Introducing Simbian and Fully Autonomous Security
Ambuj Kumar
3 min read
Ambuj Kumar
3 min read
Security is a breadth-first problem
Some problems need to be solved in their entirety. Cybersecurity is one such problem.
It does not matter that your endpoints are secure if you cannot keep users from clicking
on bad links. It does not matter that you have correctly identified your sensitive data
and have proper access control in place, if your identity provider is prone to abuse. And so on.
 
Security is a breadth-first problem that requires all its components to work together. On the other hand, the cat and mouse game between attackers and defenders means that these different components evolve at different speeds. This leads to a complex security ecosystem, famously fragmented and constantly evolving. Â
 
Security is a breadth-first problem that requires all its components to work together. On the other hand, the cat and mouse game between attackers and defenders means that these different components evolve at different speeds. This leads to a complex security ecosystem, famously fragmented and constantly evolving. Â
AI to the rescue
A typical enterprise uses 45 different
security tools to secure an average of 1061 applications, each combination requiring unique set of product knowledge, know-how, and best practices. This is a near-impossible expectation even for the world's best funded security teams. Fortunately, with the new advances in artificial intelligence (AI), especially large language models (LLM), it is possible to build a unified security layer. We can imagine a cohesive unit of intelligent AI agents that are fluent in all the products, have access to the latest threat advisories, and understand all enterprise security events. Obviously, this idea, while promising, is right at the edge of possibility with multiple challenges that would require deep technical innovations.
AI is the only rescue
The reality is that if we want to be able to control security, we must solve all the challenges in making AI-driven security a reality. In fact, AI is our only rescue. On its current trajectory, security is going to get much worse – thanks to unavoidable use of AI by nefarious actors. Automated cyber espionage, realistic phishing artifacts, LLM-driven malware generation are all going to drastically lower the cost for adversaries. Defense cannot be more expensive than the cost of attack. We must use AI to its fullest on the defense side.
Simbian's promise - Fully Autonomous Security
Today, we are pleased to launch Simbian, a mission driven company to solve security with a $10M+ over-subscribed funding round, led by Cota Capital, Icon Ventures, Firebolt, Secure Octane, and Rain Capital and supported by more than 15 of the most successful business leaders of our times.
 
Simbian is building a Fully Autonomous Security platform for enterprises. Fully autonomous security means letting humans make all the strategic decisions while AI actually implements those decisions. Today, security operators spend so much energy on product integrations, low-level APIs, triaging low-value events, and completing one-off tasks, that they find little time for important strategic decisions. It's similar to having a car with such a complicated transmission system that you spend all the time on changing gears, and forget to steer the car. Simbian's fully autonomous security will turn it upside down - letting you steer the security car, using AI for everything else.
 
Simbian is building a Fully Autonomous Security platform for enterprises. Fully autonomous security means letting humans make all the strategic decisions while AI actually implements those decisions. Today, security operators spend so much energy on product integrations, low-level APIs, triaging low-value events, and completing one-off tasks, that they find little time for important strategic decisions. It's similar to having a car with such a complicated transmission system that you spend all the time on changing gears, and forget to steer the car. Simbian's fully autonomous security will turn it upside down - letting you steer the security car, using AI for everything else.
Surmounting the AI challenges with TrustedLLM™
Using LLMs comes with its own set of risks and challenges. If an LLM hallucinates while taking a security action, it may block the wrong network port, or uninstall a production application by mistake, or worse, create a deeper security hole. Threat advisories can have hidden prompts hijacking the LLM. Security operators deal with a lot of sensitive information that must be guarded carefully. These are just some of the challenges in making LLMs reliable and trustworthy for security automation.
This is why we built the world's first hallucination-free system, TrustedLLM™, that is purpose-built for safety and reliability. This is based on our patent-pending technical innovations, our unique crowd-sourced intelligence from
a game, and our leading research.
Come, build with us
A journey of a thousand miles begins with a single step. The path towards Fully Autonomous Security that we have chosen for ourselves is the only path possible to keep organizations safe in future. We set this goal for us, not because it is hard or ambitious (although those make it fun) but because our collective well-being depends on it.
We are working with selected forward-leaning organizations as we make this vision a reality. Please contact us if this is you.
We are working with selected forward-leaning organizations as we make this vision a reality. Please contact us if this is you.