Zoom has thrown down the gauntlet in the enterprise artificial intelligence (AI) race, with the announcement that AI Companion 3.0 will be available at no additional cost to all paid users.
The strategy involves agentic AI capabilities designed to transform workplace collaboration – yet fundamental questions remain about whether Zoom is betting on the right horse.
“We want all of our customers to be able to use all of our AI features, not just a selected few,” said Steve Rafferty, Zoom’s head of EMEA and APAC, at its annual Zoomtopia conference in London ahead of the annual conference.
“That’s why AI Companion is included at no additional cost for paid Zoom Workplace customers.”
This approach tackles one of the most significant barriers to enterprise AI adoption: cost. In a market where organisations have invested $30-40bn in generative AI (GenAI) with mixed results, removing the price barrier could prove decisive.
The centrepiece of Zoom’s AI strategy is perhaps best demonstrated in its contact centre operations, where the company has become its own most compelling case study.
At Zoom’s own customer service operation, “97% of the initial inquiries into the website support chat or support calls are actually being handled by AI”, according to Ben Neo, head of contact centre and customer experience sales for EMEA at Zoom.
This isn’t merely about deflecting simple queries – the AI reportedly handles complete resolution of customer issues, from initial contact through to closure.
Implementation challenges
That success has driven a much broader roll-out of Zoom’s Customer Experience platform. Zoom Virtual Agent adapts to specific industries, with a healthcare variant due in January 2026 that connects directly with electronic health records for patient intake and scheduling. Organisations can also upload voice samples to create bespoke conversational AI that preserves their brand voice.
Contact centre innovations extend beyond customer interactions to agent wellbeing, an area where Zoom sees significant potential. Rafferty described how the platform can proactively suggest wellness breaks for agents after challenging customer interactions. “It’s so easy to burn out in a stressful environment like that,” he said. The system recognises when agents need to be pulled from the queue to reset and decompress, automatically making these decisions while ensuring administrators remain in control.
However, Zoom’s focus on front-office collaboration tools sits somewhat at odds with emerging evidence about where AI delivers the greatest return on investment. Massachusetts Institute of Technology (MIT) research suggests that while organisations typically allocate around 50% of their AI budgets to sales and marketing functions, the highest return on investment often comes from back-office automation in areas such as finance, procurement and operations. The study found that 95% of organisations are seeing zero return on their AI investments, with many stuck in pilot phases, unable to translate capabilities into measurable business outcomes.
Zoom’s response to this challenge is its “federated approach” to AI development, which differs markedly from the integrated systems offered by tech giants like Google and Microsoft. Drew Smith, Zoom’s head of government relations in the UK, explained that this flexibility allows Zoom to “be very nimble, and respond and withdraw a model and replace it with other models” if performance issues arise. The approach combines proprietary models with third-party and open-source alternatives, potentially offering best-of-breed capabilities across different use cases.
Yet this distributed model also introduces complexity, particularly in terms of compliance with regulations such as the European Union (EU) AI Act, which requires transparency in AI decision-making processes. The complexity stands in sharp contrast to the straightforward success Zoom has achieved in its contact centres. Customer service here offers clear metrics and defined outcomes, while the company’s broader AI strategy ventures into more uncertain territory through its federated approach.
Smith acknowledged the compliance challenges inherent in the federated approach. “We are then also trusting that they’ve done all of the right things in terms of risk profiling, in terms of regulatory compliance,” he said.
For European organisations operating under strict data protection and AI governance requirements, this distributed accountability model may prove problematic, particularly when AI providers are based outside the EU.
AI strategy
AI Companion 3.0 combines knowledge from internal sources – meeting transcripts, chat history and shared documents – with external market research and industry data through what Zoom terms “unified, context-aware search”. Users can generate comprehensive reports drawing from both internal and external sources, with the AI grasping their specific priorities and recent discussion topics.
For organisations requiring more sophisticated customisation, Zoom has introduced Custom AI Companion, a low-code builder priced at $12 per user per month. This add-on allows administrators to create tailored AI services and access comprehensive tooling libraries. The pricing structure suggests a more nuanced strategy than the “free” headline implies – basic features serve as a gateway to premium services where Zoom expects to generate revenue.
The timing of Zoom’s announcement is particularly significant given the broader challenges facing enterprise AI adoption. The MIT study highlights a “GenAI Divide”, where a small percentage of organisations achieve significant returns while the vast majority remain stuck with no measurable impact.
Interestingly, the research found that while only 40% of companies have purchased official AI subscriptions, 90% of employees use personal AI tools for work – a “shadow AI economy” that suggests individual adoption far outpaces corporate deployment.
Zoom’s strategy appears designed to bridge this gap by embedding AI directly into workflows where employees already operate, rather than requiring the adoption of entirely new systems. The company’s focus on collaboration tools aligns with user behaviour, and could help organisations capture the productivity gains that employees are already achieving through personal AI use.
Reality check
Yet questions remain about whether this approach addresses the fundamental challenges that have led to widespread AI implementation failures. The MIT research suggests that successful AI deployments require process-specific customisation and integration with existing business workflows – capabilities that may be better suited to back-office operations than front-office collaboration tools.
For European decision-makers, Zoom’s federated AI model presents both opportunities and risks. While the approach offers flexibility and potentially faster innovation cycles, it creates dependencies across multiple AI providers with varying compliance standards. In an environment where the EU AI Act demands clear accountability and transparency, organisations must carefully evaluate whether they’re comfortable with distributed responsibility for AI safety and data processing.
The contact centre success stories are compelling, but they represent a specific use case where AI can handle structured interactions with clear resolution paths. Whether similar results can be achieved across the broader spectrum of knowledge work remains to be seen.
The challenge for Zoom – and its customers – will be demonstrating that front-office AI tools can deliver the measurable business outcomes that have so far proven elusive for most AI adopters. Zoom’s bet on accessible, integrated AI represents a significant shift in how enterprise software companies approach artificial intelligence. By removing cost barriers and focusing on seamless integration with existing workflows, the company is positioning itself to capture organisations that have struggled with more complex AI implementations.
Whether this approach can help bridge the broader challenges of enterprise AI adoption will depend on execution – and on whether Zoom’s customers can move beyond impressive demonstrations to achieve sustained productivity gains and business outcomes.
