Why vCon?

Conversations AI can use

Making conversations useful to AI

AI can listen to a conversation, transcribe it, summarize it, and respond. The harder question begins after the interaction ends: what record travels with the conversation, who can use it, and how can people understand what happened next?

A vCon is one answer to that question. It gives a conversation a portable record that applications and AI systems can work with without leaving the context scattered across separate systems.

Connect

If you are building with Voice AI, working with conversation data, evaluating agents, or trying to understand where vCons fit in your work, connect with Jeff.

What is a vCon?

A vCon, short for virtualized conversation, is a JSON-based container for conversational data. It can bring together the people or systems involved, the dialog itself, analysis, and related attachments in one portable record.

The dialog might be a phone call, a chat, a video meeting, or another exchange. It can include a recording, a transcript, or both. The point is not to turn every conversation into an AI product. The point is to give the conversation a usable record when it needs to move between systems or inform work that follows.

A vCon is not a chatbot, a model, or a replacement for the system that carried the call. It is a way to represent the conversation and its context.

Why this matters now

As AI agents participate in more conversations, the important questions are becoming clearer. How do we preserve consent? How do we prove it? Who can use a conversation, and for what purpose? When an AI agent acts on something someone said, how do the people it represents understand what happened?

Those are not side questions. They belong in the design of an AI communication system from the start.

The IETF VCON working group is active because organizations already exchange conversations, transcripts, recordings, and related data through many incompatible systems. Its work concerns a JSON-based container for conversational data. It is active standards work, not a finished promise that a format alone resolves privacy, trust, governance, or legal obligations.

What a useful record can carry

A vCon can provide a common home for:

  • the parties involved in the interaction;
  • dialog such as recordings and transcripts;
  • analysis associated with the interaction; and
  • attachments and related material that give the interaction context.

That matters when a conversation has to survive a handoff: from a communications system to a data owner, to an analysis service, or into the work of a person trying to understand what happened.

From conversation to work

Consider a customer describing a problem. The call should not have to disappear the moment it ends. A useful record can be examined alongside the customer information, documentation, configuration data, and code or operational work needed to investigate the issue.

That does not make the record infallible. It makes the original context available for examination rather than forcing every person and system downstream to start from a partial summary.

This is where the discussion becomes practical. Voice quality, reliability, identity, permissions, the system that participated in the exchange, and the people affected by the outcome can all matter to whether a system understands the conversation responsibly.

Keep the person in view

A conversation can be valuable without becoming something to exploit. The person who spoke has a stake in what happens next.

That is why conversations about vCons bring developers, enterprise builders, attorneys, network operators, and the people responsible for customer relationships into the same room. A developer may see an implementation question. An attorney may see permission or accountability. A network operator may recognize a problem that affects the quality or reliability of the underlying interaction. Those perspectives belong together.

The work is not finished by creating a container. It starts with deciding how the people, systems, permissions, and evidence surrounding a conversation should be handled.

Sessions, Events, and AI-agent conversations

Where to learn more

Read The Pulver vCon Report for Jeff’s ongoing writing on Voice AI, vCons, consent, trust, and the future of AI communication.

Explore Conserver for vCon technical documentation and practical concepts.

Follow the IETF VCON working group for the public, evolving technical work.

Visit The vCon Foundation to learn about the education and ecosystem work around virtualized conversations.

Learn more about vCon

Read the vCon material at pulver.com for additional background on the standard and its development.

Explore vCon on pulver.com

Conference Partners