Chakra Labs
Chakra Labs builds deterministic, pixel-perfect reinforcement learning environments and large-scale trajectory datasets (PANGO) for training computer use agents, serving frontier AI research labs and CUA developers through its Dojo platform and subscription-based enterprise access.
- Company typePrivate
- Founded2024
- HeadquartersBrooklyn, United States
- Headcount1–10
- GTM typeB2B
- OfferingSoftware
What Chakra Labs does
Chakra Labs is a Brooklyn-based applied research and data infrastructure company founded in 2024 that builds deterministic, pixel-perfect reinforcement learning environments and large-scale trajectory datasets for training computer use agents (CUAs). The company positions itself as a "Frontier Data Laboratory" serving AI research labs and CUA developers, with its core products centered on the Dojo platform, the PANGO crowdsourced trajectory dataset (approximately 6x larger than comparable open-source CUA datasets), and the GLADOS-1 open-source CUA model that demonstrates the platform's training capabilities. Underlying technology emphasizes frame-accurate state control, temporal integrity preservation, deterministic system behavior, and mixed-modality (GUI plus MCP) training environments that allow agents to switch between visual navigation and programmatic commands within a single task.
Chakra's technical infrastructure includes pixel-perfect clones of production software (Notion, Figma, Canva, Gmail, LinkedIn, Salesforce, Slack, GitHub, and APAC platforms like JD.com, XiaoHongShu, and Weibo), a trajectory library containing 2,500+ hours of human computer-use data and 25 billion segmented image tokens, and first-class integrations with the Harbor, Verifiers (Prime Intellect), and Verl (Bytedance Seed) RL frameworks. The company has demonstrated measurable model performance gains: Claude Opus 4.6 reached 41.39%, ChatGPT 5.2-High 29.51%, and Gemini 2.5 Pro 9.43% on Dojo's evaluation benchmark, while GLADOS-1 improved OSWorld performance from 15% to 17% overall and action format compliance from 39% to 58%.
The business operates a subscription-based platform access model with monthly billing and a 30-day free trial, supplemented by custom enterprise arrangements for frontier AI research labs accessed through direct sales engagement. Distribution combines direct enterprise sales (via info@chakra-labs.com and partner@chakra-labs.com), a self-serve trydojo.ai entry point for researchers, and a mobile companion application (Chakra Data Scout). The company is venture-backed with a $10M funding round closed in January 2026, led by co-founders Nirmal Krishnan (CEO) and Alexander Fung (CTO), and pursues a thought leadership GTM strategy through technical publications, conference presence (ICML 2025), and open-source model releases.
Chakra Labs firmographics
Firmographics- Name
- Chakra Labs
- Legal name
- CHAKRA DIGITAL LABS, INC.
- Website
- https://chakra.dev
- Company type
- Private
- Founded year
- 2024
- Operating status
- Operating
- Headcount range
- 1–10 employees
- Short description
- Chakra Labs builds deterministic, pixel-perfect reinforcement learning environments and large-scale trajectory datasets (PANGO) for training computer use agents, serving frontier AI research labs and CUA developers through its Dojo platform and subscription-based enterprise access.
- Ownership category
- akta.pro rank
Chakra Labs industry classification
Industry- Product category
- AI Training Infrastructure
- NAICS
- Other Computer Related Services (541519), Computer Systems Design and Related Services (54151), Custom Computer Programming Services (541511)
- SIC
- Services-Computer Processing & Data Preparation (7374), Services-Prepackaged Software (7372)
- akta.pro industry
- RLHF, Human Feedback & Evaluation Data ToolsHardware, Technology & Telecom/Artificial Intelligence & Machine Learning/Synthetic Data, Data Labeling & Training Data ToolsModel Development & Training Platforms (AutoML, Notebooks, Feature Stores)Hardware, Technology & Telecom/Enterprise Platforms & Databases/Artificial Intelligence Platforms for Enterprise
Keywords
Where Chakra Labs is headquartered
LocationHeadquarters
- HQ city
- Brooklyn
- HQ country
- United States
- HQ region
- North America
Offices1 record
Markets served
Chakra Labs business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Technology or R&D, Personnel, Infrastructure, Marketing or Sales
Revenue model
- Platform Access Subscription: Monthly subscription model for access to Chakra's research-grade RL environments, trajectory datasets, and training infrastructure. Includes 30-day free trial for new users. Billing is monthly with recurring charges without prior approval for each billing cycle.
- Enterprise Platform Access: Custom enterprise arrangements for frontier labs requiring dedicated infrastructure, bespoke solutions, and specialized support. Accessed through partnership inquiries via partner@chakra-labs.com.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Subscription | Monthly | Standard Platform Access |
Go-to-market motion1 record
Distribution channels3 records
Marketing channels5 records
Chakra Labs product offering
Product offeringCore offering
Chakra Labs provides research-grade reinforcement learning (RL) environments and trajectory datasets for training and evaluating computer use agents (CUAs). The core product, Dojo, delivers deterministic, pixel-perfect clones of production software (e.g., Notion, Figma, Gmail, Salesforce) with frame-accurate state control and mixed-modality (GUI + MCP) interaction support. The company also ships the PANGO crowdsourced trajectory dataset and the GLADOS-1 CUA model.
Product overview
Chakra Labs operates as a Frontier Data Laboratory providing research-grade infrastructure for AI agent development. The core offering consists of deterministic RL environments (Dojo) for training computer use agents, paired with large-scale trajectory datasets (PANGO) containing 2,500+ hours of human computer-use data. Their proprietary GLADOS-1 model demonstrates the platform's capabilities, post-trained on PANGO trajectories. The platform emphasizes mixed-modality training supporting both GUI interaction and MCP tool-use within single tasks, enabling agents to learn workflows across productivity software including Notion, Figma, Salesforce, and similar applications.
Differentiator
Problem solved
Functional benefit
Brands
- Dojo: Collaborative RL environment suite for computer use agents (CUAs), enabling models to learn by interacting with realistic, task-driven interfaces including production-grade website and app clones.
- GLADOS-1
- Chakra Data Scout
- Chakra Data Laboratory
Products and services
- Dojo A collaborative reinforcement learning environment suite for computer use agents, providing interactive clones of production software for training, evaluating, and iterating on CUA models. Supports GUI and MCP interaction modalities.
- GLADOS-1 The first computer use agent (CUA) model post-trained using crowd-sourced trajectories from the PANGO dataset. Built on UI-TARS-7B-SFT, improves OSWorld performance through new perception training formats and online RL.
- RL Environments Deterministic, pixel-perfect reinforcement learning environments with frame-accurate state control for frontier AI research. Environments include Notion, Figma, Canva, Gmail, LinkedIn, Salesforce, Slack, GitHub, and other software clones.
- PANGO Dataset Productivity Applications with Natural GUI Observations and trajectories - a large crowdsourced dataset of real human UI interactions for training computer use agents. Approximately 6x larger than comparable open-source datasets.
- Chakra Data Scout Mobile application for accessing Chakra platform services, distributed via Apple App Store and Google Play.
Quantifiable outcome
- Model performance improvements: Claude Opus 4.6 at 41.39%, ChatGPT 5.2-High at 29.51%, Gemini 2.5 Pro at 9.43% on dojo-bench-mini
- +2 more outcomes
Companies that use Chakra Labs
Customer profileSegments2 records
Ideal customer profiles2 records
Chakra Labs technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
Integration4 records
AI capability8 records
Feature7 records
Chakra Labs partnerships and signals
Strategic signalPartnerships
Five partnerships are on record, tiered core.
- Verifiers (Prime Intellect)coreFirst-class integration with Verifiers by Prime Intellect for evaluation of CUA models trained in Dojo environments. Part of the RL framework ecosystem supporting Dojo platform.
- Verl (Bytedance Seed)coreFirst-class integration with Verl by Bytedance Seed for reinforcement learning training in Dojo environments. Enables online RL training workflows alongside evaluation.
- AnthropiccoreAI service provider used for Chakra's AI Products. Claude models integrated into the platform's AI Product offerings for CUA development.
- OpenAIcoreAI service provider used for Chakra's AI Products. GPT models integrated into the platform's AI Product offerings for CUA development.
- HarborcoreInfrastructure support for Dojo platform. Enables hosted deployments and on-premises installations of Chakra environments.
Scale indicators5 records
Recent moves6 records
Expansion highlights5 records
Chakra Labs competitors and assessment
Company assessmentEmerging players
- Adept AI: Adept builds foundation models that act on existing software (ACT-1, ACT-2) for enterprise workflow automation. Comparable as a CUA-focused AI company whose training data requirements overlap with Chakra's offering; Adept represents both a potential customer and a vertical competitor.
- MultiOn: MultiOn builds AI agents that navigate browsers and software to complete tasks autonomously. Comparable as a CUA company with overlapping trajectory data needs and customer base; MultiOn could be both a customer of Chakra's data and a competitor in agent deployment.
- Convergence AI: Convergence AI builds autonomous computer-use agents (Proxy) for general-purpose task automation. Comparable as a CUA-focused emerging player with overlapping target customers (frontier labs, enterprises deploying CUAs), though Convergence focuses on the agent layer while Chakra provides training data and environments.
Direct peers
- Prime Intellect: Prime Intellect builds decentralized AI training infrastructure including the Verifiers framework, which Chakra integrates with. Directly comparable as an RL training infrastructure provider targeting frontier-scale experiments, with partnership/competition overlap on RL environment tooling.
- Simular Research: Simular builds digital employees and computer use agents, offering agent products at $500/month per seat. Directly comparable as a CUA-focused AI company with overlapping market and ecosystem; Chakra's own data references Simular positioning.
- Surge AI: Surge AI provides high-quality RLHF data and human-preference datasets to frontier AI labs. Directly comparable as a frontier-lab-focused data provider with emphasis on RL training data quality, though Surge focuses on NLP/chat data while Chakra specializes in CUA trajectories.
Broad incumbents
- Hugging Face: Hugging Face hosts open datasets, models, and training infrastructure (including AutoTrain, RL environments via HuggingGym) widely used by AI researchers. Comparable as a host and distribution channel for training data and RL tooling, with overlapping buyer base of ML researchers and frontier labs.
- Scale AI: Scale AI is the dominant AI training data provider, offering data labeling, RLHF, evaluation, and synthetic data services to frontier labs and enterprises. Directly comparable as a data infrastructure provider for frontier AI, though Scale operates across many data modalities (vision, NLP, autonomous driving) versus Chakra's CUA/RL focus.
- Labelbox: Labelbox provides data labeling and evaluation infrastructure for AI teams, with cataloging, annotation, and RLHF support. Comparable as an AI training data platform serving frontier labs and enterprises, though broader in industry coverage than Chakra's CUA specialization.
- Snorkel AI: Snorkel AI provides data-centric AI platform including programmatic labeling, weak supervision, and RLHF tooling for enterprise ML teams. Comparable as an AI training data infrastructure company with comparable customer base, though broader in scope and less specialized in CUA.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat4 records
Key risks5 records
Key highlights6 records
Customer concentration
Chakra Labs social profiles
Digital presenceChakra Labs financial estimates
Financial estimateRevenue estimate
Valuation estimate
Chakra Labs leadership team
Management profileNumber of profiles
Profiles2 records
Chakra Labs funding detail
Funding detailFunding overview
Funding rounds1 record
Investors1 record
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Chakra Labs M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Chakra Labs
What does Chakra Labs do?
Chakra Labs provides research-grade reinforcement learning (RL) environments and trajectory datasets for training and evaluating computer use agents (CUAs). The core product, Dojo, delivers deterministic, pixel-perfect clones of production software (e.g., Notion, Figma, Gmail, Salesforce) with frame-accurate state control and mixed-modality (GUI + MCP) interaction support. The company also ships the PANGO crowdsourced trajectory dataset and the GLADOS-1 CUA model.
Is Chakra Labs a public or private company?
Chakra Labs is a private company. It is classified as venture growth investor backed and is currently operating.
When was Chakra Labs founded?
Chakra Labs was founded in 2024. It employs 1 to 10 people.
Where is Chakra Labs based?
Chakra Labs is headquartered in Brooklyn, United States, in the North America region.
How does Chakra Labs make money?
Two revenue lines are on record. Platform Access Subscription is the primary driver. The others are enterprise Platform Access.
Who are Chakra Labs's main competitors?
Emerging players on record are Adept AI, MultiOn and Convergence AI. Direct peers are Prime Intellect, Simular Research and Surge AI. Broad incumbents are Hugging Face, Scale AI, Labelbox and Snorkel AI.
Does Chakra Labs have an API?
Yes. API access is configured as part of the onboarding process. The platform includes API credentials and documentation for enterprise customers. Environments support MCP (Model Context Protocol) tool-use within single tasks, enabling agents to switch between visual navigation and programmatic commands.
What industry is Chakra Labs in?
Chakra Labs's product category is AI Training Infrastructure. Its primary akta.pro industry code is HDAAALAJ, RLHF, Human Feedback & Evaluation Data Tools, with a secondary code of HDAEANAB, Model Development & Training Platforms (AutoML, Notebooks, Feature Stores). Its NAICS code is 541519 and its SIC code is 7374.