Three AI and Enterprise Automation Developments to Watch
A TP HCM sandbox, delivery agents and newsroom AI show how businesses can choose processes to test.
Three recent reports show AI and automation entering specific processes: testing technology with public resources, handling delivery tasks and supporting newsroom work. For businesses, the points to watch are the scope of testing, the data used and who is responsible for the results.
1. TP HCM proposes allowing businesses to use public resources to test technology
On 16/11/2026, VnExpress Số hoá reported that the TP HCM Department of Science and Technology was seeking feedback on a draft resolution for controlled testing. The draft proposes giving projects approved by the city for testing time-limited access to technical and digital infrastructure, laboratories, equipment, testing spaces and public assets suited to their needs. Businesses may also be considered for access to data held by state agencies for testing purposes.
Priority areas include AI, the Internet of Things, big data, robotics and automation, along with other technology fields. To be considered, a project must show what needs to be verified in practice or which legal obstacles require a testing mechanism. Reviewers also assess implementation capacity, economic and social benefits, and the ability to control risks. Under the draft, testing may last up to three years and may be extended once.
The conditions for using data are set out in detail: define the purpose, data types, scope, duration and method of use; assign access rights, log access and apply security measures. Regulators may disconnect access or suspend the provision of resources if they detect a safety risk or use outside the approved scope. This remains a proposal under consultation, not an automatic right of access for every business.
For smaller businesses, the report suggests a practical way to prepare: describe the product to be verified, the resources still needed and the evaluation criteria before proposing a test. The more clearly an application identifies which data is needed and who may access it, the stronger the basis for assessing risk.
2. Pedestal AI uses agents to address delivery bottlenecks
On 17/11/2026, TechCrunch AI reported that Pedestal AI had announced 4.7 million USD in seed funding to develop and deploy agent technology for supply chains. One case cited in the article concerns the delivery process of a large furniture company. The system calls customers 30 minutes before delivery to collect instructions; if the call does not connect, another system sends a text message.
According to Pedestal AI, this approach reduced missed deliveries by 20% in the cited case. That is the result of a specific project, not an improvement rate that applies to every business. The article also points to a difficulty with operational data: information changes constantly and is scattered across many legacy systems. Agents can provide effective support only when they receive relevant data and pass the information they collect to the right step in the process.
For Vietnamese businesses, the significance lies in how they choose tasks to automate. A repeatable step with a measurable outcome, such as confirming a delivery schedule, can be a starting point for assessing effectiveness. Before expanding to other steps, businesses need to know which calls failed, which instructions require staff attention and which cases must be referred to the person in charge.
3. LVMH brings OpenAI tools into its media division’s work
An announcement published on 16/11/2026 said LVMH’s media division and OpenAI had signed a partnership agreement on AI use. Teams in the division will receive OpenAI tools and ChatGPT Enterprise to develop uses suited to their daily work. The tasks cited include finding documents, comparing documents and preparing initial translations. The rollout includes training tailored to each team’s needs.
The agreement also addresses the inclusion of summaries, excerpts and direct links to original articles from the division’s publications in relevant ChatGPT responses. According to the announcement, the media division has 17 brands and 1,800 employees, including nearly 700 journalists, and reaches more than 27 million people in France each month. That scale shows why the scope of use, source citations and editorial responsibility need to be defined when AI enters the workflow.
For businesses, the lesson extends beyond journalism. AI can help staff search, compare and prepare drafts from existing documents. Specialists still need to check the content, decide which version to use and take responsibility before information reaches customers or partners.
Three actions businesses should take now
- Choose a process with an observable outcome. Small and medium-sized businesses in Vietnam can start with order confirmation, document reconciliation or document searches. Record how the task is currently handled to establish a basis for evaluating the test.
- Define data and access rights before testing. List the data sources, purposes of use and people allowed to view or edit each type of information. This helps businesses control data when testing agents or connecting additional systems.
- Put a person in charge of approval at steps that affect customers. Specify when the system may prepare content and when an employee must review and decide. Businesses should also record errors so they can adjust the process after each test.
How VASCOM views these developments
VASCOM sees reviewing a process as the first step before proposing automation. Businesses need to identify which tasks take time, which departments handle the information and which outcomes can be checked. From there, VASCOM can work with a business to define an appropriate testing scope before broader deployment while data and responsibilities are still being clarified.
For different needs, VASCOM considers agents by department and places human reviewers at steps requiring judgment or approval. Access rights to operational data need to be clearly defined, and the data should remain at the business under the agreed deployment plan. This approach ties testing to actual work and provides a basis for considering expansion.
Sources
- VnExpress Số hoá, “TP HCM businesses may gain access to public resources to test new technology”, 16/11/2026, vnexpress.net.
- TechCrunch AI, “Pedestal AI raises $4.7 million to agentify the supply chain”, 17/11/2026, techcrunch.com.
- OpenAI, “LVMH’s Media Division announces strategic partnership with OpenAI”, 16/11/2026, openai.com.