Bricklayer is an innovative product that leverages generative artificial intelligence (AI) to replicate human-like processes for data retrieval, reasoning, and drawing conclusions. By employing cutting-edge AI technologies, Bricklayer aims to enhance the efficiency and effectiveness of cognitive tasks traditionally performed by humans. This blog provides insights into how Bricklayer utilizes generative AI to mimic human processes and facilitate data-driven decision-making. At Bricklayer our mantra is – Ask, Analyze, and Answer. Let’s walk through how this actually works.
Ask
The first step in any Bricklayer use-case is to simply ask the question you want an answer to. Well formed questions go a long way in terms of receiving value from Bricklayer. So hold on tight as we think about the types of questions you can ask.
Here are several questions that Bricklayer can readily answer:
- What are the latest threats to [category], [industry], [product]?
- What do we know about [malware], [vulnerability], [threat group], [IOCs]?
- Can you tell me about an incident at my company where the following IOCs were used?: “[IP address], [domain name], and [hash]”.
- Analyze these [domain names], [IP addresses], and [hashes] and provide a summary of each including: severity, summary, and existence in any organization incident reports. Provide the incident report name and date.
- A user is asking to turn on bluetooth on their company issued laptop. Based on company security policy, can they turn on bluetooth, and what are the requirements?
Analyze
Analyze performs three major functions: data retrieval, reasoning, and conclusion making. The great thing about Bricklayer is that all of these things are completely automatic. You just need to wait for them to finish… which sometimes can be hard. But be patient, the wait is well worth it.
Data Retrieval: Bricklayer harnesses generative AI algorithms to retrieve and analyze vast amounts of data from diverse sources allowing for situationally aware answers. By employing natural language understanding and machine learning, Bricklayer can swiftly and accurately access structured and unstructured data sets. This advanced data retrieval capability allows Bricklayer to efficiently gather the information necessary to provide a contextually aware answer.
- Security Blog Posts – Bricklayer consumes and analyzes thousands of cybersecurity blog posts increasing its knowledge of current events. Each day you can expect Bricklayer to have insights into what’s been happening in the cybersecurity world.
- Private Data Stores – Bricklayer’s data store feature allows you to integrate your own private organization data into the decision making power of Bricklayer AI. Whether this be your security process and policy, risk or incident reports from the last several years, Bricklayer can leverage this data to inform its reasoning and decision making processes.
- External Product Integrations – Bricklayer supports the ability to automatically integrate with third party APIs and enable a natural language interface. This opens the floodgates for integration of any/all data that an organization has at its disposal into the Bricklayer product.
Bricklayer Security Blog Posts for current news
External Product – National Vulnerability Database for latest vulnerability information
Data Store Creation
Querying documents in Private Data Stores using Natural Language
Reasoning: Bricklayer employs generative AI algorithms to mimic human-like reasoning processes. Data retrieved is used as context, enriching its knowledge about the question being put forth, and is used to deeply understand what is being asked of it. This is further aided by using a network of LLM-based agents that have specialized roles along with their own set of tools similar to a team of specialists at any company – one to plan, one to research, one to report and many more. Questions are answered and performance is improved through breaking questions down into constituent tasks that are specifically assigned to each of these. Notably, each one demonstrates the capacity to reason through thinking for themselves and observing the responses of other specialists. Here is a brief overview of some of the roles.
- Security Team Leader – leads the coordination efforts across the digital team.
- Security Researcher – researchers are responsible for retrieving relevant information required to support reasoning.
- Security Analyst – analysts leverage the outputs of researchers to perform analytic processes on data.
- Security Reporter – reporters are trained out to summarize and report and research and analysis.
Conclusion Drawing: Bricklayer is able to draw meaningful conclusions based on the information it has and its observations. The large corpus of data accessible to the models provides a wealth of background knowledge as it emulates reasoning empowering Bricklayer to help users make informed decisions and take strategic actions. Most importantly, these conclusions are reliable, verifiable and transparent.
- Reliability – The conclusions drawn by Bricklayer are based on reliable data from well-known blogs, cybersecurity textbooks, your own organization data, and established APIs.
- Verifiability – Bricklayer allows users to verify all the information it responds with through providing the sources it based the answer on. These sources can be looked at as verification and as resources for further information.
- Transparency – Bricklayer strives to be as transparent as possible by showing users all the steps it took to reach a conclusion. This is done by providing a log of all the agents and tools used along with their inputs.
Plan and Sources highlighted for trust and transparency
Answer
This is what it is all about right – Answering your question.
Bricklayer’s utilization of generative AI for data retrieval, reasoning, and conclusion drawing brings several advantages. This includes:
- Rapid sourcing of relevant structured and unstructured data from inside as well as outside of the organization at speeds far surpassing human capacities. For example, information on company policy, threat intel, et al.
- Accelerated goal-oriented processing of that information.
- Enabling data-driven decision-making through an easy-to-use UX.
Bricklayer empowers organizations to leverage data-driven insights for improved strategic planning, risk assessment, security analysis, and more. The product’s ability to replicate human-like processes while harnessing the power of AI enables users to unlock valuable intelligence and enhance decision-making across a broad range of use-cases.
How this is done
- Large Language Models – Bricklayer employs multiple generative AI models for various purposes. Each agent can have its own unique model(s) configured and model configuration and prompting is customized for each.
- Agents and Tools – The Bricklayer product supports the digital equivalent of human team members and teams. Agents, like team members, can take direction and intelligently plan and execute work while employing tools that are required to accomplish the primary and subtasks. This hierarchy closely resembles the natural world of appropriately skilled and capable people that plan and execute against problems or tasks. Furthermore, these agents and tools are natural language driven to allow for easy communication.
- External API Integrations – One of the biggest opportunities with Bricklayer is to expose existing product API’s within Bricklayer as Intelligent Services (combination of agents and tools). This allows you to ask a natural language question and for the Bricklayer AI to turn that question into API calls across multiple backend products. This coupled with Bricklayer AI’s ability to automatically summarize the response makes it easier than ever to retrieve data and take actions across the product ecosystem you already have.
- Privacy – Organizations are rightfully worried about how their private data is used within AI products. Bricklayer has architected our product to maintain user privacy throughout the process of configuration and usage. Our LLMs and cloud infrastructure are managed by Bricklayer directly and we maintain complete control over customer data. All customer data is encrypted and leveraged in a secure manner.
