AI CRM · Research Deck

Rox AI — $1.2B on $8M ARR. Stated competitor is Salesforce. Actual competitor is HubSpot Breeze and the AI-CRM flagship cohort.

Warehouse-native architecture sits in the memory layer that the cross-cohort SaaSpocalypse thesis predicts will capture value. The team ships at outlier velocity. But 11 expert calls surface a consistent “nice-to-have” pattern, and Rox’s actual customers (Ramp, Tabs, Snorkel, Rho) overlap with the ICP of Attio, Day AI, Monaco, and Reevo — where the near-term threat is HubSpot Breeze.

June 2026

1

Rox claims to span every revenue technology layer with a single “Agentic CRM” — but the market is six distinct segments with entrenched incumbents

CRM / System of Record

Salesforce, HubSpot, Dynamics

Salesforce: ~$38B rev, 150K+ customers

Sales Engagement

Outreach, SalesLoft, 11x, Artisan

Sequencing, cadences, outbound automation

Revenue Intelligence

Gong, Clari, Chorus

Gong: ~$300M ARR, conversation data moat

Data Enrichment

ZoomInfo, Clay, Apollo

ZoomInfo: ~$1B. Clay: higher switching costs

Agentic CRM ← Focus

Rox, Day AI, Monaco

Rox: ~$8M ARR · $1.2B valuation

Incumbent AI Extensions

Salesforce Agentforce, MS Copilot

Toggle-on within existing contracts

AI in Sales market: $24.6B (2024) → $145B by 2033 (~22% CAGR).[19] Rox’s bet: full-stack agentic CRM collapses point solutions into a single platform. The risk: jack-of-all-trades against entrenched specialists.

2

At 150x ARR with the wrong competitor named, Rox needs to be the memory layer for agents — or it gets bypassed by them.

Bull case — Rox sits in the layer where SaaSpocalypse value accrues

$0 → $8M ARR in <2 years; Sequoia/GC/GV re-upping at escalating valuations: Smart money stress-tested the trajectory across seed, A, and valuation rounds.[1][6]
Customer quality is not padding: OpenAI, NVIDIA, MongoDB, Ramp; GV reports teams would “revolt” if Rox were removed.[6][13]
Memory-layer position is where SaaSpocalypse value accrues: The cross-cohort thesis says agents abstract the CRM UI, value shifts upstream to memory + context layers. Rox is already there with a warehouse-native knowledge graph.[8]
Day AI’s April 2026 repositioning validates the architectural bet: Less-capitalized peer with deepest CRM founder pedigree pivoting from “Cursor of CRM” to “memory for agents” — other category-leaders converging on Rox’s lane.
2x ROI guarantee in 90 days: Bold public commitment that simplifies multi-stakeholder enterprise procurement.[8]

Bear case — Stated competitor is Salesforce; actual competitor is HubSpot + the flagship cohort

11 expert calls: “nice-to-have, not need-to-have” consensus: First cut in a downturn. Emails so poor reps copy them into ChatGPT for refinement.
HubSpot Breeze, not Agentforce, is the immediate threat: 51% of HubSpot users noticed major/significant new AI capabilities in last 12 months; 53–58% of next-gen CRM switchers would have stayed if their prior CRM had comparable AI when they left — closing the window in Rox’s growth-stage customer base.
Rox’s real customers are the same growth-stage startups the flagship cohort chases: Ramp, Tabs, Snorkel, and Rho are venture-backed tech companies, not Global 2000 accounts — so Rox competes head-to-head with Attio ($116M, 5,000+ customers), Day AI ($24M), Reevo ($90M), and Monaco ($85M), not Salesforce.
Multiple experts copy Rox output into ChatGPT/Claude for refinement: If LLMs do comparable work for free — or read the warehouse directly via MCP — Rox’s moat is workflow integration and data aggregation, not intelligence quality.

3

Four Stanford/Amazon/Confluent founders built a 14-person all-engineer team that reached $1.2B in 18 months — but the org chart has critical gaps

FOUNDERS

Name	Role	Background
Ishan Mukherjee	CEO	Amazon → Apple/Siri → Pixie Labs (acq. New Relic) → New Relic CGO→CRO
Avanika Narayan	AI Lead	Stanford PhD, Knight-Hennessy Scholar, Chris Re lab
Shriram Sridharan	CTO	Confluent data infra, pre-launch Amazon Aurora, AWS
Diogo Ribeiro	Product	Lacework, AWS, Confluent, ThousandEyes
Chris Re	Advisor	Stanford professor, data-centric AI pioneer

FUNDING

Round	Date	Lead
Pre-seed	Dec 2023	Self-funded (stealth)
Seed	Feb 2024	Sequoia[6]
Series A	Nov 2024	General Catalyst[7]
Valuation	2025	GC-led → $1.2B[1]
Total		$50M announced ($199M per PitchBook[2])

~$8M[1]

ARR (projected end-2025)

$1.2B[1]

Valuation (~150x ARR)

14-15

Team (all engineers)[8]

25M+

Agents in production

Key gap: No CFO, VP Sales, CRO, or VP Engineering named publicly. 14-person all-engineer team is world-class for product velocity but lacks enterprise sales infrastructure the GTM motion demands.

Gartner recognition: 2025 Cool Vendor for Revenue Agents. No G2 or Peer Insights coverage as of April 2026.

5

01 Company

Rox deploys per-account AI agent swarms on the customer’s own data warehouse — a structural bet that data gravity shifts away from Salesforce

THREE-LAYER ARCHITECTURE

Data Layer: Unifies CRM records, email, call transcripts, product telemetry, support tickets, and open-web signals into a warehouse-native knowledge graph. Runs on customer’s existing Snowflake/Databricks — data stays in customer infrastructure.
Intelligence Layer: One dedicated agent per customer account, continuously monitoring signals and generating recommendations. Configurable to seller, account, and organizational preferences. Always-on.
Interface Layer: Outputs surfaced as daily digests, pre-meeting briefs, in-meeting support, post-meeting follow-ups, and Command (conversational interface for ad-hoc queries).

KEY DIFFERENTIATOR

Salesforce Data Cloud requires data to reside in or mirror to Salesforce’s proprietary environment. Rox runs on the customer’s existing warehouse. Whether enterprise buyers treat this as a meaningful purchasing advantage in practice is unconfirmed.

“Every business owner across every vertical is blowing out their revenue targets for next year. Some of our customers are doubling their revenue targets, but none of them are doubling their team size.”

— Ishan Mukherjee, CEO | Bloomberg Technology[16]

“The core advantage of AI is to be able to deeply couple with the data layer. So at Rox, we built the first enterprise-ready AI agent swarm that’s powered by a warehouse-native system of record.”

— Ishan Mukherjee, CEO | Bloomberg Technology[16]

TECH STACK

Snowflake (data warehouse), AWS Bedrock (cloud), OpenAI GPT-4o mini (primary LLM), open + closed-source model flexibility, iOS mobile, Salesforce + Zendesk + Microsoft Copilot integrations.

6

01 · Company

Customers praise the engineering velocity and feature breadth, but the product is “designed by engineers, not sellers” and has no product managers on staff

CORE FEATURES

Clever Columns: Custom AI-computed enrichment columns on account/contact lists. Repeatedly cited as power-user favorite.
Plays: Templated sales motions teams can define, test, and run at scale. Top feature per Rho CRO and Snorkel AI.
Pre-meeting Briefs: Auto-generated meeting prep with attendee backgrounds, account history, competitive positioning. Most universally praised feature.
Command: Conversational agent that synthesizes across entire workspace and cites sources.
Outreach: Personalized email campaigns with automated cadences. Campaign-to-pipeline in minutes.
Autofill: Automated RFPs, InfoSec questionnaires, technical contracts. Launched at GA.

PRODUCT GAPS

No product managers or designers on staff — explains both velocity and UX gap
SOC 2 status unconfirmed — enterprise procurement typically requires Type II
Inference cost exposure unquantified — 25M+ agents × 1T+ tokens

“Rox is doing it all, and they’re doing it all in a way that is intuitive, intelligent, and it totally scales. It’s rare to find a go-to-market tool that has the engineering horsepower that Rox has.”

— VP Sales | Corporate Cards / Expense Management[3]

“The conversion rate on the Rox-influenced leads are higher than our standard funnel. We have already done break-even many times over, so it’s sort of a no-brainer.”

— CRO | Business Banking[5]

“It’s clearly designed by engineers. It’s not designed by sellers, so it’s a little bit hard and clunky to get through.”

— Head of Client Sales | Fintech

7

02 · Competitive

Rox’s stated competitor is Salesforce. Rox’s actual competitor is the AI-CRM flagship cohort fighting for the same growth-stage customers Rox is closing.

Company	Scale	Core Advantage
Incumbents — the stated competitive frame
Salesforce (Agentforce)	~$38B total; Sales Cloud ~$7-8B	150K+ customers, $6B+ R&D, embedded data gravity
Microsoft (Dynamics + Copilot)	~$4-5B est.	M365 distribution, Azure Fabric, EA relationships
HubSpot (Breeze)	~$2.6B (FY2024)	Mid-market install base; 51% of HubSpot users noticed major AI gains in last 12 months[20]
AI-native flagship cohort — the actual competitive frame
Day AI	$24M raised; ~32 headcount; GA Feb 2026[20]	Auto-captured customer memory; April 2026 repositioned to “memory for agents”; ex-HubSpot CPO + VP Product founders
Attio	$116M raised; 5,000+ customers[20]	Flexible programmable schema; MCP server; embedded AI; per-seat + credits pricing
Reevo	$90M raised; ~112 headcount[20]	Integrated GTM platform (prospecting + outreach + meeting intel + CRM); ~$2-3K/seat
Monaco (Sam Blond)	$85M raised; public beta Feb 2026; ~40 headcount[20]	Software + forward-deployed AE bundle; flat fee $25K beta / $50K list; “Sam-as-a-service”
Adjacents — same-segment but different attack surface
Gong	~$300M ARR est.	Conversation intelligence data moat; could expand into agent orchestration
ZoomInfo / Clay	$1.0B / $30-50M est.	Data enrichment; Clay is harder to switch (middleware lock-in)
Rox	~$8M ARR; $1.2B valuation[1]	Warehouse-native agent swarms, per-account intelligence, “system of intelligence”

8

02 · Competitive

Rox’s actual competitive risk is HubSpot Breeze closing the AI gap in the installed base, and Day AI claiming the memory-layer position with one-fifth the capital.

HubSpot Breeze — immediate threat to actual customer base

51% of HubSpot users noticed major or significant new AI capabilities from HubSpot in the last 12 months; 53–58% of next-gen CRM switchers say they would have stayed with their prior CRM if it had comparable AI when they left.[20]

Rox exposure: Most of Rox’s named customers (Ramp, Tabs, Snorkel, Rho, Coreweave) are growth-stage tech startups — HubSpot’s heartland. Window to differentiate on AI is narrowing.

Day AI — architectural analog at 1/5 the capital

Day AI’s April 2026 repositioning from “Cursor of CRM” to “customer memory for agents” puts it in the same architectural lane as Rox’s “system of intelligence” — both pitch themselves as the substrate agents will run on.[20]

Read-across risk: Day AI’s founders ran HubSpot’s CRM + Sales Hub; data-quality trust gap is the cautionary read for any AI-native CRM claiming autonomous writes.

Salesforce Agentforce — enterprise tier

Agentforce activates inside existing contracts at zero incremental procurement friction. Still the load-bearing threat to Rox’s stated Global 2000 ICP — but Rox’s actual deal flow does not yet show Global 2000 ACVs at scale.

Rox counter: Salesforce Data Cloud requires data in proprietary environment; Rox runs on customer’s Snowflake/Databricks. Whether enterprise buyers treat this as a purchasing advantage is unconfirmed.

ChatGPT + MCP — agents may bypass Rox entirely

Multiple experts copy Rox output into ChatGPT/Claude for refinement. If LLMs do comparable work for free — or read the warehouse directly via MCP — Rox’s moat is workflow integration and data aggregation, not intelligence quality.

Memory-layer bear: SaaSpocalypse may shift value upstream to LLMs + orchestration; Day AI runs on top of HubSpot, demonstrating an agent-substrate layer can be built without owning the CRM.

“I think it isn’t that sticky. I think people will move around a lot in this space.”

— Strategic Account Executive | Database Infrastructure

“I’ll copy and paste the email out. I’ll start putting it within Gemini or Anthropic or whatever, then I’ll start having the conversation with the LLM. It’s better.”

— Strategic Account Executive | Database Infrastructure

9

03 · Risks & Signals

ARR growth, HubSpot Breeze adoption, MCP positioning, and Day AI’s agent-substrate landing are the four signals that determine whether 150x ARR was prescient or premature.

MoM ARR growth rate — the single most decisive metric. Sustained above 15% keeps the valuation thesis viable. Below 10% for two consecutive quarters is a red flag independent of product quality. At 150x ARR, the growth rate IS the investment thesis.[1]
HubSpot Breeze adoption + AI gap-closure metrics — the more immediate competitive clock than Agentforce. 51% of HubSpot users noticed major/significant new AI capabilities in last 12 months; 53–58% of next-gen CRM switchers would have stayed if their prior CRM had comparable AI. Given Rox’s actual customer-base composition, HubSpot AI feature parity is the displacement risk.[20]
Day AI’s “memory for agents” repositioning landing. If Day AI captures the agent-substrate framing at one-fifth Rox’s capital with deeper CRM founder pedigree, the “system of intelligence” narrative position transfers. Watch for Day AI MCP integrations into customer-side agent frameworks; queryable-API usage as a third-party memory layer.[20]
MCP server / agent-framework integrations. In a world where customers want to query their CRM via Claude or Cursor, the question is whether Rox is the layer agents call or the layer agents bypass. Attio already enables this; Reevo’s missing MCP server was a “huge complaint” from its customer base. Rox needs to be on the right side.[20]
Salesforce earnings + Global 2000 logos + VP Sales hire. Agentforce attach-rate disclosure remains an external clock on the enterprise tier of Rox’s displacement narrative. WSP is the only confirmed Global 2000 public reference. CEO’s signature credential (New Relic) was a PLG motion; Global 2000 field sales is a different competency.[15]
First independent reviews + NRR cohort data. G2, Gartner Peer Insights expected H2 2026; quality and quantity will indicate real-world adoption beyond promotional testimonials. NRR cohort data for 12+ month customers is the decisive metric for the infrastructure-vs-discretionary debate — if early cohorts churn, the “nice-to-have” thesis hardens from opinion to data.

10

Sources

EXPERT CALLS (N=11, PROPRIETARY)

11 proprietary expert calls across 6 companies. Experts are current or former Rox users at enterprise and growth-stage companies. Profiles: Heads of Sales, CROs, Enterprise AEs, Sales Strategy Managers, Directors of Sales Acceleration. Sourced April 2025 through January 2026. All experts anonymized per policy.

PUBLIC INTERVIEWS & WEB RESEARCH

14 product/customer videos (Rox YouTube channel). Bloomberg Technology CEO interview. Investor articles from Sequoia, General Catalyst, GV, and Not Boring. Web sources: TechCrunch, PitchBook, Kavout, OpenAI partner listing. Altis did not have access to Rox management or internal documents.

NUMBERED REFERENCES

11

Legal Notices

1. Informational Purposes Only. This report is provided for informational purposes only and does not constitute investment, legal, tax, accounting, or other professional advice. Altis Ventures, Inc. (“Altis”) is not an investment advisor.

2. No Recommendation or Solicitation. Nothing in this report should be interpreted as a recommendation to buy, sell, or hold any security.

3. No Reliance. Recipients may not rely on this report as the basis for any investment decision.

4. No Guarantee of Accuracy. This report may include information from third party sources, expert interviews, and AI-assisted analysis. Altis does not independently verify all data.

5. AI-Assisted Analysis. Portions of this report may have been produced or synthesized through use of AI. All AI-assisted content is subject to human review.

6. Forward-Looking Statements. Views expressed are those of Altis as of the date indicated and are subject to change.

7. No Duty to Update. 8. No Redistribution. 9. No Fiduciary Relationship.

10. Conflicts and Interests. Altis may have relationships with entities referenced in this report.

11. Warranty Disclaimer. THE REPORT IS PROVIDED ON AN “AS IS” BASIS.

12. Limitation of Liability. Altis’s aggregate liability shall not exceed $100.

12

Thank you

13