This site uses cookies to ensure the best viewing experience for our readers.
Why OpenAI and xAI angels are betting on Israeli AI startup Tzafon

Why OpenAI and xAI angels are betting on Israeli AI startup Tzafon

Tzafon is building foundational models that interact with computers the way we do. It has already secured funding from angels at top AI companies and signed a deal with Google Cloud.

James Spiro | 06:39, 25.08.25

When angels from two of the most talked-about names in artificial intelligence — OpenAI and Elon Musk’s xAI — take notice of a young Israeli startup, people pay attention. Tzafon, a Tel Aviv-based company with offices in Stockholm and headquarters in the US, is developing technology that could reshape the way machines reason and interact with humans.

The company is building foundational models that interact with computers like humans, such as clicking, scrolling and inputting text. Its infrastructure can support autonomous AI agents across multiple fronts that all work together to address challenges and interact with operating systems, applications, and web browsers. This year, Tzafon raised a $9.7 million pre-Seed round, including angel investment from players at xAI and OpenAI.

The Tzafon team. The Tzafon team. The Tzafon team.

“I think the first thing they really liked about us was our team,” said Mark Rogers, co-founder and CTO. “They said that our team is very good and also really like the vision of improving computer use agents with humans in the loop. They seem to resonate with this very strongly.” According to Rogers, Tzafon currently has 15 employees, including former workers from companies such as DeepMind, Palantir and Character AI - and there is no sign of using the money to grow the team more than necessary. “We believe very strongly in the lean startup,” he added. “We prefer to select really good engineers and pay them really well.”

AI agents are autonomous software systems that use artificial intelligence to perceive their environment, reason, make decisions, and perform actions to achieve specific goals without constant human intervention. Analysts have pointed out that the next major evolution in AI may come not from larger models, but from how those models are deployed as agents capable of performing multi-step tasks. Companies like Adept and Inflection have made headlines for similar visions, but Tzafon’s cross-platform and explainability-first approach sets it apart.

The race for AI dominance is no longer just about building bigger models; it’s about solving the challenge of explainability and trust. That global players are watching Tzafon is proof that the startup may be onto something. Co-founded in 2023 by Rogers and CEO Noah Löfquist, the startup is working on its primary feature, Lightcone, as an agent that acts on a user’s behalf across apps and platforms.

For an everyday user, this could mean asking an AI to book a flight, process an insurance claim and collect receipts, or prepare a report. Then, the user could watch it interact with the browser and applications just like a human assistant would. Unlike LLMs like ChatGPT that simply provide text answers, Lightcone is designed to actually ‘do the work’ and improve explainability and trust in the process.

Related articles:

The product is downloadable and runs locally on a computer like an app. According to one of the founding engineers, Noga Bregman, one way to ensure explainability and build trust is to have open models where everything is stored locally. “This is like one thing we are doing to handle this concern,” she said. “It's definitely like something we have internal discussions on, and we really keep that in mind. All of us come from AI safety worlds and really care about AI safety, so I think it's definitely something we bear in mind.”

Tzafon is already born into an Israeli ecosystem that has been thriving through the AI revolution, adapting to a changing cybersecurity world, and adjusting to the needs of a world that is being engulfed by new technology. The country has often held a lead in cybersecurity, and more AI companies are leaning into safety and reliability: two themes that resonate strongly in today’s debates over artificial intelligence.

A recent partnership with Google Cloud to help build the next generation of agentic machine intelligence is proof that Startup Nation is attracting attention from almost every leading player in the game.

According to Google, “Tzafon will partner with Google Cloud to access the compute resources and cloud services it needs to train its new multi-agent models – and to develop new automation frameworks that will allow Tzafon’s agents to collaborate more quickly and seamlessly.” The partnership will also provide Tzafon with access to NVIDIA GPUs, Google Kubernetes Engine, BigQuery, and Google Cloud’s complete AI stack.

The company had originally secured a $4.4 million pre-Seed round led by Streamlined with participation from Kakao VC, Oliver Jung, and angels from OpenAI and xAI. But the extension from HV Capital brought its total funding to the $9.7 million mentioned above. Teams will remain across its three locations in Tel Aviv, Stockholm, and San Francisco, and continue to attract eyes from all big players. As AI giants look beyond LLMs, Israeli startups like Tzafon could shape the industry’s next leap.

share on facebook share on twitter share on linkedin share on whatsapp share on mail

TAGS