Bryan Vann is the co-founder and CTO at Kindo where he specializes in AI security, LLM application architectures, RAG-based applications, agentic behavior and MLOps. During his previous 16 years at Google, he was a co-architect of Google Drive and the senior director of engineering where he grew the backend Engineering team from 5 engineers, to a department with more than 230 people. He also served as a key technical leader for Google Workspace, leading security and compliance initiatives, as it grew from zero to $10 billion in annual revenue.
Could you tell me more about your background?
I received both my undergrad and masters degree in computer science from UCLA. After having worked at a couple startups before grad school, after grad school I joined Google. I initially worked on the Google Photos backend, but after a year I joined up with another colleague to be the first individual contributor on a backend storage system that would eventually become the backend for Google Drive. I was one of the key architects that designed and scaled it up to a billion user product, and eventually shifted into management and scaled up the engineering organization. The last few years of my time on Drive and Workspace were focused on enterprise security, governance, and compliance. Features like client side encryption, data regions, access controls, legal hold, among others were some of my key focus areas. By the time I left Google I had an org of 230 engineers. I left Google to join Kindo as CTO and co-founder because I was incredibly excited the advancements in AI and I’d always planned on going back to startups. I really believed in the vision of Kindo, that there was a substantial market need, and incredibly interesting technical problems for me to work on.
How did the idea for Kindo come together?
Ron, our CEO was the CSO of several unicorn startups such as Riot Games, Bird Scooters, etc. His hypothesis given his background as a CSO was that enterprises are going to be pressured to adopt new AI technologies (either through a desire to transform their business or simply a need to compete), but that they’d have significant concerns around the security and governance of their data. So he set out to build the product he’d want to buy in his role as a CSO to have a secure way to adopt AI with strong governance and auditability. The vision has since built on that foundation to include helping those teams not only protect against the threats of AI, but to leverage it in their own day-to-day operations to get the transformative efficiency benefits of AI. So our product offers automations for critical workflows for DevOps and SecOps teams so that they can cover more ground with their limited resources, as well as level up the knowledge base and expertise of their teams.
What does Kindo do?
Our AI orchestration and security platform is designed to transform manual DevSecOps processes into automated workflows, while maintaining enterprise-grade security and compliance standards. Features include:
- AI-Powered Runbook Automations – automates workflows for DevSecOps teams, handles incident response procedures, manages vulnerability assessment and remediation, and integrates with existing infrastructure.
- WhiteRabbitNeo AI Model – specialized AI model for offensive cybersecurity, was trained on Indicator of Compromise (IoC) data, threat actor intelligence, CVE database, National Vulnerability database, and enterprise infrastructure documentation.
- No-Code AI Agents – offering an OpenAI-compliant API, a Dev Copilot that’s compatible with multiple IDEs, an automated first-level alert analysis, and system-of-record execution capabilities.
- Enterprise Security Controls – Data Loss Prevention (DLP) filters, Access Control Lists (ACL), Single Sign-On (SSO) integration, comprehensive audit logging, and PII anonymization and redaction capabilities.
- Compliance Framework – includes SOC 2, HIPAA, GDPR, ITAR, FINRA, OWASP Standards, NIST AI RMF, EU AI Act compliance, among others.
What challenges have you faced recently and how did you overcome those challenges?
One challenge has been the dichotomy between how fast the innovation in AI technology is moving vs how quickly large enterprises are adapting to it. As someone deeply steeped in AI technology, it’s easy to start assuming this is common knowledge and that the rest of the world is on the same page. But in GTM we’ve realized that while companies know they need to adopt AI technology to compete and transform their businesses, they are often not sure where best to apply AI to get the maximum benefits. Ron’s hyptothesis about the concerns around AI have also been proven true, there are signficant concerns around training on their data, and other new threats that are specific to AI such as prompt injection or toxic content. Fortunately solving those problems is core to what we offer, but we’ve realized a big part of the role we need to play in GTM is working as advisors to educate and help our customers figure out where AI can be best applied to increase the efficiency of their operations.
How has the Kindo evolved since launching?
The pace of innovation in the AI field has been staggering over the past 2 year, much faster than any technological wave we’ve ever seen before. Every week there are a slew of new advancements – new models, new frameworks, new agentic techniques, new bleeding edge research on things like infinite context window. It’s part of our core values at Kindo that we are a learning company. We spend hours and hours per week reading up on the latest advancements, and running experiments on the most promising changes. We have a bi-weekly R&D day where we don’t do any roadmap work, and instead the engineering team can pick anything they’ve been interested in learning about to just experiment and build prototypes. A ton of great features have come out of that. But overall, our product has evolved significantly since we first formed. We had built an early prototype of a no-code workflow builder around the time I joined a year and a half ago. The richness of the feature set for our workflow builder (now called agents) has increased dramatically, as has the introduction of chat. Adding the ability to sync your Google Drive and search over it plus searching the web for answers, are key use cases in chat that are general purpose. But we’ve added the ability to create chat bots and train them on a set of knowledge (by uploading documents) has been another really powerful paradigm. More recently we added automations that can be triggered in the background by a ticketing system. Customers can set up agents that run on a trigger so they’ll watch for changes in a ticketing system and then automatically process them through that agent when they come in. We’ve also added a ton of admin functionality such as DLP, dozens of models to select from, comprehensive audit logging, etc.
What differentiates Kindo from its competition?
We’re hyper-focused on DevSecOps (SecOps, DevOps) and are the first AI startup to deliver solutions that can immediately solve multiple pain points for those teams. Being able to bundle WhiteRabbitNeo, the leading uncensored, open source AI model for offensive cybersecurity, has allowed us to build an enticing product that helps them create Runbook automations and offload their time-consuming tasks. And finally, our management team has deep experience in cybersecurity and enterprise solutions.
What does 2025 hold for Kindo?
Our mission is to become the most trusted AI company on the planet. By doing so, we plan to grow into the de facto choice for AI security, when it comes to serving enterprise clients.
