Rox pricing

About

Rox builds an AI “agent swarm” (RoxAgents) that acts as an autonomous copilot and account executive for sales reps, working inside the tools reps already use — Slack, iMessage, and CRM systems like Salesforce and HubSpot. The agents handle account research, people insights, pipeline generation, and meeting briefs on behalf of the rep. Rox Data Corp is headquartered in San Francisco (251 Rhode Island St, San Francisco, CA 94103).

Rox positions its product around “Revenue on Autopilot” — agents that continuously research accounts, monitor for signals, and discover contacts so high-performing sales teams can operate at the highest level. The platform maintains SOC 2 Type II compliance and enriches records through a wide roster of third-party data providers (HG Insights, FullEnrich, ContactOut, People Data Labs, RocketReach, and others).

Rox was founded in 2024 by Ishan Mukherjee — previously chief growth officer at New Relic and co-founder of Pixie, the observability startup New Relic acquired in 2020. It exited stealth in November 2024 with a $50M raise (a Sequoia-led seed, a General Catalyst-led Series A, and participation from GV), and by March 2026 was reported by TechCrunch to have reached a ~$1.2B valuation on roughly $8M of projected 2025 ARR, with 35+ enterprise customers including MongoDB, Confluent, and Ramp. It competes with conventional sales-engagement and AI-SDR tooling, but positions agents as autonomous labor rather than a rep productivity add-on.

Rox sells to individual reps up through enterprise sales organizations. Pricing is published openly: a free Starter tier for light users, a $50/month Core tier for active sellers, and a custom Enterprise tier for large teams.

Pricing summary : how Rox meters AI sales agents in Agent Actions

Rox uses a blended outcome-based and usage-based model metered in a single unit called Agent Actions, with two dimensions:

1. Monthly plan (base fee + action pool): Starter is Free with 2,000 Agent Actions/month; Core “starts at $50/month” with 5,000 Agent Actions/month; Enterprise is custom (Contact Sales). The allotment refreshes each billing cycle and does not roll over. The free tier makes this a freemium model at the entry point.
2. Agent Action consumption (the meter): Each agent task — account research, people insights, pipeline generation, meeting briefs — costs a defined number of Agent Actions. The action cost scales with task complexity and the AI model / data processing involved, so heavier tasks burn more of the pool.

What makes this different: Rox does not gate features by tier — every plan gets full feature access, and the only upgrade lever is action volume. The “value metric” is the work the agents do (outcomes), not seats or raw data processed. This is a deliberate departure from the per-seat licensing that has defined sales tooling, and part of a broader shift away from per-user pricing in AI products toward metering the value metric directly.

Pricing by product

RoxAgents (all plans)

Sales motions across products: PLG / self-serve for Starter and Core (“Try now” / “Start now”); sales-led for Enterprise (Contact Sales).

How Agent Actions are metered

Agent Actions are the fundamental billing unit. Each task an agent performs is counted as a specific number of actions, and that count “reflects the complexity and cost of the underlying processes” — tasks needing more advanced AI models, more data processing, or more detailed outputs consume more actions. Allotments reset on the renewal date each cycle and unused actions expire (no rollover) on both Starter and Core. Always-on features like Insights also draw down the pool, but their consumption can be customized to stay within the monthly allowance.

The list of metered tasks spans Rox’s four agent workflows: Research (account information, teams and technologies, custom research sections), Discover (clever columns, contacts, pipe-gen actions), Engage (email generation, LinkedIn message generation, pre-meeting briefings, conversational meeting briefings), and Monitor (Insights on public and private data). Because the per-action cost is published behind the “See full pricing list” accordion rather than as a single headline rate, the real cost driver is the action burn rate per task — see Hidden costs below.

Hidden costs : when the Agent Action pool runs dry mid-month

Rox’s headline is “starts at $50/month,” but the real bill is governed by how fast a rep’s agents burn through the 5,000-action Core pool. Per-action costs sit behind the “See full pricing list” accordion, and the published page does not name a single per-action dollar figure — so the cost risk is volume, not the base fee. Two archetypes show where the pool gives out.

Archetype A — a single active SDR running continuous monitoring. One seller covering ~30–40 named accounts with always-on Insights, weekly research refreshes, and contact discovery can exhaust a 5,000-action pool well before month-end, because continuous “Monitor” tasks draw down the pool even when the rep is idle.

Because actions do not roll over and the only way to top up is to renew early (which resets the full allotment at the base price), a heavy user effectively pays the base fee multiple times in one calendar month. The sticker “$50” can become $150+ for a single power user.

Archetype B — a 5-rep team on Core, mixed usage. Five reps each on their own Core plan, three of them light and two heavy, illustrates how per-seat-equivalent spend lands even without a per-seat price.

The lesson mirrors the broader bill-shock risk in usage-based AI products: the variance lives in the meter, not the plan name, and a team’s spend is set by its heaviest agents, not its average ones. Buyers used to predictable per-seat sales tooling should model the burn rate before committing.

Want to estimate your own outcome-metered AI bill? Model your monthly cost from action volume with the Rox pricing calculator, compare against a per-resolution outcome model with the Gorgias pricing calculator, and ground your assumptions in our guide to choosing the right usage metric.

Pricing evolution : Core priced up 2.5× and tiers de-gated since beta

Rox is young — it exited stealth in November 2024 — so the archived record is short but it captures one clean, material repricing. Archived snapshots from December 2024 through February 2025 are byte-identical; the live page in June 2026 tells a different story.

Cadence

Tracked range: 2024-11–2026-06. The 2025 Q2–2026 Q1 window is uncovered by deduped monthly snapshots, so the exact date Core moved to $50 is unknown — the change is bracketed between the Feb 2025 archive and the Jun 2026 live capture.

Notable changes

• 2024-11 — Rox exits stealth, launches public beta, and announces a $50M raise (Sequoia seed; General Catalyst Series A; GV participating), per Bloomberg and Sequoia. Launch pricing: Core “starts at $20/mo.”
• 2024-12 → 2025-02 — Pricing held stable; tiers gated premium data enrichment, advanced integrations (Salesforce Service Cloud, Zendesk, cloud warehouses), and governance (audit logs, org admin) by plan, with hard account caps (10 / 100 / unlimited).
• By 2026-06 — Core repriced to $50/mo and the model simplified: account caps and the feature matrix were removed so every plan now includes all features, leaving Agent Action volume as the only difference between tiers.
• 2026-03 — TechCrunch reported Rox reached a ~$1.2B valuation (round closed 2025), with ~$8M ARR projected for year-end 2025 and 35+ enterprise customers including MongoDB, Confluent, and Ramp.

The repricing in detail

The shift from $20 to $50 on Core is a 2.5× base-fee increase, but the more consequential change is structural. At beta, Rox looked like conventional tiered SaaS: a comparison matrix gated premium enrichment, governance, and integrations, and each tier had a hard account cap (10 → 100 → unlimited), so a rep could outgrow Starter on account count alone. The current page deletes both levers. Every plan now advertises full feature access and there is no published account cap — the only thing $50 buys over Free is a larger pool of Agent Actions (5,000 vs 2,000). This converts Rox from feature-gated tiers into an almost pure consumption ladder, where the differentiator is metered work rather than unlocked capability — a textbook example of the move from entitlement-gating to credit-style metering.

What’s unique : a single Agent Action meter for an agent swarm

One unit for everything an agent does. Rox collapses heterogeneous work — research, enrichment, message drafting, meeting briefs, continuous monitoring — into a single billing unit, the Agent Action. Most AI sales tools still price on seats or credits scoped to one feature; Rox normalizes all agent output to one meter whose per-task cost floats with model and data intensity. This puts it firmly in the outcome-based pricing camp rather than per-seat licensing.

No feature gating — capacity is the only lever. Every plan, including free Starter, ships all features. The only reason to upgrade is to buy more actions. This is rare: it removes the classic “I need the Pro feature, not the Pro volume” friction and makes the upgrade decision purely about consumption. It also means Rox’s expansion revenue is tied directly to how much work customers’ agents actually do.

Always-on monitoring draws down the same pool. Rox’s “Monitor/Insights” agents run continuously, consuming actions even when the rep is asleep. Few tools meter passive, background AI work against the same budget as active tasks — it makes the value metric genuinely usage-driven, but also means idle accounts still burn budget unless customers tune what agents watch.

Early renewal instead of overage billing. Rather than charging per-action overages when the pool runs out, Core users “renew early” to reset the full allotment at the base price. This is an unusual top-up mechanic — it avoids surprise overage invoices but effectively charges the base fee multiple times per month for heavy users, a quirk that shapes the real cost (see Hidden costs).

Built into the rep’s existing surfaces. Agents operate inside Slack, iMessage, email, a MacOS app, and CRMs (Salesforce, HubSpot, Zendesk) rather than a standalone dashboard, so the metered “work” happens where reps already are. This distribution choice is itself part of the pricing story: the agent swarm is positioned as labor, not software, reinforcing the shift away from per-seat licensing.

Strengths & weaknesses

Billing UX : agent activity log, admin view, early renewal

• Agent activity log — outlines the actions taken by AI agents on your behalf; you can track usage per action, including on-demand and continuous tasks.
• Admin / organization view — aggregates usage information across the team for easier organizational tracking.
• Early renewal (Core) — Core users who hit their Agent Action limit before month-end can renew early to immediately reset their allotment instead of waiting for the next cycle.
• Always-on feature controls — consumption from continuous features like Insights can be customized (what agents monitor and prioritize) to stay within the monthly Agent Action allowance.

Strategic wins : how Rox priced an agent swarm as labor

1. Picked one meter for a multi-function product

Rather than shipping separate credits for research, enrichment, and messaging, Rox normalized every agent task to one unit. This sidesteps the “five different credit balances” confusion that plagues multi-feature AI tools and gives sales the simplest possible pitch: “you only pay for agent work.” Choosing one defensible value metric is the single highest-leverage pricing decision an AI company makes, and Rox got it right early.

2. Removed feature gating to make capacity the only lever

By moving from a beta-era feature matrix to “all features on every plan,” Rox eliminated the most common cause of stalled upgrades — the buyer who wants one gated feature but not the higher volume. The upgrade conversation is now purely about consumption, which is far easier to forecast and to expand. It also makes the free tier a true product trial, not a crippled demo.

3. Free tier doubles as a wedge into enterprise accounts

A full-featured Starter plan lets individual reps inside large companies adopt Rox bottom-up, generating internal proof before a sales-led Enterprise motion. With 35+ enterprise logos (MongoDB, Confluent, Ramp) reported by early 2026, the PLG-into-enterprise land-and-expand path is clearly working alongside the top-down Enterprise sale.

4. Outcome framing matches the funding and category narrative

Pricing “agent work” rather than seats reinforces Rox’s “Revenue on Autopilot” / agent-swarm positioning, which is what attracted a $50M raise and a $1.2B valuation. The pricing model and the product story are the same story — agents as labor — which is exactly the outcome-based pricing revolution investors are betting on.

Areas to improve : the transparency and predictability gaps

1. Publish a per-action cost reference

The biggest gap is that no per-action dollar figure appears on the public page — the cost-per-task table sits behind a “See full pricing list” accordion and is not committed to a headline number. Buyers cannot estimate spend without talking to sales, which undercuts the “transparent, aligned with value” claim Rox makes in its own FAQ. A published “typical actions per task” table (research = N actions, meeting brief = M actions) would let prospects model cost the way they can with token-based AI pricing.

2. Offer rollover or a soft overage instead of forced early-renewal

Forfeiting unused actions and forcing a full early-renewal at the base price punishes both light and heavy users — light users lose what they paid for, heavy users pay the base fee multiple times a month. A modest rollover cap or a metered overage rate would smooth spend and reduce the bill-shock risk that makes finance teams wary of usage-based tools.

3. Add team/org plans, not just stacked individual Core seats

Today a team appears to be N individual Core plans, which means a 5-person team manages five separate pools and five renewal dates. A pooled team plan with shared actions and one admin budget would match how the Enterprise tier already aggregates usage, and would remove the friction of per-rep pool management documented in Hidden costs.

4. Communicate the $20→$50 change proactively to early adopters

The 2.5× Core increase and the Starter action cut (2,500→2,000) happened quietly between the Feb 2025 archive and today. Early beta adopters who anchored on $20 may feel the change. A public changelog explaining the repricing — and what the extra value buys — would protect trust, the way the best usage-based vendors handle pricing migrations.

Key takeaways

1. One unit beats many. Rox normalizes every agent task to a single Agent Action rather than juggling per-feature credits. If your product does several kinds of work, find the one unit that spans them — it makes the pitch, the bill, and expansion all simpler.
2. De-gating can be a feature. Putting all features on every plan and charging only for volume removes upgrade friction and turns the free tier into a real trial. Feature gating is not the only way to monetize tiers; capacity tiers can do the job with less customer resentment.
3. Always-on AI needs a metering story. Continuous background agents consume budget passively. If you ship always-on AI, decide deliberately whether passive work draws the same pool — and give customers controls — or you create silent bill-shock.
4. Top-up mechanics shape the real price. Rox’s “early renewal” avoids overage invoices but makes heavy users pay the base fee multiple times a month. How customers top up is as load-bearing as the sticker price; model it before you ship it.
5. Reprice transparently or pay in trust. A quiet 2.5× base-fee increase since beta is the kind of change early adopters remember. Pair any usage-pricing change with a public explanation of the added value.

UBP implications

1. Outcome metering is reaching sales tooling. Sales software has been per-seat for decades; Rox pricing “agent work” is a signal that the outcome-based model is spreading from customer support (Intercom’s per-resolution Fin) into revenue teams.
2. The agent-swarm pattern needs a composite meter. When one product runs dozens of autonomous agents doing different jobs, a single normalized unit (actions) whose cost floats with model/data intensity is emerging as the natural meter — a template other agentic vendors will copy.
3. Free-tier-as-trial plus consumption tiers is a durable AI GTM. Full-featured free entry, capacity-only upgrades, and a sales-led enterprise top end combine PLG and enterprise motions without feature-gating — a packaging shape well-suited to AI products where the marginal value is volume of work, not unlocked capability.

Sources

• Rox pricing page (accessed 2026-06-05)
• Rox homepage — product, agent workflows, integrations (accessed 2026-06-05)
• Rox docs (accessed 2026-06-05)
• Rox pricing page, archived Dec 2024 / Jan 2025 / Feb 2025 via the Wayback Machine — Core “$20/mo,” 2,500/5,000 action caps, feature-comparison matrix (accessed 2026-06-05)

Bottom line

Rox prices an autonomous sales-agent swarm the way you’d price labor: a single “Agent Action” meter, full features on every plan, and a base fee that only buys more capacity. Since its $50M-backed beta it has raised Core 2.5× ($20→$50) and dropped feature gates and account caps, betting that buyers will accept a consumption ladder in exchange for transparency — provided they can stomach the no-rollover, early-renewal mechanics that govern the real bill.

Want to compare Rox against other AI sales-tooling and outcome-based pricing? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Core repriced $20→$50; account caps and feature gates removed

Jun 2026

Core base fee raised from $20/mo to $50/mo (5,000 actions unchanged). Starter monthly actions cut from 2,500 to 2,000. The plan-comparison feature matrix and per-tier account caps were removed in favor of 'all features on every plan'; action volume becomes the sole upgrade lever.

Feature-gated tiers with account caps (stable)

Feb 2025

Wayback snapshots Dec 2024–Feb 2025 are identical: Starter $0 (2,500 actions, 10 accounts), Core $20/mo (5,000 actions, 100 accounts), Enterprise custom. Tiers gated premium data enrichment, integrations, and governance features by plan.

Public beta + $50M raise; Core at $20/mo

Nov 2024

Rox exits stealth, launches public beta and announces a $50M raise (Sequoia seed, General Catalyst Series A, GV). Pricing page shows Starter Free (2,500 Agent Actions/mo, 10 accounts), Core 'starts at $20/mo' (5,000 actions, 100 accounts), Enterprise custom — with a full feature-gated plan-comparison matrix.