications under its banner.
Discussing these during an earnings call, Eric Yuan, CEO of Zoom, said:
With Claude guiding agents toward trustworthy resolutions and powering self-service for end-users, companies will be able to take customer relationships to the next level.
As Yuan suggests, the first application will support contact center agents in finding answers to customer queries that sprawl across numerous systems.
Meanwhile, the second will leverage Claude as a self-service tool that customers can use to autonomously search for solutions to their issues.
Unfortunately, Zoom remains tight-lipped about what these applications will look like in practice.
However, the latter application may involve an integration between Claude and the Zoom Virtual Agent, a native solution to the Zoom Contact Center.
Such an integration may increase the scope of bots, allowing bots to answer questions that go beyond the conversational flows the contact center has already created. Yet, this is only one possibility.
When Zoom brings it to market, the innovation could look different.
That is part of the excitement that swirls around these new LLM-driven AI assistants; they bring many more possibilities to life.
From auto-summarizing customer conversations to auto-generating knowledge articles, LLMs have opened up significant room for contact center innovation.
As such, Zoom customers can likely expect much more innovation to come, especially with the vendor investing an undisclosed amount of money into Anthropic through its venture arm.
Get to Know Claude
Claude delivers a similar user experience to ChatGPT, which is unsurprising given its creators left OpenAI soon after the launch of ChatGPT-3.
In doing so, they established Anthropic and built Claude as a challenger to technology that took the world by storm.
Yet, the two models differ in one critical area: how they were trained.
Most LLMs – including ChatGPT – learn from human feedback. So, teams give them prompts, rank the outputs, and – in doing so – train the model.
Eventually, the model picks up on human preferences and optimizes its responses.
However, Anthropic took a different approach when training Claude, known as “Constitutional AI”.
This approach skips over human input and has AI supervising AI.
The supervising AI follows a set of rules – i.e. “constitution” – to automatically identify when responses may be unethical, prejudicial, or harmful.
As such, the only input from humans during the training process is creating that set of rules.
Anthropic hopes this approach allows them to innovate faster – as will Zoom, which may gain an advantage over many of its rivals that prioritize GPT-based innovation.
Nevertheless, Zoom is far from throwing all its eggs into one basket.
Zoom’s AI Strategy
While Zoom seems set to implement Claude across its portfolio, the vendor also aims to establish relationships with other AI pioneers.
In doing so, Zoom will look for new ways to combine its in-house AI models with those from AI innovators and its customers.
The vendor has already done this in the latest release of its Zoom IQ solution – which combines AI from Anthropic, OpenAI, and Zoom’s NLP.
By building up such a deep portfolio, Zoom may deliver on a “horses for courses” strategy.
For instance, if one model provides excellent speech-to-text accuracy, Zoom may deploy this for meeting transcriptions. Meanwhile, if another does a superb job of gauging conversational context, it may use this for contact center use cases.
In taking this open approach to AI, Zoom hopes to pick and choose from a plethora of models as a means of accelerating its innovation curve.
Perhaps this may add complexity to integrating systems, sharing data sets, and building UIs.
As such, Zoom must avoid passing this complexity onto its customers, which could jeopardize its conventionally high brand loyalty rates.
Nevertheless, Zoom has undoubtedly weighed up the pros and cons. After doing so, it likely sees differentiated AI as a critical way to stand apart from the crowd.
Such differentiators are critical within the CCaaS and UCaaS markets that it operates within.
