The Operational Twin

A new layer of business software.

A few years ago, almost no one had a conversation with software. People clicked. They filled forms. They scrolled.

Then ChatGPT happened. Then Claude. In two years, hundreds of millions of people learned a new way of using software: you tell it what you want, and it does it.

This shift is going to reach every company, faster than most companies are ready for. The customer who sends a 200-word voice note to a brand on WhatsApp is the same customer who used to fill out a form. They are not going back to the form.

There is a parallel here worth thinking about.

Around 2008, people started saying "data is the new oil." Companies spent the next decade collecting it. Most got it wrong. They built data warehouses on the side and treated data as something to be extracted later. Companies like Amazon, Netflix, and Stripe did something different. They built data into the operating layer of the business itself. They won the decade.

Conversations and AI agents are now what data was then. The companies that win the next decade will not bolt them on. They will build them in.

What it is

The Operational Twin is a live map of your business. It holds your customers, your conversations, your deals, your loans, your service requests, in one structured place. Humans can read from it and write to it. AI agents can read from it and write to it. Both work on the same picture.

That sounds simple. Nobody actually builds it.

A CRM is a place for sales reps to type. A CDP collects data for marketers to run campaigns. An ERP records financial transactions. None of these were designed for AI agents to do real work in. None of them treat conversation as a first-class thing.

Most AI products today sit on top of these systems and pretend. The agent reads from the CRM, replies in a chat tool, stores its memory in a vector database, and hopes the three never disagree. They always do. The chatbot vendor, the agent platform, the conversation analytics tool, the customer 360 dashboard: every one of them is a workaround for a missing layer.

The Twin replaces the pretending with structure.

Why now

Three things are converging.

Customers are starting to message instead of click. The volume of inbound conversations is going to multiply across every business that sells anything to anyone. WhatsApp is already the front door for hundreds of millions of customers in India alone.

AI agents are starting to do real work. Not "summarize this email" but "handle this customer's loan onboarding from offer to disbursement." That kind of work requires holding state across days or weeks. It cannot live inside a chat tool.

The cost of integration has collapsed. What used to be a six-month engineering project to connect a CRM or a WhatsApp number is now a configuration change.

Each of these alone makes a business a little better. Together they invert the assumption underneath every business application built so far: that the human is the one operating it.

What lives in it

The Twin holds three kinds of things.

It holds your customers. The real humans behind every interaction. One record per real person, even when they touch you on five channels.

It holds your business objects. Whatever your business actually cares about. A dealership cares about test drives, vehicles, service appointments. A bank cares about loans, EMIs, accounts. A telecom cares about plans, devices, complaints. The Twin shapes itself to match.

It holds your conversations. Every one of them. Structured, scored for sentiment and intent, available to any human or AI agent that needs the context to act.

Underneath all of this, every action gets recorded automatically. The Twin always knows who did what, whether the actor was a human, an AI agent, or an external system. This part matters because in a world where AI does real work, accountability is the bottleneck on adoption. No CIO is going to deploy AI agents at scale without it.

Where it fits

The Twin is not trying to replace your existing systems.

It can run alongside a Salesforce, an SAP, a core banking system, or a data warehouse. It plugs into them and stays in sync. Some companies will use the Twin instead of a CRM. Some will run it next to one. Both setups work.

What the Twin actually replaces is the pile of bolted-on AI products. The standalone chatbot. The conversation analytics tool. The agent platform. The customer 360 dashboard. None of these exist as separate products in five years. They get absorbed into the operating layer of the business, which is the Twin.

The bet

Every category of business software has been shaped by who does the work.

Spreadsheets, when individuals worked alone. CRMs, when sales teams emerged. CDPs, when marketing went cross-channel. ERPs, when finance and operations needed one ledger.

AI agents and humans now do the work together. They need a shared substrate. That substrate is going to be the most valuable layer of business software built in this decade.

That layer is what we are building.