Loading organizations...
§ Private Profile · Tel Aviv, Israel
Hirundo is a technology company.
Hirundo offers a machine unlearning platform for artificial intelligence models. It enables organizations to surgically remove problematic data, biases, and vulnerabilities like hallucinations or private data memorization directly from their AI systems. This technology precisely modifies model parameters, providing efficient remediation and ensuring compliance and performance without extensive retraining.
Founded in 2023 by Ben Luria (CEO), Michael Leybovich (CTO), and Professor Oded Shmueli (Chief Scientist), Hirundo recognized existing AI risk management solutions were inadequate. Their expertise pioneered machine unlearning, developing a fundamental approach to quickly remediate problematic AI outputs and enhance enterprise AI integrity.
Hirundo's platform assists enterprises deploying complex AI models, especially large language models, in production. It helps organizations maintain trustworthy, compliant AI systems by swiftly addressing risks. The vision is to integrate machine unlearning as a core component of AI governance, fostering secure, responsible AI adoption industry-wide.
Hirundo has raised $8.0M across 1 funding round.
Hirundo has raised $8.0M in total across 1 funding round.
Hirundo has raised $8.0M across 1 funding round. Most recently, it raised $8.0M Seed in June 2025.
| Date | Round | Lead Investors | Other Investors | Status |
|---|---|---|---|---|
| Jun 1, 2025 | $8M Seed | Maverick Ventures Israel | SuperSeed, Mads Jensen, Yaron Carni, AI Fund, Alpha Intelligence Capital, Plug And Play Ventures, Tachles VC | Announced |
Hirundo has raised $8.0M in total across 1 funding round.
Hirundo's investors include Yaron Carni, SuperSeed, Mads Jensen, AI Fund, Alpha Intelligence Capital, Plug and Play Ventures, Tachles VC.
Hirundo is a Tel Aviv- and London-based AI startup founded in 2023 that builds the world's first Machine Unlearning Platform, enabling AI models to "forget" unwanted data, biases, hallucinations, and vulnerabilities without retraining.[1][2][3][4][5] The platform serves enterprises in high-stakes sectors like finance, healthcare, defense, and consumer-facing industries, solving core AI deployment risks such as privacy violations, regulatory non-compliance (e.g., "right to be forgotten"), poisoned data, and inaccuracies that block production readiness.[2][3][4][5] It delivers results in under an hour at less than 5% of retraining costs, with proven outcomes like 85% reduction in jailbreaks, 70% bias reduction, and 100% PII removal, while supporting LLMs, computer vision, radar, LiDAR, NLP, and more via API, SaaS, VPC, or on-premises deployment.[3][4][5]
In June 2025, Hirundo raised $8 million in seed funding led by Maverick Ventures Israel, with participation from SuperSeed, Alpha Intelligence Capital, Tachles VC, AI.FUND, and Plug and Play Tech Center, to commercialize its patent-pending technologies.[2][4] Already piloting with multinational corporations and government agencies, the company merges academic AI research with practical tools, positioning it for rapid growth in the exploding demand for trustworthy AI.[1][4]
Hirundo was founded in 2023 by Ben Luria (CEO, Rhodes Scholar, serial entrepreneur with prior roles at Oxford and co-founder of fintech Worqly and nonprofit ScholarsIL), Emeritus Professor Oded Shmueli (former Dean of Computer Science and EVP at Israel's Technion, with AI work since 1995 at IBM, HP, AT&T, and more), and Michael Leybovich (graduate CS researcher at Technion and award-winning R&D officer).[1][2][4] The idea emerged from the limitations of traditional AI training—a "one-way track" where removing bad data required costly full retrains—pioneering "machine unlearning" to surgically excise issues like biases, hallucinations, and sensitive data.[1][3][4]
Early traction built on the founders' expertise: Luria's entrepreneurial experience, Shmueli's decades in AI academia, and Leybovich's data lineage skills enabled rapid development of patent-pending solutions.[1][2][4] By 2025, this led to the $8M seed round and pilots, marking a pivotal shift from research to enterprise product amid rising AI regulations and safety needs.[2][3][4]
Hirundo rides the trustworthy AI wave, addressing AI's "unresolved core issues" like hallucinations, biases, and privacy as models scale and regulations (e.g., EU "right to be forgotten," GDPR) intensify—market forces de-risking enterprise adoption where most mission-critical AI fails production.[1][3][4][5] Timing is ideal post-2023 AI boom, with unlearning emerging as essential for generative and non-generative systems amid vulnerabilities from poisoned data and fine-tuning.[2][3]
It influences the ecosystem by enabling safer AI in regulated sectors, reducing retrain barriers (key for cost-sensitive firms), and boosting accuracy via automated QA—pioneering a field that could standardize "AI hygiene" like data cleaning, fostering broader, responsible deployment.[3][4][5]
Hirundo is poised to dominate machine unlearning as AI safety becomes table stakes, with its seed funding fueling commercialization and pilots converting to revenue amid exploding demand for bias-free, compliant models.[2][4] Next: Expand to more modalities, scale enterprise wins, and leverage regulations/trends like multimodal AI and federal AI mandates; potential for acquisitions by hyperscalers needing remediation tools.[3][5]
Its influence could evolve from pioneer to infrastructure layer, "fixing AI at its core" to unlock trillions in stalled deployments—ensuring models know only what they should, as AI permeates everything.[1][5]