Rox Raises $100M Series B at $1.3B Valuation to Scale AI Sales Agents

Rox has raised $100 million in a Series B round that values the San Francisco startup at $1.3 billion, Reuters reported on November 11, 2026. General Catalyst and March Capital co-led the financing, with existing investors Sequoia Capital and Factory participating alongside new backers Databricks Ventures and Prosperity7 Ventures.

The company plans to spend the capital on AI research and custom-model development while expanding its enterprise customer base across healthcare, industrial, financial services, energy and technology. That's a broad target list for a company that only launched commercially about a year ago — and one that claims annual recurring revenue grew more than tenfold during that first year. Its customers include Puma, Groq and Ramp.

What does Rox actually sell?

Rox is pitching more than outbound lead generation. Its platform documentation describes one connected system covering prospecting, qualification, engagement, closing, renewal and account expansion — meaning a single deployment handles both new-business development and revenue from existing accounts.

The platform can draw on CRM records, email, calendars, Slack, call transcripts, documents and internet sources. A knowledge graph resolves entities and conflicting values across those sources and infers relationships, while a governance layer applies a single access model across systems. The stated goal: give AI agents a shared account context instead of leaving them to work from isolated sales records.

Before any contact enters an outreach workflow, the Outbound Agent validates the person's current role and email address, supplementing internal records with data from more than 17 enrichment providers, according to the company's documentation. Search results carry a matched or rejected label with reasons for exclusion. Approved matches transfer into an outreach sequence, contact list or lead list, and the agent can handle responses and book meetings.

Control stays with humans. Teams can keep every send under manual review, pause autopilot and inspect booked meetings. Changes to a running agent remain drafts until the user saves or discards them, and Rox states its automation runs within user-defined parameters requiring initial configuration and continuing oversight.

What results have customers reported?

Rox's customer-story index offers a handful of numbers:

• MongoDB took 45 days to move from prototype to production
• Ramp's conversion rate rose 10% to 15%
• Couchbase reached adoption above 90% among more than 250 representatives
• Puma Energy uses Rox to monitor customers in key markets

The examples span deployment speed, conversion, workforce adoption and account monitoring — not a single uniform measure of value. Rox itself cautions that the figures reflect individual configurations and may not represent other deployments. Actual results vary with data quality, CRM configuration, outreach volume, market conditions and target audience, so the cases don't establish a standard outcome for new customers.

Who is building it — and who's backing them a second time?

The research spending fits an agenda already laid out on Rox's research page: task-specific models and agent architectures, systems for evaluating agent behavior, and production infrastructure intended to make agents reliable in real-world revenue environments. Custom-model development sits inside that broader engineering program rather than standing alone as a model-building exercise.

Rox's company page identifies the founding team: Ishan Mukherjee as cofounder and CEO, Shriram Sridharan as cofounder and CTO, Diogo Ribeiro as cofounder and CGO, and Christopher Ré as cofounder for research.

General Catalyst is a returning lead. On November 19, 2024, it announced it had led Rox's Series A, saying more than 35 enterprise sales teams had adopted the product during its public beta — including MongoDB, Confluent and Ramp.

The Series B gives Rox capital to pursue model development and a larger enterprise footprint at the same time. The central test now is whether its early revenue growth and individually reported customer outcomes can be reproduced across the wider range of industries it's targeting.