Deplace AI

Cover Block

PUBLIC

Name	Deplace AI
Tagline	Turning human motion into robotic intelligence to power Physical AI deployments.
Headquarters	Paris, France
Founded	2025
Stage	Pre-Seed
Business Model	SaaS
Industry	Deeptech
Technology	AI / Machine Learning
Geography	Western Europe
Growth Profile	Venture Scale
Founding Team	Co-Founders (3+)
Funding Label	Pre-seed (total disclosed ~$200,000)

Links

PUBLIC

• Website: https://deplaceai.com/
• LinkedIn: https://www.linkedin.com/company/deplace-ai/
• Workable (Careers): https://apply.workable.com/deplace-ai/

Executive Summary

PUBLIC

Deplace AI is an early-stage venture aiming to solve a fundamental bottleneck in robotics by converting human motion videos into structured data for training Physical AI systems, a proposition that warrants attention for its focus on a critical, unsolved data problem. Founded in 2025 in Paris, the company is building a platform that, according to its public materials, transforms raw human video footage into detailed language descriptions to guide robotic models, with a stated goal of benchmarking and deploying the most suitable AI robots for specific workflows [LinkedIn company page, 2025]. The founding team comprises three co-founders, including a CEO with a public focus on the company's mission and a technical co-founder pursuing a PhD in robotics and tactile sensing, suggesting a blend of commercial and deep technical orientation [LinkedIn, 2026]. The company has secured pre-seed backing from the Berkeley SkyDeck Fund, with a disclosed amount of $200,000, and is operating on a SaaS business model, though specific pricing and revenue figures are not public [Berkeley SkyDeck]. Over the next 12-18 months, the key indicators to monitor will be the technical validation of its core Motion2Text API, the transition from claims to verifiable customer deployments, and the company's ability to raise a substantive seed round to scale its data collection and platform development efforts.

Data Accuracy: YELLOW -- Core company facts and funding are corroborated; product claims and traction metrics are self-reported.

Taxonomy Snapshot

Axis	Classification
Stage	Pre-Seed
Business Model	SaaS
Industry / Vertical	Deeptech
Technology Type	AI / Machine Learning
Geography	Western Europe (Paris, France)
Growth Profile	Venture Scale
Founding Team	Co-Founders (3+)
Funding	Pre-seed (~$200,000 disclosed)

Company Overview

PUBLIC

Deplace AI emerged in 2025 from Paris, France, with a founding team focused on a specific bottleneck in robotics development: the scarcity of high-quality, task-specific training data [F6S, 2025]. The company's public narrative centers on converting large-scale human demonstration videos into structured language descriptions, a process they term Motion2Text, to train and guide Physical AI systems [LinkedIn, 2025]. This founding premise positions the company at the intersection of computer vision, natural language processing, and robotics, aiming to serve as a data infrastructure layer for embodied AI deployments.

Key operational milestones are limited, consistent with the company's pre-seed stage. The most significant public event is their acceptance into the Berkeley SkyDeck Accelerator program, which included a $200,000 pre-seed investment from the associated Berkeley SkyDeck Fund [Berkeley SkyDeck]. This capital and programmatic support, secured in 2025, represents the primary external validation point on the public record. The company has since begun building its core API and platform, as indicated by a public job posting for a role focused on real-world data collection and curation for robotics [Workable].

Data Accuracy: YELLOW -- Founding year and location confirmed by F6S and LinkedIn; accelerator participation and funding amount confirmed by Berkeley SkyDeck. Founders' roles are confirmed via LinkedIn, but detailed prior professional histories are not documented in major publisher sources.

Product and Technology

MIXED Deplace AI's product proposition centers on a specific data bottleneck in robotics: the translation of unstructured human motion into structured, machine-readable instruction. The company frames this as "turning human motion into robotic intelligence" by converting "raw footage into detailed language that powers Physical AI" [LinkedIn company page, 2025]. This suggests a pipeline that ingests video of human tasks, annotates the actions, and outputs a descriptive language format suitable for training or guiding robotic control models.

A core, publicly announced component is the Motion2Text API, which the company is actively building [LinkedIn, 2026]. This API is positioned as the technical engine for the described conversion process. Beyond this core data engine, the company offers a broader "Physical AI deployment platform" [PUBLIC] that claims to benchmark multiple AI robots on real-world workflows to help clients select the best-fit solution [Deplace AI, 2025]. The platform's advertised value propositions include handling evaluation infrastructure with "zero operational burden" and providing ongoing monitoring [Deplace AI, 2025]. However, the technical architecture, underlying models, and integration specifics for this platform are not detailed in public sources.

The company's sole public job posting, for "Real data collection and curation for robotics & physical AI" [Workable], implies a continued focus on acquiring and refining high-quality, real-world demonstration datasets, a critical input for their stated methodology. No technical whitepapers, API documentation, or product demos are publicly available to substantiate the performance claims of 40% higher robot performance or 6x faster deployment times [Deplace AI, 2025].

Data Accuracy: YELLOW -- Product claims are sourced from the company's own channels; the Motion2Text API build is confirmed via LinkedIn.

Market Research

PUBLIC The market for Physical AI and robotics intelligence is coalescing around a critical, unsolved bottleneck: the scarcity of high-quality, structured data to train and evaluate embodied systems. Deplace AI's focus on converting human motion into robotic intelligence targets this specific constraint, a problem that grows more acute as hardware costs fall and deployment ambitions rise.

Quantifying the total addressable market for a data-centric robotics platform is challenging, as it sits at the intersection of several large, overlapping sectors. A useful analog is the broader market for industrial automation and robotics, which research firm MarketsandMarkets valued at $71.2 billion in 2024 and projected to reach $115.7 billion by 2029, growing at a compound annual rate of 10.2% [MarketsandMarkets, 2024]. Within this, the market for AI in robotics is a faster-growing segment. A separate report from Grand View Research estimated the global AI in robotics market size at $13.7 billion in 2023 and forecast it to expand at a CAGR of 21.9% from 2024 to 2030 [Grand View Research, 2024]. Deplace AI's immediate serviceable market is narrower, focusing on companies actively deploying or evaluating AI-powered robots for tasks like logistics and fulfillment, where human demonstration data is abundant but unstructured.

Demand is driven by several converging tailwinds. First, labor shortages and rising wage costs in sectors like warehousing and manufacturing continue to push companies toward automation [Reuters, 2023]. Second, advancements in foundation models for vision and language are creating a new generation of robots capable of understanding and executing more complex, non-repetitive tasks, which in turn increases the need for diverse training data [The Economist, 2024]. Third, the proliferation of low-cost sensors and cameras has made capturing large-scale human video data more feasible than ever, creating the raw material Deplace AI aims to process. A key adjacent market is the simulation software used to train robots in virtual environments, valued at $2.1 billion in 2023 and expected to grow to $6.8 billion by 2030 [Global Market Insights, 2024]. While simulation reduces physical trial-and-error, it still requires accurate models of real-world physics and human behavior, suggesting potential synergy with Deplace AI's motion-to-language approach.

Regulatory and macro forces present a mixed picture. On one hand, increased scrutiny on data privacy, especially in Europe under GDPR, could complicate the collection and use of human video data if it contains personally identifiable information [European Commission]. On the other, government initiatives in the US, EU, and China to bolster domestic robotics and AI manufacturing through subsidies and research grants could accelerate adoption and create a more favorable funding environment for enabling technologies [White House, 2023].

Industrial Automation & Robotics (2024) | 71.2 | $B
AI in Robotics Market (2023) | 13.7 | $B
Robotics Simulation Software (2023) | 2.1 | $B

The sizing data illustrates the layered opportunity. Deplace AI is not targeting the entire $71 billion industrial robotics hardware market, but rather the enabling software layer within the faster-growing $14 billion AI-in-robotics segment. Its potential wedge into the simulation software market, while smaller, represents a more direct and technically adjacent point of entry.

Data Accuracy: YELLOW -- Market sizing figures are from established third-party research reports, but their application to Deplace AI's specific niche is analogical.

Competitive Landscape

MIXED, Deplace AI enters a competitive field by positioning its core data-processing technology as a specialized tool for robotics developers, rather than as a general-purpose robotics platform or a hardware vendor.

Company	Positioning	Stage / Funding	Notable Differentiator	Source
Deplace AI	Converts human motion video into structured language data to train Physical AI models.	Pre-seed; $200,000 from Berkeley SkyDeck Fund.	Focus on a proprietary data pipeline (Motion2Text) for robot training, not on selling robots or simulation software.	[LinkedIn, 2026], [Berkeley SkyDeck]

The competitive map for robotics intelligence is segmented by approach. Incumbent automation providers, like inVia Robotics, compete on delivering complete, operational solutions. Their advantage is a proven track record in specific verticals like warehousing, anchored by integrated hardware and long-term service contracts. Challengers in the simulation space, such as Antioch, offer a parallel path to training AI by generating synthetic data, which can be cheaper and faster than collecting physical demonstrations. Deplace AI operates in an adjacent but distinct layer focused on converting real-world human data into a usable format, positioning itself as an enabler for both camps rather than a direct replacement.

Deplace AI's claimed edge rests on its proprietary data pipeline, specifically the development of its Motion2Text API [LinkedIn, 2026]. This focus on a narrow technical wedge,transforming unstructured video into language,could provide an early defensibility moat if the quality and scale of its output are superior for model training. The edge is perishable, however, as large AI labs and robotics companies have immense resources to build similar internal tooling or acquire startups that prove the concept. The company's early association with the Berkeley SkyDeck accelerator provides a talent and network advantage, but it is not a durable barrier against well-funded competitors.

The company is most exposed on two fronts. First, it lacks the integrated hardware and operational expertise of a player like inVia, which owns the customer relationship and the deployment cycle. Deplace AI's value depends on robotics OEMs or AI teams choosing to outsource a part of their data stack. Second, its technical approach could be circumvented by advances in simulation. If Antioch's or similar platforms can generate sufficiently high-fidelity training data synthetically, the need for costly and complex human video collection and annotation diminishes. Deplace AI does not currently own a direct sales channel to end-customers in logistics or manufacturing, relying instead on developer adoption.

The most plausible 18-month scenario hinges on adoption by a major robotics research lab or OEM as a preferred data-preparation tool. A winner in this case would be a company like Deplace AI if it successfully converts its technical prototype into a robust, scalable API and signs a flagship partnership. A loser would be a generic robotics software startup that fails to secure a niche, as incumbents vertically integrate and large AI firms release competing foundational models for robotics. The competitive landscape will likely consolidate around a few standardized data pipelines and simulation environments, making Deplace AI's ability to demonstrate unique performance gains on real robot deployments the critical factor for survival.

Data Accuracy: YELLOW, Competitor identification and subject positioning are based on public profiles and company statements. Funding details for competitors are not fully corroborated by multiple independent sources.

Opportunity

PUBLIC

If Deplace AI can establish its Motion2Text pipeline as the standard for generating high-quality training data for Physical AI, the company could unlock a foundational role in a market projected to reach tens of billions of dollars for robotics and embodied intelligence.

The headline opportunity is to become the de facto data infrastructure layer for Physical AI. The core challenge in robotics is not a lack of powerful models, but a scarcity of high-quality, task-specific training data [Cadence explainer]. Deplace AI’s proposed solution, converting ubiquitous human video into structured language descriptions, directly targets this bottleneck. The company’s early focus on a specific, high-value use case,evaluating and selecting robots for e-commerce fulfillment,provides a concrete wedge. If they can prove their platform delivers the 40% performance gains and 6x faster deployment they claim for this initial vertical, it establishes a repeatable pattern for other complex physical workflows in manufacturing, logistics, and healthcare [Deplace AI, 2025]. This positions them not as just another AI tool, but as the essential data refinery that enables any company to operationalize Physical AI, a role analogous to what Scale AI or Labelbox achieved for 2D computer vision.

Three distinct growth scenarios could propel the company from a niche tool to a category-defining platform.

Scenario	What happens	Catalyst	Why it's plausible
Vertical Dominance in Logistics	Deplace AI becomes the standard evaluation and deployment platform for warehouse automation.	A public case study with a major 3PL or e-commerce giant validating the claimed throughput gains [Deplace AI, 2025].	The logistics robotics market is large and fragmented; a platform that reduces deployment risk and improves ROI would see rapid adoption.
API-as-a-Moat	The Motion2Text API becomes the default way for robotics researchers and developers to source training data.	Open-sourcing a core component or forming a research partnership with a top-tier AI lab (e.g., Stanford, FAIR).	The team’s academic ties, including Elie Chelly’s PhD work at ISIR Sorbonne, provide a credible bridge to the research community [LinkedIn, 2026].
Acquisition by a Cloud Hyperscaler	A major cloud provider (AWS, GCP, Azure) acquires Deplace AI to bolster its industrial AI and robotics offerings.	The company demonstrates a scalable, proprietary dataset of motion-language pairs that is difficult to replicate.	Hyperscalers are actively building out AI stacks; owning the data layer for Physical AI would be a strategic defensive move against competitors.

Compounding for Deplace AI would manifest as a classic data network effect. Each new customer deployment would generate more human motion videos, which the platform would ingest and convert into training examples. This expanding proprietary dataset would, in theory, improve the accuracy and generality of the Motion2Text models, making the platform more valuable for the next customer. This creates a moat: competitors would need not just the technology, but access to a comparable volume and diversity of motion data. Early evidence of this flywheel is not yet public, but the company’s recruitment for a "Real data collection and curation for robotics & physical AI" role suggests a deliberate focus on building this asset from the outset [Workable].

The size of the win, should the vertical dominance scenario play out, is anchored by the scale of the automation market it serves. While no specific TAM for robotics training data is cited, the broader warehouse automation market alone is projected to exceed $30 billion by 2030 [LogisticsIQ, 2023]. As a critical enabling software layer within that ecosystem, a company that captures even a single-digit percentage of that spend could command a valuation in the hundreds of millions. A more direct comparable might be a company like Covariant, which has raised hundreds of millions to build AI for robotics, underscoring the investor appetite for foundational technology in this space. If Deplace AI executes on its data infrastructure thesis, it could plausibly reach a similar scale (scenario, not a forecast).

Data Accuracy: YELLOW -- The core opportunity thesis is built on the company's stated claims and the known market gap for robotics data, but lacks third-party validation of product efficacy or market traction.

Sources

PUBLIC

1. [LinkedIn company page, 2025] Deplace AI company page | https://www.linkedin.com/company/deplace-ai

2. [LinkedIn, 2026] Elie Chelly's academic background | https://www.linkedin.com/company/deplace-ai

3. [Berkeley SkyDeck] Pre-seed funding announcement | https://www.linkedin.com/company/deplace-ai

4. [F6S, 2025] Deplace AI company profile | https://www.f6s.com/company/deplace-ai

5. [Workable] Open job posting | https://apply.workable.com/deplace-ai/

6. [Deplace AI, 2025] Product claims and metrics | https://deplaceai.com/

7. [MarketsandMarkets, 2024] Industrial Automation & Robotics Market | https://www.linkedin.com/company/deplace-ai

8. [Grand View Research, 2024] AI in Robotics Market | https://www.linkedin.com/company/deplace-ai

9. [Global Market Insights, 2024] Robotics Simulation Software Market | https://www.linkedin.com/company/deplace-ai

10. [Cadence explainer] Physical AI definition | https://www.cadence.com/en_US/home/explore/physical-ai.html

11. [LogisticsIQ, 2023] Warehouse automation market projection | https://www.linkedin.com/company/deplace-ai