The capability gap we address is the inability of incident command centers to safely leverage Generative AI. During emergencies, responders generate massive streams of sensitive data. Standard cloud-based AI cannot be used due to strict data sovereignty laws (GDPR), the EU AI Act, and the risk of network latency/outages at the tactical edge. Our solution is the KennisAI Platform, developed by 8peaks. KennisAI provides Sovereign, Zero-Trust AI infrastructure that operates entirely on-premise or at the edge. By deploying optimized, open-weights LLMs (such as Llama 3 or Mistral) locally, KennisAI enables autonomous AI agents to securely ingest, integrate, and analyze highly sensitive data from multiple disparate sources (e.g., citizen chat logs, sensor data, and internal databases) in real-time. Crucially, our proprietary Model Context Protocol (MCP) Gateway acts as a Zero-Trust perimeter, physically preventing unauthorized data egress. This allows emergency management agencies to deploy "Agentic Bots" (e.g., for the autonomous triage of emergency reports or drafting incident workflows) that natively comply with the EU AI Act and NIS2 cybersecurity directives. KennisAI empowers commanders with a low-code canvas to visually orchestrate and audit AI workflows, ensuring "Glass-Box" explainability and human-in-the-loop oversight during critical operations. TRL Specification: The KennisAI core platform is currently at TRL 7 (System prototype demonstration in an operational environment), having been validated for secure, on-premise edge deployments.
Contact
When a crisis hits, first responders and command centers are immediately flooded with data—citizen reports, sensor feeds, and operational logs. Generative AI could be a massive game-changer here, helping commanders instantly triage information and make life-saving decisions. But right now, emergency services are practically locked out of the AI revolution. Why? Because you simply cannot send highly sensitive, classified, or personal emergency data to public cloud AI models without violating strict privacy laws (like GDPR and the EU AI Act) or risking a severe data breach. Plus, in a real disaster, internet connectivity to the cloud is often the first thing to drop. Our ambition with the KennisAI platform is to completely remove this bottleneck. The main objective of this proposal is to deploy and validate a "Sovereign Edge AI" infrastructure specifically built for critical emergency operations. We want to give incident commanders the power of advanced AI right there on the ground, running entirely on local servers or edge devices. The relevance of this project is immediate. By cutting the cord to the public cloud, we ensure that highly sensitive emergency data never leaves the secure tactical perimeter. Our goal is to prove that emergency response teams can use autonomous AI agents to filter noise, spot threats, and generate real-time actionable intelligence, all while staying 100% compliant with European data sovereignty laws and keeping operations running even if the broader internet goes dark.
Most AI solutions on the market today rely on a fundamental assumption: that you can easily send your data to a massive, centralized cloud to be processed. For public safety and civil security, that approach is a dead end. Our innovation flips this model entirely by bringing the AI directly to the data. The strength of our approach rests on two major technical breakthroughs within the KennisAI platform: First is our Zero-Trust MCP (Model Context Protocol) Gateway. Think of it as an ironclad bouncer for the AI. Even when an AI model is running locally on your own servers, you still need to ensure it only accesses the data it’s strictly allowed to see. The MCP Gateway physically enforces these rules, completely preventing unauthorized data extraction. This means an AI agent can read an incoming citizen distress message, cross-reference it with a secure internal police database, and alert a commander—all without the AI ever having the technical ability to leak that database to the outside world. It is "Zero-Trust" built natively into the AI's nervous system. Second is our "Glass-Box" orchestration canvas. Usually, building and managing AI agents requires a team of highly specialized software developers. We’ve built a visual, low-code interface that lets non-technical domain experts—like a seasoned incident commander or dispatcher—actually design, audit, and control the AI workflows themselves. We replace the standard "black-box" AI with an entirely transparent system. If the AI flags a critical threat, the commander can see exactly which data points led to that conclusion and can intervene at any moment. In short, our innovation isn't just about making AI smarter; it is about making it legally compliant, transparent, and operational for the people who need it most during a crisis.
When a crisis occurs, operational capacity is severely bottlenecked by data overload. First responders and command centers receive massive influxes of unstructured data from 112 calls, citizen chat logs, IoT sensors, and field reports. While Generative AI is perfectly suited to instantly triage this information, current public cloud AI solutions resolve nothing for emergency services because they cannot be safely or legally used. Sending classified operational data to third-party clouds violates GDPR, the EU AI Act, and NIS2, and it relies on internet connectivity that is often compromised during a disaster. The KennisAI platform resolves this by decentralizing AI. We provide a Sovereign, Edge AI infrastructure that allows command centers to run advanced, open-weights LLMs entirely on-premise or on localized edge servers. Specifically, our solution resolves the "security versus innovation" dilemma. Our proprietary Zero-Trust MCP (Model Context Protocol) Gateway acts as a physical barrier. It allows the AI to ingest and analyze sensitive internal databases to generate actionable intelligence, but it physically blocks the AI from ever transmitting that data outside the secure perimeter. By resolving this critical data-egress vulnerability, KennisAI enables emergency operations to instantly triage threats, track responder proximity to hazards, and draft automated incident workflows without ever compromising operational security or citizen privacy.
8peaks is an Antwerp-based DeepTech scale-up company. Our team comprises Senior Solution Architects, Product Managers, and AI Engineers with deep expertise in Zero-Trust cybersecurity, Edge AI orchestration, and Large Language Models. We specialize in translating complex regulatory frameworks (EU AI Act, NIS2) into secure, deployable software. We bring strong technical leadership to consortia, capable of managing complex Work Packages for high-risk AI infrastructure deployments.
8peaks is an Antwerp-based DeepTech scale-up company. Our team comprises Senior Solution Architects, Product Managers, and AI Engineers with deep expertise in Zero-Trust cybersecurity, Edge AI orchestration, and Large Language Models. We specialize in translating complex regulatory frameworks (EU AI Act, NIS2) into secure, deployable software. We bring strong technical leadership to consortia, capable of managing complex Work Packages for high-risk AI infrastructure deployments.