Introducing Israel’s GenAI Startup Map

The value of such a map is not only to understand the competitive landscape. For that, one can refer to global landscapes that showcase companies in the field from all over the world. Rather, the goal is to shine a spotlight on the Israeli GenAI ecosystem – to comprehend how it is evolving, which companies can serve as role models and sought after for valuable advice, who competes for local talent and local capital, whether Israel holds a relative advantage in certain domains, and what the maturity level of the Israeli GenAI ecosystem is as a whole.

The map is interactive, allowing anyone to add companies and modify their details (subject to review). It is continuously evolving and updated regularly. It can be organized by different criteria, based on the user’s needs:

1. By layers – Foundation Models, Infrastructure (LLM/ML Ops, Data, Cybersecurity, etc.), Application.
2. By sectors – Business Tools, Cybersecurity, DevTools, Fintech, etc.
3. By the main data type the company processes/generates – Image, Text, Video, etc.

The map displays only startups that meet all of the following criteria:

1. Startups where GenAI/infrastructure for GenAI is an integral part of their core technology – not just a feature. For the purpose of this map, GenAI is defined as AI systems that use pre-existing content to generate new & original content in response to a query and rely on foundation models for that purpose. Analytical/ traditional AI is not and will not be represented.
2. Startups that are actively using foundation models in production (only relevant for application layer companies).
3. Infrastructure startups can be considered as GenAI startups if their central value proposition is closely tied to enabling or enhancing the capabilities of GenAI technology.
4. Independent startups that have not yet exited and are still active.
5. Startups at any funding stage (including stealth startups who granted permission to be included).
6. Israeli-related startups (where at least one of the founders is/was Israeli).

A few key takeaways from the current landscape presented on the map:

1. Startup Distribution Across Layers

When dividing the map by layers, we can see that 70% of Israeli GenAI startups operate within the application layer, 28% within the infrastructure layer (including MLOps, Data, Cybersecurity, etc.) and 2% (1 company – AI21) at the foundation model layer.

This shouldn’t come as a surprise. Firstly, new technology initially emerges at the application layer due to a faster time to market. Secondly, just for the sake of context, in general (outside the GenAI context) there are typically more software companies operating in the application layer compared to the infrastructure layer. Lastly, from a demand perspective there is a surplus when it comes to real world applications. As the adoption of GenAI continues to grow, customers will gradually come to recognize the pains and limitations within their infrastructure (this is true only when discussing infrastructure for integrating GenAI models within companies, rather than replacing existing infrastructures entirely, using GenAI). Therefore, it is reasonable to anticipate that over time, as GenAI technology matures, we will witness a rise in the number of companies emerging within the infrastructure layer, particularly in the layer dedicated to the integration of this technology.

Still, it’s important to highlight that more than 12 Israeli companies are opreating within the infrastructure layer, a notable figure in itself. Various factors contribute to this, including Israel’s strong foothold in the infrastructure domain, evident not only in AI but also in other sectors. Additionally, companies specializing in classical AI infrastructure can also offer solutions for GenAI. Notably, approximately 65% of the infrastructure companies on the map were active even prior to the GenAI surge (Aporia, Run:ai, PineCone etc.). They played a role in implementing AI models and their technology is adaptable to GenAI as well. Furthermore, the infrastructure layer encompasses the cybersecurity sector, enhancing its scope. Worth mentioning is the considerable number of companies operating in the domain of Cybersecurity for LLMs, though most of them chose to stay in stealth mode, thus not currently visible on the map.

2. Key Sectors: Marketing, DevTools and Business Tools

Certain industries seem tailor-made for GenAI disruption. This hinges on two factors: (1) how much of a given task can be accomplished by AI and (2) the value that this task provides. 

When categorizing application layer startups by sectors, it becomes evident that the dominant sectors are Marketing (27%), Business Tools such as productivity/writing assistant/sales enablement (20%) and Dev Tools (17%).

Marketing tools are characterized by the presence of the first factor: they need to produce novel content of text, images, and videos. They also need to generate highly personalized content at scale for each individual customer. On top of that, utilizing this technology doesn’t require technical knowledge and can be done effortlessly by marketing teams. The same is true for some of the general business tools (productivity/writing assistant).

As for Dev Tools, both of the factors are relevant: (1) We are witnessing a noteworthy transformation, where an AI-powered assistant tool such as Github Copilot or CodiumAI can handle a substantial portion, roughly 60-70%, of coding tasks. This phenomenon is rooted in the inherent compatibility between the logical and mathematical structures of programming languages and the functioning of language models. The anticipation and recurrence of patterns in structured coding languages exhibit a higher degree of predictability compared to natural languages. This reality enables the automated generation of a significant proportion of code, as long as the system is appropriately guided by prompts. While software developers remain responsible for devising the solution’s architectural blueprint and framework, the majority of the foundational code can be expertly synthesized with the assistance of AI. (2) Considering that developers are costly and that the output of their work lies at the core of the company’s business, it’s evident why optimizing their work is of utmost importance.

As one can see, GenAI models perform well with data that is organized, structured and contains context. In domains where information follows patterns, rules and has a clear relationship between elements. This is why we can expect a real revolution in areas characterized by this, such as legal tools, accounting, grammar tools and more.

———

So far we’ve covered the Israeli GenAI startup map. Its goal is to track the evolution of the Israeli industry in this field over time and enable entrepreneurs to learn from others who have followed a similar path. While there are notable clusters of startups in specific areas, the map will continue to monitor the industry’s progress and update accordingly.

Invitation for Startups

We have been investing in AI since day one, and the same is true for GenAI. Our GenAI investments include Run:AI,, Aporia, CodiumAI, and three recent investments still in stealth mode. We also backed classical AI companies such as Aidoc, Buildots, DeepCure, Immunai and GoodOnes. 

Indeed, there is a lot of buzz around the topic, and all parties should carefully distinguish between fact and fiction. However, we are witnessing a huge technological leap that holds many opportunities, and we have no doubt that exceptional companies will arise from this domain in Israel. So if you are a founder and would like to meet, shoot us a note at [email protected].

*If your company aligns with the specified criteria and is not yet listed, please feel free to add it or reach out to me.