Is Prox profitable in 2026?

No. Prox is a 7-person Y Combinator Fall 2025 startup with no independently verified revenue or named paying customers as of July 2026. Job listings mention 'first deployments' but name no clients, so profitability cannot be confirmed from public sources.

Who is Prox's biggest competitor?

Aquant, an agentic-AI field-service platform operating since 2016 with named industrial logos including Makino, Comfort Systems, and Powerscreen. Prox publicly frames Zendesk as its incumbent contrast, but Aquant is the closer, more advanced rival in complex-equipment support as of 2026.

Could AI replace Prox itself?

Partly. Prox's core claimed moat is proprietary multimodal knowledge-graphs for diagram and CAD reasoning that general LLMs allegedly cannot yet handle, but frontier multimodal models (GPT, Claude, Gemini) are actively closing that gap, threatening to commoditize Prox's differentiation within 12-24 months.

Who owns or funds Prox?

Prox is private, backed by Y Combinator and Bloomberg Beta ($450M AUM as of June 2026), plus a job listing from April 2026 naming Paul Graham, SV Angel, and Burst Capital as backers. No confirmed funding amount or valuation has been publicly reported.

What does Prox actually build?

Prox builds AI technical support tools for manufacturers of complex physical products like powersports, agriculture, HVAC, firearms, and construction equipment, ingesting manuals and diagrams into a knowledge graph deployed on product pages, dealer tablets, and via ChatGPT, Claude, Copilot, and Gemini connectors as of July 2026.

How many people work at Prox?

Y Combinator's company profile and multiple 2026 job postings consistently list Prox's team size as 7, as of July 2026, while actively recruiting for Founding Engineer, Forward-Deployed Engineer, Chief of Staff, and Agent Evals roles.

YC Fall 2025 · United States

Prox

AI technical support tool for complex physical products: powersports, ag, HVAC, firearms.

Losing its protection

Assessed 27 July 2026 · Evidence as of 27 July 2026

The structural protection is eroding.

On the recordSince 27 July 2026 · Checkable from within 12 months of 2026-07 · Every call on the record Checkpoint: Prox's first exclusive-data manufacturer contract and named paying customer, or its Series A outcome

The call

Prox has a real niche in complex-product support but no proven data lock-in yet, while both frontier AI models and a funded rival close in from opposite directions.

Prox builds software that helps technicians fix complicated machines like dirt bikes, farm equipment, and heating systems by using AI to read diagrams and manuals faster. The company is trying to make this hard to replace by locking in exclusive data from manufacturers and dealers, but this is unproven. Aquant, a competitor that started in 2016, is already selling similar tools to factories and repair networks, and bigger companies like Salesforce are slowly building their own versions. What matters most in the next year is whether Prox can sign exclusive deals with equipment makers before the AI systems everyone uses (like ChatGPT and Claude) get good enough at reading diagrams on their own, which would make Prox's main advantage disappear.

Why

AI pressure: Frontier multimodal models (GPT-5, Claude, Gemini). These models are closing the diagram and CAD reasoning gap Prox calls its main differentiator, threatening to make its knowledge graph redundant. Could AI do the core job? Yes, likely.

Bigger players: Aquant. Aquant, funded since 2016, already runs an agentic AI platform for industrial and complex-equipment servicing, overlapping Prox's target buyers. Could a giant copy it? Yes, easily.

What protects it. Locking in exclusive manufacturer data (service bulletins, wiring diagrams) and dealer-tablet workflows before rivals or bigger models do the same.

What would change this call

Prox announces an exclusive manufacturer data deal or named paying customer within 12 months. Would show Prox locking in the data moat it needs before rivals close the gap.
Aquant or ServiceMax launches a diagram/CAD-reasoning feature for dealer tablets or product pages. Would show incumbents closing the technical gap Prox relies on for differentiation.
GPT-5, Claude 4, or Gemini 3 demonstrate strong wiring-diagram or CAD parsing accuracy. Would show frontier multimodal models commoditizing Prox's core technical claim.

Key numbers

Team size: 7 people (2026-07)
Backed by YC and Bloomberg Beta ($450M AUM) (2026-06-14)
Aquant has shipped agentic AI platform since 2016 (2026-07)

Named risks

No independently verified revenue, customers, or named enterprise logos as of July 2026. Severity: high
Aquant, a funded 2016-vintage competitor, already has shipped platform and named industrial logos. Severity: high
Claimed diagram/CAD reasoning moat is a closing capability gap as frontier multimodal LLMs improve. Severity: high
Seed-to-Series A conversion fell to about 9% in 2025, raising follow-on funding risk for early-stage AI startups. Severity: medium
Residual outdated ShipBob logistics-claims copy on YC page suggests recent, possibly unsettled pivot. Severity: medium

Rivals, as read in July 2026

Rival	Direction	Evidence
Aquant	losing to	Aquant has shipped agentic AI platform since 2016 with named logos Makino, Comfort Systems, Powerscreen (as of 2026-07-08)
Salesforce/ServiceMax/IFS	losing to	Incumbent FSM vendors increasingly integrating AI features, pressuring specialists per 2025-07-26 analysis
Zendesk	contesting	Prox claims Zendesk solved simple-product support but not complex technical support (vendor claim, n.d.)
Frontier multimodal LLMs (GPT/Claude/Gemini)	gaining on	Models are closing the diagram and CAD reasoning gap Prox cites as its core differentiator, as of 2026-07

Questions people ask

Is Prox profitable in 2026?: No. Prox is a 7-person Y Combinator Fall 2025 startup with no independently verified revenue or named paying customers as of July 2026. Job listings mention 'first deployments' but name no clients, so profitability cannot be confirmed from public sources.
Who is Prox's biggest competitor?: Aquant, an agentic-AI field-service platform operating since 2016 with named industrial logos including Makino, Comfort Systems, and Powerscreen. Prox publicly frames Zendesk as its incumbent contrast, but Aquant is the closer, more advanced rival in complex-equipment support as of 2026.
Could AI replace Prox itself?: Partly. Prox's core claimed moat is proprietary multimodal knowledge-graphs for diagram and CAD reasoning that general LLMs allegedly cannot yet handle, but frontier multimodal models (GPT, Claude, Gemini) are actively closing that gap, threatening to commoditize Prox's differentiation within 12-24 months.
Who owns or funds Prox?: Prox is private, backed by Y Combinator and Bloomberg Beta ($450M AUM as of June 2026), plus a job listing from April 2026 naming Paul Graham, SV Angel, and Burst Capital as backers. No confirmed funding amount or valuation has been publicly reported.
What does Prox actually build?: Prox builds AI technical support tools for manufacturers of complex physical products like powersports, agriculture, HVAC, firearms, and construction equipment, ingesting manuals and diagrams into a knowledge graph deployed on product pages, dealer tablets, and via ChatGPT, Claude, Copilot, and Gemini connectors as of July 2026.
How many people work at Prox?: Y Combinator's company profile and multiple 2026 job postings consistently list Prox's team size as 7, as of July 2026, while actively recruiting for Founding Engineer, Forward-Deployed Engineer, Chief of Staff, and Agent Evals roles.

Timeline

2026-07-22Prox listed as API provider enabling ChatGPT, Claude, Copilot, Gemini to query manufacturer knowledge via MCP connector
2026-07-06Forward-Deployed Engineer job post mentions first deployments and pilot rollouts, names no customers
2026-04-27Chief of Staff job listing names Paul Graham, SV Angel, Burst Capital as backers
2026-04-03Founding Engineer job post details roadmap: voice AI, video processing, CAD sharing, multimodal knowledge graphs
2025-09Prox joins Y Combinator Fall 2025 (F25) batch, San Francisco

Revision history

27 July 2026

First assessed. Losing its protection. The text above is fixed as of this date and is not edited afterwards.

In the same band

Nexus Terrain Metorial