Welcome to Reality. Leave Your “AI Magic” at the Door.

The AI hype train is moving at terminal velocity, but the tracks are missing. We have vendors pitching artificial general intelligence that will solve world peace, and executives panicking because they think a conversational wrapper around a large language model is a strategy. In the latest episode of CRMKonvos, Ralf sat down with Dan Miller, formerly of Opus Research to discuss something that actually matters: infrastructure. Specifically, we are talking about vCons, or Virtual Conversations. It is an IETF standard that threatens to finally bring architectural integrity to the chaotic mess we currently call conversational AI.

TL;DR

If you want to watch the full CRMKonvo, please go ahead here (optimized for smartphones) or here (optimized for tablets/computers).

Else, be my guest and continue to read.

Or do both …

The “PDF Problem”: Why Your Call Recordings Are Useless

For decades, the contact center has relied on the clunky mechanics of Automatic Call Distributors, green screen terminals, and audio recordings. As Dan rightly points out, a traditional call recording is essentially the conversational equivalent of a PDF. You get a static document or an audio file that you cannot easily manipulate, query, or extract meaningful context from. It captures one part of the conversation at a specific point in time, freezes it, and historically required overnight batch processing just to transcribe it for basic analytics.

These days, we have swarms of AI agents acting on our behalf, yet the enterprise plumbing remains grossly neglected. You are trying to deliver a data stream simultaneously with historical information about that stream to build a customer profile, but the underlying data format is a relic.

Enter the vCon: Plumbing Over Poetry

The Internet Engineering Task Force (IETF) is currently finalizing the vCon standard. While the W3C handles how things look on the web, the IETF handles the engineering guts that actually make communications possible. Think of a vCon as a standardized, portable container for a conversation, regardless of whether it happened via voice, text, messaging, or a mix of multiple media over time.

This container defines the content (the literal words spoken or written) and the metadata associated with it. It makes the conversation seamlessly available to CRM and other enterprise systems, CDPs, AI analytics engines, and CCaaS platforms. It is the foundation of the “conversational graph”, a neural network-like web of nodes representing the people involved, their ideas, and the resources utilized. This is not sexy generative poetry; it is hardcore data plumbing.

And plumbing is what makes a house livable.

Flipping the Script: From B2C to C2B and the Consent Mandate

I have little interest in the standard B2C relationship where the business dictates the terms of engagement. Similar to MyTerms, the real power of the vCon lies in enabling a C2B (Customer-to-Business) model. The vCon container allows the customer to express preferences and attach explicit instructions regarding privacy and data usage.

Consider the right to be forgotten under the California Privacy Act or GDPR. If a customer wants their data scrubbed after a sensitive call with a pharmacy like Walgreens, the enterprise is obligated to comply. Traditional legacy systems cannot easily isolate what was said, identify the PII, and retroactively throw it out. A vCon makes it possible to pinpoint the exact container holding that interaction’s content and instructions, providing an auditable trail for consent and compliance. As Dan noted, you can draw a hard line on what is permitted for AI training purposes based on the customer’s explicitly captured preferences.

Revenue Acceleration: The Dealership Reality

Let us look at actual business cases instead of theory, or even science fiction. Dan highlights Strolid  and its spin-off Vconic, a company providing back-office support for automobile dealerships. By capturing conversations across a customer’s entire lifecycle within vCons, a sales manager does not just see a phone number; they see a persistent, historical record.

When a customer calls, the system knows they previously inquired about an SUV, understands their past service issues, and detects if they are getting closer to a purchase decision. This leads directly to what Dan calls “revenue acceleration”. It qualifies leads, moves deals forward, and stops the customer from feeling like an unknown entity every time they interact with your brand. If your system has had 500 phone calls with a customer and still does not know how they tick, your architecture has failed.

The Reality Check for Buyers: Three Mandates for Enterprise AI

If you are an enterprise buyer looking to invest in conversational AI for CX, stop looking at the shiny front-end and start inspecting the foundation. Here are your three mandates based on the vCon reality.

Fix Your Plumbing Before Buying More AI

Stop buying generative AI wrappers and start looking at your data infrastructure. You cannot train an intelligent system or a Retrieval-Augmented Generation (RAG) setup on dirty, siloed data. A standard like vCon provides the necessary container to make your conversational data portable, structured, and chronologically sound across all your systems. If your vendor’s idea of data integration is dumping flat audio files into an AWS bucket, show them the door.

Consent is Not a Feature; It is a Baseline Requirement

With regulations like GDPR and the California Privacy Act, you must be able to audit, manage, and delete conversational data with surgical precision. Dan rightly identified “consent metadata” as the most critical non-optional field for enterprise CX. If your current vendor cannot pinpoint a specific multi-channel conversation and scrub the PII upon a customer’s request, you are carrying massive, unnecessary compliance risk. Buy systems that respect the C2B dynamic.

Co-working is the Unavoidable Reality

We are not replacing human agents with autonomous swarms tomorrow, even though some are trying. The near future is about co-working. Agents and AI assistants must collaborate, and the AI needs accurate historical context to be useful. Do not fall victim to the “deer in the headlights” paralysis where you buy a disconnected AI tool just to appease the board. Define the playpen, establish the bumper cars, and ensure your human agents have standardized conversational memory (like vCons) to actually get the job done.