Nội dung
- Google expands Managed Agents in Gemini API: agent infrastructure enters a more practical phase
- HubSpot continues to push Breeze and automation: AI moves straight into daily operations
- 3 AI agents for marketing workflows: effectiveness comes from the data foundation, not just the tool
- The AI discovery layer is changing how marketers measure performance
- A perspective for the Vietnamese market
- References
Four new updates from Google, HubSpot, and MarTech analysis show AI moving deeper into marketing operations, from agent infrastructure and automation to how users discover information. For Vietnamese marketers, this is no longer about “using AI to write faster,” but about building systems, measuring impact, and allocating resources in an era where search is being redefined by AI.
-
Key points:
- Google is expanding Managed Agents in Gemini API with new features such as background execution for asynchronous tasks, remote MCP connections, custom functions, and credential refresh between interactions.
- HubSpot continues to push Breeze and automation, showing that AI is being integrated directly into the content, data, and operations workflows of marketing teams.
- Marketing teams that use AI agents effectively often start with a unified “source of truth” for brand, product, and response standards, rather than deploying tools first and fixing the data later.
- The AI discovery layer is changing how users search for information, forcing marketers to rethink measurement models, funnels, and customer acquisition strategy.
Google expands Managed Agents in Gemini API: agent infrastructure enters a more practical phase
In the latest update to Gemini API, Google added a series of capabilities for Managed Agents, most notably background execution for asynchronous interactions, easier connections to remote MCP servers, support for custom functions, and a credential refresh mechanism between work sessions. In essence, this signals that Google wants to move agents from an experimental concept to a tool that can operate reliably in real-world environments.
What matters to marketers is not the technical jargon, but how agents can handle long-running tasks that depend on multiple systems. When an agent can work in the background, stay connected to external data sources, and continue steps after the user leaves the session, it becomes a better fit for multi-stage marketing workflows such as data collection, lead classification, CRM syncing, or customer support.
This is also a sign that the AI race is shifting from “which model is smarter” to “who can build the better integration layer.” For businesses, the value lies in the ability to plug agents into existing infrastructure rather than replacing the entire system. Source: Google AI Blog.
HubSpot continues to push Breeze and automation: AI moves straight into daily operations
For Vietnamese businesses, these four updates point to one clear trend: AI is moving from a support tool to an operational and customer-discovery layer. Marketing teams should prioritize three things: standardizing brand data so AI can use it safely; testing agents in repetitive, low-risk tasks; and revisiting how performance is measured as users increasingly find information through AI assistants rather than only through Google.
HubSpot’s May 2026 update shows AI being embedded more deeply into the tasks marketing and admin teams use every day. Changes related to Breeze Assistant, including access to campaign data, document creation and editing, and email creation and refinement, reflect the goal of turning AI into an execution assistant rather than just a suggestion tool.
From an operations perspective, the more notable development is the agentic automation builder. While MarTech does not describe this as a full revolution, the term suggests that HubSpot is moving toward more agent-like automation: the system is no longer just running simple if-then rules, but gradually progressing toward more flexible action chains. For busy marketing teams, this can shorten the time needed to handle repetitive tasks while reducing dependence on manual work.
However, the key lesson is that AI is only truly useful when the data is well organized. Features such as cross-object filtering, archive property options, and help desk routing rules show that businesses still need data discipline and clear system governance for AI to operate reliably. Source: MarTech.
3 AI agents for marketing workflows: effectiveness comes from the data foundation, not just the tool
Another MarTech article shares experience deploying three AI agents to improve marketing workflows without replacing the entire platform or adding headcount. The key point is that each agent is assigned repetitive or information-gathering tasks, while the human team still retains responsibility for messaging, quality, and strategic direction.
The most valuable message here is that AI agent effectiveness does not come from simply “turning it on and letting it run,” but from the input foundation. The author emphasizes building a unified “single source of truth” for the brand, including tone of voice, core values, product information, competitor comparisons, and the standards by which AI is allowed to represent the brand. This is something many businesses overlook when they rush to put agents into operation.
For marketers, the practical lesson is to start with data, processes, and quality-control standards. The clearer the task assigned to an AI agent, and the cleaner the input, the more useful it becomes. Otherwise, the system may create speed, but not necessarily the right outcome. Source: MarTech.
The AI discovery layer is changing how marketers measure performance
A MarTech analysis of the AI discovery layer raises a major question: if users no longer begin their journey with a traditional search query, how should marketers change the way they measure and optimize? As AI assistants increasingly act as intermediaries in answering questions, the information discovery layer is no longer contained entirely within search engines.
This undermines the familiar funnel model built on pulling traffic from search results to a website. If AI acts as an information filter, marketers may lose some control over the first touchpoint with users. As a result, metrics such as organic traffic or keyword rankings, while still important, will not be enough to fully reflect brand visibility in an AI environment.
Instead of optimizing only for “top blue links,” businesses need to think more broadly about visibility across AI-assisted discovery layers. This is a structural change, not just a tactical SEO adjustment. Source: MarTech.
A perspective for the Vietnamese market
In other words, the competitive advantage no longer lies in whether a business “uses AI or not,” but in how well it organizes data, processes, and customer touchpoints so AI can amplify existing capabilities. This will be a race between organizations that know how to integrate and those that stop at experimentation.
See more marketing news and guides at https://marketing365.vn.
Follow more updates from Marketing365 to stay on top of the latest marketing trends.
Read more articles in the Digital Trends category.
This article focuses on the latest AI news with a perspective for the Vietnamese market.
References
- Expanding Managed Agents in Gemini API: background tasks, remote MCP and more
- HubSpot’s May 2026 updates: Breeze gets serious, automation gets smarter, and your data gets cleaner
- 3 AI agents to improve marketing workflows
- How AI discovery is changing everything marketers measure
- MarTech — HubSpot’s May 2026 updates: Breeze gets serious, automation gets smarter, and your data gets cleaner
