Startup brief
Security Consolidation and AI Frontiers in Tel Aviv
As Cisco pursues a discounted buyout of Zafran, early-stage capital concentrates around AI-native security and brain-decoding models.
Startup
Proven founders are commanding premium seed rounds to tackle AI-identity security gaps — Oak is betting machine-to-machine permission management is the next major defense category.
BackgroundOak was co-founded by Shai Morag and Tal Marom, 2 highly successful serial cybersecurity founders. Morag previously co-founded Ermetic, which was acquired by Tenable for $265M in 2023, and Secdo, which Palo Alto Networks bought in 2018.
- Oak's platform uses artificial intelligence to evaluate access permissions based on active software usage, instantly revoking stale clearance to prevent attackers from exploiting dormant accounts.
- The $60M seed round stands as 1 of Israel's largest early-stage cybersecurity rounds in 2026, signaling that venture capital still aggregates heavily around proven serial founders.
- The technology specifically addresses the rising threat of machine identity compromises, which have multiplied as organizations integrate hundreds of automated AI agents into their software stacks.
Startup
Cisco's discounted pursuit of Zafran shows that the era of inflated cybersecurity exit premiums is over — forcing even well-funded startups to accept flat valuations to survive.
BackgroundZafran Security was founded in 2022 and quickly raised over $130M from top-tier venture firms including Sequoia Capital, Menlo Ventures, and Cyberstarts. The company had achieved a valuation exceeding $200M during its Series C round in late 2025.
- The acquisition talks come amid internal friction, including the sudden departure of co-founder Snir Havdala, signaling deep strategic rifts over the startup's exit path.
- Investors are bracing for a thin return on capital as the proposed $150M buyout barely covers the $130M raised to date, leaving little profit for early backers.
- The discounted deal highlights a broader consolidation wave in the cybersecurity market where mature platforms absorb cash-strapped point solutions that can no longer raise independent growth capital.
Startup
Hemispheric is applying large language model scale to non-invasive brain activity — decoding neural signals without surgical implants could commoditize clinical brain diagnostics.
BackgroundHemispheric was co-founded by computational neuroscientist Dr. Hagai Lalazar and AI entrepreneur Gidi Littwin. Littwin previously co-founded RealFace, the facial recognition startup that Apple acquired in 2017 to form the basis of its Face ID technology.
- The $52M funding round was backed by prominent local and global venture firms including Hanaco and OurCrowd, securing the startup a multi-year runway for deep-tech clinical research.
- The Descartes foundation model translates complex, non-invasive electrical brain signals into quantitative insights, allowing pharmaceutical and clinical partners to assess neurological drug efficacy without surgical implants.
- The startup aims to establish a standardized AI translation layer for neural activity, bypassing invasive brain-computer interfaces to accelerate the commercialization of brain-reading applications.
Unlock the full brief
Sign in to read every signal, takeaway, and source. Free account — Apple, Google, or email.