Artificial intelligence entering business operations
The visit to Alibaba in Hangzhou brought a discussion about what happens when artificial intelligence starts taking part in running a business. Researching a market, organizing requests, finding suppliers and preparing a proposal appeared as parts of a connected workflow.
Agents, collaboration between agents and OPCs were the most interesting parts of that conversation. For founders, they raise a question about the structure needed to turn an idea into an operation: how much work can one person coordinate with systems that can carry out tasks?
From an answer to a completed task
The presentation described the evolution of AI through its ability to do things on a computer. An agent can receive a goal, access tools and carry out steps in a task. The expected output might be organized research, a prepared proposal or a monitoring routine.
To do that, the system needs access to the applications where the work takes place. The conversation covered connections through APIs, MCP and interaction with interfaces. These approaches help connect the model to operational data and tools.
The ability to take action makes process design even more important. The goal, permitted access, expected output and points where someone should check the result all need to be defined.
Several agents sharing the work
Another central topic was the coordination of specialized agents. A larger task can be broken into steps, with agents working in different roles, sharing information and bringing their results together.
Research, execution and review each need their own criteria. Organizing these roles also means managing how context moves between steps: what an agent needs to receive, how its output will be assessed and what should happen when a question or an error arises.
For a small business, this organization can expand its capacity to get work done. The benefit needs to be assessed across the full process, including review and rework, to understand how much of the task was actually resolved.
An example from the inbox to the proposal
One example in the presentation connected requests received by email with finding suppliers, preparing proposals and responding to customers. The supplier described this sequence as work that one person could coordinate with AI support.
The example is useful because it connects agents with a familiar routine. A request arrives, needs to be understood, compared with sourcing options and turned into a commercial response. Each step depends on context produced by the previous one.
This was a use case presented by the supplier. Assessing it in a business means checking proposal quality, the data used, the terms offered and the exceptions that still need human intervention.
Alibaba as a case of commercial application
This direction also appears in the publicly described Accio Work product, announced by Alibaba International in April 2026. The company describes a platform with specialized agents for tasks such as research, design, sourcing and store operations. It illustrates how commercial knowledge and tools can be organized around a business goal. Official Accio Work announcement.
Alibaba's context helps make the discussion concrete: commerce, suppliers and operational routines provide defined problems for agents to address. The public product description adds context to the visit, without turning the scenarios presented into measured results.
OPC and the capacity one person can coordinate
OPC, short for One Person Company, came up in this discussion about a business run by one person. With AI support, that person can coordinate tasks performed by agents and access capabilities that previously required a larger organization.
The example started with someone who had a good idea or a product, but no marketing, sales or manufacturing teams of their own. The proposal was to connect these functions through agents, tools and other services.
This possibility expands the choices available to someone starting a business. The founder remains responsible for understanding the customer, setting priorities and delivering. Access to more capacity changes what they can test and put into operation before expanding the organization.
A2A as a direction for the next step
The conversation also covered A2A, agent to agent, as a vision of agents interacting with those of other businesses. This extends beyond sharing tasks within one operation: it involves exchanging information and passing work between systems at different companies.
This was presented as a future direction. A commercial relationship requires identity, permission, clear terms and accountability for what has been agreed. Technical coordination is one part of that work; significant decisions and commitments need defined controls.
Choosing models and setting limits on execution
In the closing questions, the presenter described choosing models according to the task, using different options depending on complexity and the type of work. This brings the discussion to operational economics: cost, quality and time need to make sense for each output.
The question about payments also revealed an important limit. The response indicated that direct processing still depended on integration, compliance and development requirements. The conversation did not establish a ready autonomous payment workflow.
For businesses considering agents, a useful starting point is a well-defined routine, clear execution permissions and observation of results before access is expanded. Sending a proposal, making a purchase or changing data each needs authorization appropriate to its consequences.
A connection to the supplier network
At the end, 1688 added the sourcing dimension. Alibaba's domestic wholesale platform brings together suppliers from different sectors. The presentation used stocked products and small orders as examples of access to this network, subject to each supplier's terms. It complements the discussion of OPCs: turning an idea into a product also depends on connections with those who can supply it. Official 1688 profile.
The lesson for people building a business
The visit leaves founders with a practical question: which part of the operation can be organized as a clear sequence of tasks, data, tools and decisions? This is the level where agents can begin to produce an observable change.
Alibaba's case connects the discussion with concrete commercial activities. The opportunity is to combine knowledge of the problem with execution capacity and access to a network of services. For an OPC or a small team, that combination can expand what is possible to build, test and deliver.