As enterprise software vendors race to put AI agents into every workflow, the cost of running those agents is becoming as important as what they can do. Zoom believes it has an answer in context. According to the company’s internal benchmarks, its new Unified Context Layer reduced costs by between 45 and 65 per cent compared with standalone agents that have to repeatedly parse the same documents to understand a task.
The layer, unveiled at Zoom’s annual Zoomtopia conference in San Jose this week, links conversation history from meetings, phone, chat and contact centre interactions with relevant documents and data, including activity from connected third-party applications. The aim is to give both people and AI agents enough of the picture to act on what was discussed rather than reconstruct it.
Speaking to GEC Newswire, Mohannad AlKalash, Vice President of Zoom for the Middle East, Türkiye, Africa and Pakistan (METAP), said the company’s case rests on depth rather than breadth.
“Our approach is not to compete through a checklist of similar features, but to solve customer problems in depth and help people move from conversation to completion,” he said.
Competing on depth, not a checklist
Zoom is already a part of a rather crowded market. Most of its largest rivals have announced similar agentic capabilities over the past year, and the company’s launches arrive in a market where buyers are already weighing overlapping promises from much bigger platforms.
AlKalash did not dismiss the pressure.
“Competition is healthy because it raises the bar for the entire industry,” he said. What Zoom is offering instead, he explained, is a layer that “brings together documented business knowledge, general model knowledge, and near-real-time context from meetings, phone, chat, contact centre, events, and revenue workflows.”
The practical value, in his view, lies in continuity. “This shared intelligence helps people and agents understand what was discussed, retain the relevant context, and take the next action,” he said.
He also pointed to openness as a deliberate choice. Rather than positioning Zoom as a replacement for core business systems, the company is building it to sit alongside them. “We are also deliberately open,” he said.
“Zoom complements systems of record such as Salesforce, ServiceNow, and Workday, while our MCP servers and APIs make Zoom capabilities and context available through third-party AI platforms.”
At Zoomtopia, the company announced connections to AI assistants including Claude, Codex, Copilot and Muse. AlKalash summed up the pitch as a combination rather than a single feature. “That combination of embedded intelligence, dynamic conversational context, openness, and enterprise-grade governance is where we believe Zoom stands apart,” he said.
Control built in from the start
The more authority AI agents are given, the more enterprises want to know what those agents can see and do. AlKalash was clear that this was not an afterthought. “Privacy, security, and responsible AI are built into our products from the outset,” he said.
The Unified Context Layer, he explained, “is designed to respect existing permissions, sensitivity labels, and compliance policies as it brings together business knowledge, model knowledge, and dynamic context from conversations.” On top of that sit administrative controls.
“Administrators can apply custom output guardrails, manage AI centrally, review sources and audit trails, require human oversight for high-stakes actions, and shut down agents when needed,” he said.
Several of these controls arrive on a staggered timeline. Centralised AI management and deepfake risk detection, designed to flag potential synthetic media in hiring, customer service and meetings, are expected in December. Account-level AI guardrails are due in the first quarter of 2027.
Synchronisation with Microsoft Purview sensitivity labels is already available to select Zoom Workplace customers, and support in Zoom Compliance Manager for archiving and monitoring AI-generated summaries, notes and documents from My Notes and Canvas is expected in January.
“Our goal is to make AI genuinely useful and transparent, while giving organisations and their people meaningful control over how it is used,” AlKalash said.
From meetings to action
The governance push sits alongside a broader set of launches that show how far Zoom wants to extend beyond video calls. New tools include personalised transcripts for people who miss parts of a meeting, notes that link shared screens and slides to moments in a searchable transcript, workflows that turn meeting summaries into follow-up tasks in tools such as Salesforce, and ZoomMate Agent Teams, in which specialised AI agents work together to complete tasks.
On the customer experience side, Zoom is adding dynamic customer profiles and kiosk-based virtual agents for physical retail.
The company says users of its AI tools report completing tasks twice as fast, with 72 per cent citing high or extremely high business impact, against a market average of 63 per cent, according to its own 2026 survey of US knowledge workers.
Whether those gains hold up across the METAP region will depend less on how many features Zoom ships and more on whether enterprises trust its agents with real decisions, which is precisely where AlKalash is placing the company’s bet.
