Agents, voice and social capital: what comes after AI that answers

I am in Beijing, China, attending iRTE 2026, an Agora event bringing together more than 7,000 people around artificial intelligence and real-time interaction. During the panel on personal agents, Tony Wang, Agora’s cofounder, raised a question that stayed with me: if my agent can talk to thousands of people, what still makes a relationship valuable?

“What’s the next scarce resource?”

Tony Wang, Agora’s cofounder, speaking during the panel. Originally in English.

He imagined an agent contacting thousands of people on LinkedIn and compared it with someone taking the time to write a message personally. The question opens a discussion about social capital: the trust we build, the commitments between people and the value of a relationship that continues after the first introduction.

My bet is that we will express more goals in natural language and delegate larger parts of their execution. Voice could accelerate this shift. Alongside that capability, we will need to think more carefully about the work we want to automate and the relationships we want to cultivate.

What a connection becomes worth

In his response, Eugene Wei noted that removing friction has guided much of digital product development. Yet certain kinds of effort also bring people closer. Mutual commitment and the work of doing something together help form communities. Removing every difficulty cannot be the only criterion.

This perspective interests me a great deal. We can use AI to organize a meeting, recover the context of a conversation and prepare a presentation. The time saved can create room to listen more carefully to someone, follow through on a commitment or help with a problem they are facing.

At the same time, multiplying messages without paying attention to context also multiplies the effort required of the people receiving them. Reaching someone and building a relationship with that person are different stages. Having a name on a list still tells us little about what we can do together.

That is how I see social capital in this setting: relationships that allow us to learn, ask for help, contribute and build with trust. Technology can help bring people together. The value of that connection depends on what we do with it.

When we built Shawee’s community in Silicon Valley, bringing people together came before expanding the commercial offering. Meetings with developers created space to share knowledge and get to know the ecosystem. That experience helps me approach today’s discussion: tools can make the first contact easier, but building a relationship still requires participation and contribution.

What changes when conversation becomes the interface

The panel also discussed the next access point to technology: phones, glasses, earphones and devices capable of understanding their surroundings. The question that interests me is how much of our context a system can perceive and what it can do with it.

Someone might show a component while explaining a problem or correct a piece of information halfway through a sentence. The interaction combines speech, images and intent. Designing for this requires understanding how people converse and work.

The official program brings these areas together: conversational AI, multimodality, infrastructure, applications and devices. The discussion encompasses the entire experience.

AI needs to know when to stop talking

In a conversation, we need to interrupt, return to a point or correct information. The system needs to incorporate that change without making us start all over again.

If the AI takes too long, it feels as though the connection has dropped. If it responds too early, it cuts the explanation short. If it talks over a correction, the person loses control of the conversation.

Agora works at this layer with interruption handling, noise reduction, echo cancellation and connections to different models. These are capabilities described by the company, to be evaluated in each application.

For anyone building a product, here is a useful test: interrupt the AI, change a detail and ask it to continue. Repeat with background noise and a weaker connection. Does the task get completed?

From customer support to execution

Imagine someone wanting to reschedule a service appointment. They explain why, provide their available times and mention a restriction on access to the building. An assistant can explain where to find the form. An agent connected to the operation can check the schedule, verify the conditions, present options and, after confirmation, record the change.

The conversation connects with data, rules and tools capable of producing a result.

The system needs to know which record to consult, what it can change, when to ask for confirmation and what to do if the schedule is unavailable. It needs to distinguish intent from authorization and an attempt from a completed action.

I see a major opportunity for entrepreneurs in this transition. Understanding a process deeply can be more valuable than adding another generic AI feature. People who understand a business’s exceptions can see where a task gets stuck and what needs to happen for it to be completed.

The agent needs to earn the next permission

Allowing AI to change an appointment, send a proposal or take on an obligation creates a greater responsibility than asking it for a suggestion.

I believe in autonomy built in stages. The agent observes, suggests and prepares an action for review. For clearly defined tasks, it receives permission to execute. The results help determine where to expand that permission and where to keep a person involved.

That trust depends on something very concrete: being able to understand what happened. Did the agent consult the right information? Was the change recorded? Did something fail? Was the case passed to someone with enough context to continue?

A reassuring response can conceal incomplete information. The product needs to make uncertainty visible and know when to ask for help.

Infrastructure is also part of the experience

Agora, Inc. is a Nasdaq-listed company trading under the ticker API. The group brings together the operations of the Agora brand, serving the global market, and Shengwang, serving the Chinese market. Its background in real-time communication helps explain its place in this conversation about AI.

The experience depends on communication, the model’s response time, data retrieval and the systems that execute tasks.

For the user, it all feels like a single experience. If a query gets stuck or the person has to repeat the story to someone else, the task remains difficult.

There is also the cost of running the operation. I would track cost per resolved task, rework and the quality perceived by the people using it. That is how a technical possibility starts to become a sustainable business.

The opportunity I see for Brazil

Looking at what is happening here raises a practical question: which problems in our own context can we solve better with these capabilities?

There is room in customer support, education, commerce and internal workflows. A team can reduce the effort needed to retrieve scattered information. A student can get support to practice a conversation. A company can better organize the path from a customer’s request to delivery. Each situation requires testing language, data, consent and limits in the context where the product will be used.

Bringing people who understand these problems together with people who can build solutions is part of NewHack’s work and AI House Brasil’s mission. Agora is the first sponsor of AI House Brasil and NewHack’s strategic institutional partner. This relationship connects our ecosystem with a global infrastructure and conversational AI company.

For me, this partnership becomes meaningful when it creates the conditions to learn, experiment and build. It is also an application of the discussion about social capital: bringing people, knowledge and infrastructure together around problems worth solving. The value needs to show up in the work we can accomplish together.

I leave this discussion in Beijing with two questions: what task will someone be able to trust your product to handle from beginning to end? And what will we do with the time and capacity this technology frees up?

I want to see this capability lead to better products and more people able to build. I also want it to create more room for relationships that need our presence. The next phase of AI may become increasingly good at conversing and acting. It is up to us to decide what is worth doing with it and whom we want to build with.

An account by Rodrigo Terron and NewHack, expanded with context and information from external sources.