Knostic

𝗧𝗵𝗲 𝗗𝗼𝘄𝗻𝗳𝗮𝗹𝗹 𝗼𝗳 𝗖𝗼𝗱𝗲𝗿𝘀 𝗮𝗻𝗱 𝘁𝗵𝗲 𝗥𝗶𝘀𝗲 𝗼𝗳 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝘀 A year ago, I still had candidates write code as part of the hiring process. Today, that feels like 𝗮 𝘄𝗮𝘀𝘁𝗲 𝗼𝗳 𝘁𝗶𝗺𝗲 - both for the interviewer and the candidate. With AI handling much of the coding, the real skill to evaluate is not how well they can labor and write code, but how well they understand and assess code to create quality products. Studies suggest that heavy reliance on AI erodes critical thinking, which is 𝗲𝘅𝗮𝗰𝘁𝗹𝘆 what we need to evaluate when assessing candidates. Successful developers must have a solid grasp of software infrastructure, best practices, design patterns, and the internals of both programming languages and operating systems - this is what makes them software engineers. However, in the AI age, they must also ensure that their instructions to the AI coding agents don’t contain hidden faults and that the AI-generated code is robust, efficient, and secure. In an era dominated by AI, a developer’s problem-solving approach, and how that developer guides an AI coding assistant, is even more important than the developer’s ability to code. 𝗖𝗼𝗱𝗲 𝗥𝗲𝘃𝗶𝗲𝘄 𝗮𝘀 𝗮 𝗛𝗶𝗿𝗶𝗻𝗴 𝗧𝗲𝘀𝘁: 𝗔 𝗥𝗲𝘁𝘂𝗿𝗻 𝘁𝗼 𝗙𝘂𝗻𝗱𝗮𝗺𝗲𝗻𝘁𝗮𝗹𝘀 Instead of a traditional coding test, at Knostic, we now give candidates a premade code snippet. It’s a small, functional piece that integrates with common technologies (e.g., Redis, async IO) but contains about ten issues - bugs and potential bugs(performance, scale. race conditions), or just poor code quality and standards. Candidates receive a printed version and a limited amount of time to review it, marking problems, and suggesting improvements. Their performance is scored on: (1) The number of issues they identify (bugs, scalability risks, bad practices) (2) Their reasoning and technical depth (3) The quality of their suggested fixes I admit this is a work in progress, and I’m still refining the exercise. However, the last three hires all went through it, and it’s proving far more relevant than traditional coding tests in evaluating that critical thinking and problem-solving capability. Simple coding exercises leave us with less control over the kinds of issues we need them to recognize, like performance bottlenecks and race conditions. Some argue that AI agents are already advanced enough to avoid these mistakes. But that’s not entirely true - AI still makes errors. Even when it doesn’t, inexperienced developers might fail to provide AI with the full relevant context - leading to suboptimal solutions. Maybe in the future, AI will get so good that we’ll need to rethink hiring again. But AI is already changing how we evaluate human capabilities. #softwareengineering #AI #softwaredevelopment #softwarearchitecture #ArtificialIntelligence 𝘐𝘯 𝘵𝘩𝘦 𝘱𝘪𝘤𝘵𝘶𝘳𝘦: 𝘒𝘯𝘰𝘴𝘵𝘪𝘤 𝘴𝘸𝘢𝘨 𝘵𝘰𝘶𝘳𝘪𝘯𝘨 𝘕𝘰𝘳𝘵𝘩𝘦𝘳𝘯 𝘍𝘪𝘯𝘭𝘢𝘯𝘥.

Knostic has raised $11 million, bringing us to a total of 14.3. It's time to end LLM oversharing so enterprises can safely adopt technologies like Microsoft 365 Copilot and Glean, without worrying about data leaks. Beyond our first 9 deployments, this year has been incredible as far as validation goes. From winning RSAC's Launch Pad and Black Hat's Startup Spotlight competitions to being chosen as SVCI's bet in the space -- we are grateful. Thank you Fernando Martins 🇨🇵🇵🇹 and the Bright Pixel Capital crew for believing in us, DNX Ventures - US for always stepping in to help, Seedcamp for speaking truth - especially when it's uncomfortable, and SVCI - Silicon Valley CISO Investments for being one of the best vertical investor and support systems I've seen. Also joining us for the round are angel investors Kevin Patrick Mahaffey, who has doubled down on his investment yet again (now three times!), Gerhard Eschelbeck, Keith B., Zach Jakob, and Elad Cohen. We also appreciate all our investors so far, thank you for trusting us early on. Startups aren't easy, in fact - they're stress, embodied. But, nothing compares: the team, the mission, and getting things done for customers. I wouldn't have it any other way. Thank you Sounil Yu and to the whole team for joining me on this journey. https://lnkd.in/d3S3GffE #informationsecurity #cybersecurity #LLM #AI #artificialintelligence #copilot #oversharing #hacking #startup #fundraising

The promise of AI-powered productivity tools like Microsoft 365 Copilot is undeniable – but so are the risks. As organizations rush to enable these powerful capabilities, traditional data security approaches fall short in an AI-enabled world. The challenge isn't just about protecting data anymore; it's about securing your organization's knowledge. Join experts from Microsoft and Knostic for an expert-led discussion on preparing your organization for the safe adoption of Microsoft 365 Copilot. We'll explore how AI is fundamentally changing information flows within organizations, why traditional security controls may be insufficient, and what concrete steps you need to take before enabling these powerful tools. In this session, you'll learn: ✅ Why AI requires a fundamental shift from data security to knowledge security ✅ How to identify and protect your organization's most sensitive knowledge areas ✅ Critical security assessments needed before enabling Copilot ✅ Practical strategies for balancing productivity gains with security requirements ✅ Real-world examples of AI security challenges and how to address them Register below 👇

The promise of AI-powered productivity tools like Microsoft 365 Copilot is undeniable – but so are the risks. As organizations rush to enable these powerful capabilities, traditional data security approaches fall short in an AI-enabled world. The challenge isn't just about protecting data anymore; it's about securing your organization's knowledge. Join experts from Microsoft and Knostic for an expert-led discussion on preparing your organization for the safe adoption of Microsoft 365 Copilot. We'll explore how AI is fundamentally changing information flows within organizations, why traditional security controls may be insufficient, and what concrete steps you need to take before enabling these powerful tools. In this session, you'll learn: ✅ Why AI requires a fundamental shift from data security to knowledge security ✅ How to identify and protect your organization's most sensitive knowledge areas ✅ Critical security assessments needed before enabling Copilot ✅ Practical strategies for balancing productivity gains with security requirements ✅ Real-world examples of AI security challenges and how to address them Register below 👇