The Good AI Lab

What The Good AI Lab does

The Good AI Lab is an independent non-profit AI research collective founded in 2024 and headquartered in Stockholm, Sweden, with a distributed team of approximately 50 named contributors across leading universities and companies globally. The organization operates two intertwined workstreams: an "AI Lab" focused on foundational academic research in computer vision, open-vocabulary semantic segmentation, and domain adaptation, and a "Good AI" stream that applies these capabilities to humanitarian applications. Core technical outputs include SemLA (CVPR 2025), a training-free test-time domain adaptation framework using LoRA adapter retrieval and fusion; VocAlign (BMVC 2025), the first framework for source-free domain adaptation in open-vocabulary segmentation; HAMLET (ICCV 2023), a real-time adaptation framework achieving 29+ FPS on consumer GPUs; and UPR Matcher, an NLP tool for tracking UN Universal Periodic Review recommendations. All code is published openly on GitHub under Apache 2.0 or MIT licenses.

The organization does not sell commercial products or operate a traditional revenue model. Instead, it sustains operations through research collaborations, partnership work with humanitarian organizations (Doctors Without Borders/MSF Italy, FMSI), academic publication channels, and identified EU research funding (PRIN2022, NRRP). Its customer base consists of humanitarian NGOs applying its tools, the academic research community consuming its publications and code, and graduate students participating in its research mentorship program. The primary go-to-market is community-led: publishing at top AI venues, co-hosting the CV4GOOD workshop at ECCV 2026 with MSF, and building open-source tools that humanitarian organizations can adopt directly. Governing board and advisory positions are held by an AMD Silo AI Chair, professors at KTH, Bologna, Modena, and CTU Prague, and a Director of Research at Google Zurich.

Name

The Good AI Lab

Legal name

The Good AI Lab

Website

https://thegoodailab.org

Company type

Private

Operating status

Operating

Headcount range

1–10 employees

Short description

The Good AI Lab is a Stockholm-based non-profit AI research collective publishing open-source computer vision and NLP frameworks for humanitarian applications, in partnership with MSF and FMSI, serving NGOs, academic researchers, and graduate students globally.

Ownership category

akta.pro rank

Product category

AI Research and Computer Vision

NAICS

Computer Systems Design and Related Services (54151), Research and Development in the Social Sciences and Humanities (54172)

SIC

Services-Computer Integrated Systems Design (7373), Services-Commercial Physical & Biological Research (8731)

akta.pro primary industry

End-to-End Enterprise AI Platforms (MLOps & Model Lifecycle Management) (HDAEANAA)

akta.pro secondary industries

Responsible AI, Security & Privacy Platforms (Safety, Guardrails, PII) (HDAEANAG), Responsible AI, AI Governance & Compliance Services (BPAEAHAJ), Regulatory Readiness & Audit Automation (e.g., EU AI Act, NIST AI RMF, ISO/IEC 42001) (HDAAAMAE)

Keywords

• Computer vision research
• Open-vocabulary segmentation
• Domain adaptation frameworks
• Humanitarian AI applications
• Open-source AI research

Headquarters

HQ city

Distributed/Virtual

Offices1 record

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Operations, Infrastructure, Marketing or Sales, Others

Revenue model

1. Non-profit Research Funding: The organization is a non-profit AI research collective funded through partnerships, collaborations, and potentially grants. Revenue is not generated from product sales but from collaborative research projects with humanitarian organizations and academic partnerships.

Go-to-market motion3 records

Distribution channels3 records

Marketing channels8 records

Core offering

The Good AI Lab is an independent non-profit AI research collective that conducts foundational academic AI research—primarily in computer vision, domain adaptation, and open-vocabulary semantic segmentation—and publishes the resulting open-source frameworks and code. The lab collaborates with humanitarian organizations such as Doctors Without Borders (MSF) and FMSI to co-develop applied AI tools (e.g., UPR Matcher) and runs the CV4GOOD workshop at ECCV 2026 to bridge computer vision research with humanitarian practice.

Product overview

The Good AI Lab is an independent non-profit AI research collective focused on foundational academic AI research for social good. The lab operates as a research-first organization rather than a commercial product company, developing open-source frameworks published at top-tier AI conferences. Their core offerings include SemLA (CVPR 2025) and VocAlign (BMVC 2025) for open-vocabulary semantic segmentation with domain adaptation, HAMLET (ICCV 2023) for real-time adaptation, and UPR Matcher for humanitarian NLP applications. The lab also runs the CV4GOOD workshop at ECCV 2026 in partnership with MSF. All code is shared openly on GitHub, and the lab collaborates with universities, research institutes, and non-profit organizations worldwide.

Differentiator

Problem solved

Functional benefit

Products and services

• SemLA (Semantic Library Adaptation) A training-free, test-time domain adaptation framework for open-vocabulary semantic segmentation that dynamically retrieves and merges LoRA adapters based on target domain proximity in CLIP embedding space; published at CVPR 2025 and available as open source on GitHub for researchers and practitioners.
• VocAlign (Vocabulary Alignment) The first framework for source-free domain adaptation in open-vocabulary semantic segmentation, using vocabulary alignment, LoRA-based efficiency, and top-K class selection; published at BMVC 2025 and open-sourced on GitHub for researchers and practitioners.
• HAMLET (Hardware-Aware Modular Least Expensive Training) Real-time domain adaptation framework for semantic segmentation combining hardware-aware back-propagation orchestration (HAMT) with domain-shift detection, achieving semantic segmentation with simultaneous adaptation at 29+ FPS on a single consumer-grade GPU; published at ICCV 2023 and open-sourced on GitHub.
• UPR Matcher AI-powered tool developed with FMSI that automatically identifies and surfaces meaningful connections between civil society recommendations and UN Universal Periodic Review recommendations, using semantic matching with BGE embeddings and MiniLM cross-encoder; designed for humanitarian advocacy organizations tracking human rights policy.
• CV4GOOD Workshop ECCV 2026 workshop on computer vision for humanitarian action, co-hosted with Médecins Sans Frontières, bringing computer vision researchers and frontline humanitarian practitioners together to shape a field-informed research agenda; serves as an engagement channel for the AI-for-good community.

Quantifiable outcome

• +6 mIoU improvement on CityScapes benchmark with VocAlign
• +2 more outcomes

Named customers2 records

Segments3 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability9 records

Feature5 records

Partnerships

Eleven partnerships are on record, tiered flagship and core.

• Doctors Without Borders (MSF Italy)flagshipStrategic or Co-development Partner · 17 October 2025The Good AI Lab partnered with MSF Italy in October 2025 to explore how AI can strengthen humanitarian action responsibly. The collaboration focuses on two major workstreams: AI Training (sessions with MSF personnel from directors to field operators) and AI Strategy (determining how AI can be responsibly integrated into MSF operations and long-term planning). The partnership includes co-hosting the CV4GOOD workshop at ECCV 2026.
• FMSI (Fondazione Marista per la Solidarietà Internazionale)coreStrategic or Co-development Partner · 23 July 2025Partnership to develop UPR Matcher - an AI tool that automatically identifies and surfaces meaningful connections between civil society recommendations and UN Universal Periodic Review recommendations. FMSI contributes domain expertise in child rights and advocacy work within the UPR process, while The Good AI Lab provides the technical implementation using semantic matching with BGE embeddings and MiniLM cross-encoder.
• Médecins Sans Frontières (MSF)flagshipStrategic or Co-development Partner · 1 January 2025Co-host of CV4GOOD workshop at ECCV 2026 - a half-day workshop bringing computer vision researchers face to face with frontline humanitarian practice. MSF contributes practitioner perspectives, needs-from-the-field framing, and governance expertise alongside researchers from Conflict Ecology Lab at Oregon State, MIT, and others.
• KTH Royal Institute of TechnologycoreTechnology or IntegrationAcademic collaboration partner. Team member Hedvig Kjellström is a professor at KTH. The organization describes itself as bringing together scientists, professors, engineers, researchers, and developers from top universities including KTH.
• University of TorontocoreTechnology or IntegrationAcademic collaboration partner. Researcher Gianluca Villani holds a PhD from University of Toronto. Team members include researchers from this institution.
• Technical University of MunichcoreTechnology or IntegrationAcademic collaboration partner. Multiple papers include co-authors from TU Munich (Daniel Cremers, Linus Härenstam-Nielsen) and the Munich Center for Machine Learning.
• AMD Silo AIcoreTechnology or IntegrationIndustry collaboration partner. Multiple team members (Jussi Karlgren, Pier Luigi Dovesi, Carl Persson, Adrian Yijie Xu, Alexander Aurell, George Stampoulidis) are employed at AMD Silo AI. The organization benefits from GPU expertise and computational resources.
• University of BolognacoreTechnology or IntegrationAcademic collaboration partner. Matteo Poggi is an assistant professor at University of Bologna, and multiple papers include co-authors from this institution.
• ETH ZurichcoreTechnology or IntegrationAcademic collaboration partner. Silvio Mazzucco was previously at Computer Vision Lab at ETH Zurich, Mattia Segù is affiliated with ETH Zurich, and multiple co-authors from this institution.
• GooglecoreTechnology or IntegrationIndustry collaboration partner. Federico Tombari is Director of Research at Google Zurich, Mattia Segù is a Research Scientist at Google, and Marc Botet Colomer is a Software Engineer at Google.
• Sapienza University of RomecoreTechnology or IntegrationAcademic collaboration partner. Multiple team members (Donato Crisostomi, Maria Rosaria Briglia, Indro Spinelli, Alessio Palma) are affiliated with Sapienza University of Rome.

Scale indicators4 records

Recent moves6 records

Expansion highlights5 records

Emerging players

• DataKind: Non-profit organization mobilizing data scientists for humanitarian and social impact projects — directly comparable end-customer overlap (NGOs, humanitarian agencies) and pro-bono engagement model.
• Hugging Face: Open-source AI community platform hosting model and dataset repositories — comparable open-source distribution model and community-led GTM, though commercially oriented.
• Partnership on AI: Non-profit multi-stakeholder organization convening researchers, NGOs, and industry on responsible AI — overlaps with The Good AI Lab's mission to channel AI research toward humanitarian and ethical applications.
• Omdena: AI-for-social-impact collaborative running global community builds for humanitarian and development organizations — comparable focus on AI applied to humanitarian challenges via distributed teams.

Direct peers

• DAIR Institute: Distributed AI Research institute focused on independent, community-rooted AI research outside big tech — aligns with The Good AI Lab's decentralized, values-driven research model.
• EleutherAI: Open-source AI research collective producing large language models and publishing at major venues — comparable decentralized, volunteer-driven research model with no commercial product.
• Mila (Quebec AI Institute): Non-profit AI research collective with 1,000+ researchers across academic institutions producing foundational research — comparable structure, open-science ethos, and deep learning publication track.
• Allen Institute for AI (AI2): Non-profit AI research institute producing open-source models and top-venue publications, structurally the closest analogue to The Good AI Lab as a mission-driven research organization publishing in CV/ML.

Broad incumbents

• MIT CSAIL: Large-scale academic AI research laboratory producing foundational CV and ML work at top venues — comparable research output but embedded in a single major university rather than a distributed collective.
• Stanford HAI (Human-Centered AI Institute): Established academic AI institute focused on human-centered, multidisciplinary AI research — comparable mission to advance AI for societal benefit, but with vastly larger institutional backing.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat4 records

Key risks6 records

Key highlights7 records

Customer concentration

• GitHubgithub.com/The-Good-AI-Lab
• LinkedInlinkedin.com/company/the-good-ai-lab
• Instagraminstagram.com/thegoodailab

Revenue estimate

Valuation estimate

Number of profiles

Profiles10 records

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about The Good AI Lab