> For the complete documentation index, see [llms.txt](https://docs.rox.com/development/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.rox.com/development/about-rox/readme/revenue-operating-system.md). # Revenue Operating System Connecting enterprise data with agents that act on it, and the architecture behind that vision (originally written around Rox’s GA launch) #### **Shriram Sridharan, Co-founder/CTO** Enterprise software changes when the infrastructure underneath it changes. In **Cloud 1.0**, moving from on-premises software to the cloud changed how companies built and operated business applications. Teams could provision capacity as they needed it, and CRMs could scale without each customer maintaining the infrastructure. In **Cloud 2.0**, warehouses such as Snowflake, Redshift, and BigQuery brought data from those applications into a shared analytical environment. The CRM became one source among many. Companies gained a broader view of their business, but sellers still moved between separate tools for research, contacts, meetings, and enrichment. Much of that work reached the CRM only after it happened. AI creates an opportunity to bring those activities together. Models can interpret information and work out the steps needed to complete a task. They also need relevant context and access to tools. The system of record and the system of engagement can become more closely connected, with agents using customer information to carry out work. That is what we mean by a **Revenue Operating System**. #### The Revenue Operating System At its foundation is the **System of Context**, which brings private enterprise data together with public information. Customer context spans CRM records, support conversations, product usage, email, and Slack, alongside news, filings, and job postings. We designed the foundation to be warehouse-native and event-driven, so it can work with the enterprise data stack. Bringing this information together gives an agent a fuller picture of the customer. A meeting transcript can explain a priority that a CRM field does not capture. Product usage can add context to a renewal conversation. Public research can reveal a change that the account team has not yet discussed. Our **Revenue Agents** work on top of that foundation. They use context to perform tasks on behalf of sellers, with the aim of carrying useful information from one activity into the next. The architecture has to support both the data those agents need and the actions they take. You can see that direction in the product today. Outbound agents find prospects and assess fit, with an option to enroll good matches into sequences automatically. Agentflows take instructions and a defined set of tools, then work through the task. Rox can also turn account context into decks, documents, and reports. These capabilities put the same foundation to work on different parts of a seller’s day. We want that work to fit where sellers already spend their time. Rox is available across web, mobile, desktop, and Slack, with [voice](https://docs.rox.com/development/about-rox/release-notes#august-12th-2026) providing another way to work with the system. The interface should make it easier to use the customer context behind it. #### Built for Security and Scale As agents take on more work, access controls become part of how that work gets done. [Rox’s governance](https://docs.rox.com/development/about-rox/release-notes#july-15th-2026) applies permissions when information is requested, and agents inherit the permissions of the user they act for. Access Provenance helps explain why a user can see a record. Scale matters for the same reason. Researching one account is a different operational problem from supporting agents across an organization’s customer base. The system has to manage the data and execution behind that work as usage grows. By our public beta, we had deployed Rox across multiple regions. At our general availability launch in September 2025, we reported 450 billion tokens under consumption, 25 million agents in production, and 150 TB of governed data. Those figures describe that launch milestone, rather than current usage. #### Looking Ahead Rox became generally available on September 16, 2025. Since then, more of the work we envisioned has become part of the product. The architectural goal remains the same: connect customer context to agents that can use it, within the organization’s controls. For sellers, this should mean less time assembling information and carrying it between tools. For the teams responsible for enterprise systems, it means a way to put their data to work in the daily activities of the revenue organization. Our ambition is to earn the right to run the world’s revenue data and supercharge every seller with intelligent agents. Earning that responsibility requires a product that does useful work reliably, and an architecture customers can trust as they give it more to do. I’m proud of the team building it. They bring experience from AWS, StreamSets, Confluent, and Ramp, together with research and engineering backgrounds that have shaped how we approach these problems. Their work makes this ambition possible. If this future excites you, I’d love to hear from you: .