Prompted by ELX is our bi-weekly industry-wide briefing on the AI developments that matter to the business of events. We cut through product announcements and rumour to explain what has genuinely changed in the artificial intelligence landscape, and what it really means for event leaders and their teams.
New announcements from Zoom, Microsoft and IBM bring AI closer to the systems where work happens. For event leaders, the opportunity is to improve the handoffs between conversations, decisions and delivery.
The practical opportunity in this reporting period sits between tasks. A meeting produces decisions. Those decisions inform a brief. The brief shapes communications, supplier instructions and reporting. Each handoff takes time and creates room for information to disappear.
Three recent announcements address different parts of that chain: retaining business context, running agents across models, and controlling their access. Together, they suggest a useful direction for experimentation. Start with a recurring operational problem and assess whether AI can help resolve it across the complete workflow.
This builds on our previous edition’s focus on oversight. The new development is the arrival of more infrastructure intended to make that oversight usable within everyday work.
Conversations become a source for follow-through
On 20 October 2026, Zoom announced a Unified Context Layer connecting conversation history with documents and data. Its announcement also described event analytics as available, while event setup automation was expected in November. Those are different levels of readiness.
Across business, the proposition is continuity: giving people and agents access to information that would otherwise sit across separate conversations and applications.
For event teams, the potential extends beyond producing another meeting summary. A controlled pilot could test whether an internal planning discussion leads to a more accurate action list, with each decision connected to its source and assigned to the right owner.
That is an ELX application of the announcement, rather than a demonstrated event-industry outcome. The test should include changed decisions, conflicting instructions and incomplete information. A system that confidently carries forward an outdated decision can create more work than it saves.
Choose a routine internal meeting with agreed recording permissions. Compare the resulting actions with the team’s reviewed record and measure correction time. Keep confidential peer discussions outside the experiment unless their use has been expressly agreed.
The system around the model becomes a buying question
On 28 October 2026, Microsoft announced availability of its Copilot harness through Microsoft Foundry. A harness is the software around an AI model that supplies instructions, tools, an execution environment and controls. Microsoft describes its offering as supporting work across multiple models.
The wider significance is architectural. Organisations are being offered a managed way to build agents without assembling every supporting component themselves. Microsoft’s performance claims remain vendor claims; availability alone does not establish reliability for a particular business process.
For event leaders, this changes the procurement conversation. Ask a prospective supplier to demonstrate an entire workflow using representative information, including an exception that requires human judgement.
A speaker-management demonstration, for example, should show what happens when approval is missing or two documents disagree about a session title. Evaluate whether the system pauses, identifies the conflict and preserves a usable record.
The event function does not need to select the underlying architecture independently. It does need to define acceptable outcomes alongside IT, procurement and the people doing the work.
Agents acquire identities and limits
On 27 October 2026, IBM announced agent identity and access capabilities, including generally available Agent Identity and Vault 2.2. Its cross-platform agent discovery and containment capabilities were described as technology previews.
These distinctions matter. A preview can inform a technology roadmap, but should not be treated as an established production safeguard.
For businesses, identity and access controls help specify who owns an agent and what authority it receives. For events, the useful distinction is between reading information and changing it.
An assistant preparing a registration report may need permission to read selected records. It does not automatically need permission to edit those records or contact attendees. That separation should be explicit in the brief and demonstrable in testing.
ELX’s interpretation is that these capabilities could support more proportionate delegation. Their effectiveness within any event environment still depends on implementation, integration and testing.
ELX leadership take
The strongest opportunity is to make a complete piece of work more dependable. That requires a shared definition of success across the event team and its technology partners.
Choose one recurring handoff, such as turning an approved planning meeting into an action register. Establish the current time spent, correction rate and human approval point. Then run a bounded comparison.
The decision to expand should rest on accurate completion, manageable exceptions and a clear improvement in the team’s work.
Posted November 24, 2026
Authored by Hannah Elston, Senior Manager of Partnerships, Marketing and Communications at ELX.
SOURCES
- Zoom, “Zoom advances its AI strategy with a new Unified Context Layer that helps turn meeting conversations into completed work”, 20 October 2026.
- Microsoft, “Introducing the Microsoft Copilot harness: Unlocking the potential of truly multi-model agentic work”, 28 October 2026.
- IBM, “IBM Introduces Identity and Control for AI Agents, Securing Frontier Intelligence for the Enterprise”, 27 October 2026.
.png)
.svg)
.png)
-1.png)
-1.png)
.png)
.png)
.svg)
