What it is
Rox is an agentic CRM — a “revenue operating system” that deploys a fleet of AI agents on top of your existing CRM rather than asking reps to feed one. You assign each named account an agent; it ingests email, calendar, call recordings, and CRM history, holds a live picture of every deal, researches prospects, drafts outreach, and writes its findings back to Salesforce or HubSpot on its own. Reps work in natural language instead of forms and dropdowns. Founded in 2024 by ex-New Relic growth lead Ishan Mukherjee and backed by Sequoia, General Catalyst, and GV, Rox closed a round led by General Catalyst at a $1.2B valuation in March 2026 — about two years from founding, on a reported ~$8M ARR run-rate. Its stated target is the Global 2000.
Why it shows up in RevOps stacks
- It acts on the CRM instead of being the CRM. Rox doesn’t replace Salesforce; it sits on top and runs the busywork — account monitoring, prospect research, meeting briefs, CRM hygiene, RFP autofill — as autonomous Agent Actions. The pitch is that the system of record stops being a data-entry tax on reps.
- One agent per account, always on. Each agent watches its portfolio for buying signals and risk, surfaces the next action, and updates fields without a human prompting it. That is the structural difference from a copilot you have to ask.
- It plugs into the stack you already run. Salesforce, HubSpot, Zendesk, Slack, Gmail/Outlook, Google Workspace/M365, and warehouses (Snowflake, Redshift, BigQuery), plus MCP and an API for custom wiring. MacOS and iOS apps put the agents where reps work.
Pricing reality
Rox is one of the few agentic-CRM vendors that publishes a price. Billing is usage- and outcome-based, metered in Agent Actions — one research task, one meeting brief, one insight each burns actions:
- Starter — free, ~2,000 Agent Actions/mo, ~10 accounts
- Core — from $50/mo, ~5,000 Agent Actions/mo
- Enterprise — custom; unlimited accounts, advanced controls, the tier the Global 2000 target actually buys
The published tiers make Rox cheap to pilot, but the real deployments are Enterprise contracts negotiated on action volume and seats — so model your spend on how many agents you will run against how many accounts, not the $50 headline. Unused actions do not roll over, and a Starter team that exhausts its allotment loses new tasks until renewal.
Best for
RevOps and sales leaders at mid-market-to-enterprise teams who already live in Salesforce or HubSpot and want the data-entry and account-research load taken off reps — without ripping out the system of record. It is the right call specifically when CRM hygiene and rep admin time are your bottleneck, not top-of-funnel volume.
Skip it if you want a single consolidated platform rather than a layer (you will be running Rox and your CRM), if your team is too small to justify managing an agent fleet, or if your gap is high-volume cold outbound — Rox is built around account intelligence and expansion, not spray-and-pray sending.
Versus the alternatives
The queued head-to-head is Clay: pick Clay when you want to build enrichment and outbound workflows yourself in a spreadsheet you control, and Rox when you want autonomous agents acting on accounts without you wiring each step. The incumbents are the CRMs themselves — Salesforce with Agentforce and HubSpot with Breeze; choose them when you would rather consolidate AI agents inside the platform you already pay for than add a vendor on top. The fastest-growing native-agentic pole is Day.ai, an AI-built CRM that rebuilds the system of record around agents instead of layering over it — weigh it when you are willing to leave your current CRM rather than augment it.
Watch-outs
- You are adding a layer, not removing one. Rox runs on top of Salesforce/HubSpot, so you keep paying for and administering both. Guard: scope Rox to the specific admin and research load you are offloading, and price the combined stack — not Rox alone — in the business case.
- Agent Actions are a real budget you can blow through. Always-on features like Insights consume actions continuously, and they do not roll over. Guard: start on a constrained account set, watch the action burn for a full cycle, and tune always-on features before scaling the fleet.
- The category and the company are young. Rox is ~2 years old at ~$8M ARR, and autonomous agents writing to your CRM is a new failure surface. Guard: keep a human review gate on agent-written fields and outreach for the first quarter, and pilot on one segment against your current CRM-hygiene baseline before trusting it across the book.