I started Bricklayer in March of 2023, a year ago today. When I started, Generative AI (GAI) was barely a discussion in the cybersecurity industry. We have come so far in the last year, both within and outside of cybersecurity.
I thought then and still believe today that GAI is going to be a linchpin for the most revolutionary technical advancements that mankind has ever seen. The impact it will have will surpass what we saw with the internet and will transform every industry.
People often ask me why generative AI is unique/different from other advancements we have seen in Machine Learning (ML) and AI (pre GAI). My answer is that GAI allows people and machines to understand each other. When we can instruct machines to do things in flexible terms and communicate with us in kind, we have a new paradigm that allows us to interact, work together in a flexible manner, and solve problems with our AI counterparts.
In Star Trek, the Universal Translator played a crucial role in the formation of the United Federation of Planets. It allowed for communication between different species and cultures, overcoming language barriers and promoting understanding and diplomacy. This ability to communicate was foundational in establishing alliances, negotiating peace, and fostering cooperation among the diverse members of the Federation, leading to its formation and ongoing success. Thus, I see GAI as our universal translator. Today, we are using it to allow computers and people to understand each other, but I foresee our AI counterparts also deciding that natural language is a great way to communicate amongst themselves, thereby bridging the communication gap between machines as well.
While others have been building GAI capabilities into their current software (Co-Pilots) and or using GAI to invent/reinvent a solution to a particular task or create a new tool, we decided to go in a different direction.
We looked at what makes the cybersecurity industry powerful and modeled it – People are the most critical element of the cybersecurity industry. Without them we wouldn’t have the people (obviously) or the processes or the technology we need to bring together the powerful combination of “people, process and technology”. So rather than start with technology, we started with people. Feels strange to say that as a technology company, but the truth is that we are applying a technical solution to a not so technical problem.
While others were looking for people to drive their AI capabilities (requiring more pilots), we built software to allow our customers to create and hire cybersecurity-trained AI Agents, integrate them within their existing security operations processes and technologies, and have them work for them – same as people do.
An autonomous AI agent is a specialist with the ability to plan, make decisions, and communicate with other AI agents or human co-workers. Think of each one as a member of a team, with specific skills and a particular job to do. Autonomous planning and decision making is at their core and is what makes them AI Agents instead of yet another GAI or Large Language Model (LLM) based use-case. It’s their ability to think and act that makes them really powerful.
Now, one could argue that they could have one agent that can do it all. For example, I create a Security Analyst AI Agent that can span all my cybersecurity needs. This means that this one Agent would have experience in every area or role, access to all pertinent organizational knowledge, be capable of planning for any security workflow, and would integrate into all tools required. This could be done, but I think we can all agree that a better way of handling this is through separation of duties, specialization of each, and the ability to employ multiple agents together in a collaborative way. Afterall, this is how people work, and that works really well.
By combining multiple AI agents, each with their tools and knowledge, you can dissect a larger task into sub-tasks and tackle them simultaneously with specialized capabilities. This separation allows for more focused processing for each task, and allows for multiple tasks to be achieved as part of a larger task. It’s the difference between a solo expert working alone and a team of experts collaborating.
We see a future where the security organization can hire cybersecurity trained AI Agents to take on specific tasks and processes that previously could only be done by their human workforce. We don’t see this as replacing human workers, given the massive cybersecurity talent shortage that exists. This allows existing cybersecurity teams to relocate their workforce away from the tasks that are monotonous and leading to burnout and retention issues, while saving time not needing to constantly hire and rehire human workers into these lower level tasks.
