Two days of inspiring keynotes, hands-on workshops and new connections at the premier summit for Artificial Intelligence in Greece. March 13–14, 2025 · OTEAcademy, Athens.
As AI moves beyond traditional automation and generative capabilities, ensuring its reliability, transparency and ethical integrity becomes a pressing challenge. This session introduces AI-Assurance — a new paradigm for validating and governing Agentic AI systems that leverage Computer Use Automation (CUA), Chain of Thought (CoT) and GraphRAG to enhance model-driven decision-making and adaptive, agent-based execution. Covers large action models (LAM), AI-driven software agents across the SDLC, and governance frameworks such as ISO 42001, AI TRiSM, NIST and the EU AI Act.
Agentic systems are a new software paradigm — products that act autonomously, adapt to dynamic environments and make decisions in real time. Building production-grade agentic systems means solving for reliability, safety and performance. This session covers the core principles and best practices for trustworthy agentic solutions, a practical roadmap for upskilling teams and fostering AI literacy across engineering, product and design, plus the organizational shifts — restructured workflows, redefined metrics, an AI-first culture — needed to support agentic products at scale.
In a fast-evolving AI landscape, separating genuine advances from noise is crucial. The non-deterministic nature of LLM-driven applications makes common software-testing practices hard to apply and performance hard to quantify. This session digs into RAGAS — a framework designed to quantitatively measure the performance of LLM-driven applications — drawing on established software-testing practices and including a short live demo of how RAGAS translates LLM effectiveness into real-world impact.
A hands-on workshop for AI professionals, testers and developers eager to use cutting-edge AI to enhance testing strategies, automate processes and uncover bugs with precision. Blending theory with practical exercises, attendees generate comprehensive test cases, surface bugs in unexpected places, and master the art of prompt-engineering tuning — learning to craft prompts that guide AI through highly specialized testing tasks and integrate these tools into daily workflows. (Part I)
The second half of the hands-on session: deeper, guided exercises and live demonstrations that put AI-augmented testing into practice. Attendees continue building specialized prompts, automating complex scenarios, and integrating AI-driven tools to elevate the quality of their testing and the products they help create. (Part II)
Agentic and compound AI systems are reshaping LLM applications, driving real productivity gains in enterprise workflows — but adopting them at scale poses challenges in consistency, reliability and scalability. Central to solving these is memory. This talk examines the architecture of agentic systems and how working memory, data stores, profilers and toolboxes — across short-term and long-term memory — enable adaptive learning, entity profiling and personalized interactions in robust, scalable AI solutions.
Generative AI has unlocked remarkable capabilities in natural-language processing, but deploying LLMs efficiently in production reveals a landscape of challenges and technical debt. Ethically, LLMs face bias amplification, misinformation and privacy risks; societally, they reshape employment. Ahmed highlights the key challenges and technical debt of LLM deployment — the customization and sophisticated engineering not readily available in broad-use ML libraries or inference engines.
A tour of AI applications and why they matter — covering bias in AI (its types, impacts and mitigation), transparency through explainable AI and open-source models, and reliability through thorough testing and error handling. Addresses challenges like complexity and data privacy alongside solutions such as regulatory frameworks, with a practical case study on brAInbank, a NextGen Knowledge Management System powered by Agentic AI, plus a look at emerging trends and a Q&A.
How Generative AI is transforming the developer experience by speeding up tasks across the development lifecycle — researching and evaluating system designs, building secure and scalable applications, upgrading existing ones and optimizing performance. See how developers use GenAI to create specification documents, generate documentation, and understand complex code, streamlining the whole process.
The general availability of LLMs like ChatGPT and Gemini has made the disruptive importance of AI for testing impossible to ignore — bringing new technologies, techniques and required skillsets, and the ability for AI to partially take over testers' work. What does this mean for the role of testing over the next few years? What will it look like in ten years, will software testing as we know it survive, and will we still need humans? This talk reviews the current state, sketches future scenarios, and seeks plausible answers.
An exclusive hands-on workshop where you build your own Generative AI application using Retrieval-Augmented Generation. You'll create an AI chatbot that retrieves relevant information from contextualized data, gain practical experience storing vector embeddings in a vector database for efficient similarity search, and explore the RAG architecture — how combining retrieval mechanisms with generative models boosts the accuracy and relevance of AI responses. (Session I)
Prompt engineering is the emerging discipline of creating and refining prompts to get the best out of language models, diffusion models and other advanced technologies. The Crafting AI Prompts Framework splits into two parts — IPE and IVPE — helping you use Generative AI effectively for both written and visual content while promoting ethical, responsible use. (Part I)
Building on Part I, this session applies the IVPE side of the Crafting AI Prompts Framework — extending the method to visual content generation and putting the full framework into practice with worked examples for responsible, effective prompting. (Part II)
The continuation of the hands-on RAG build: deepen your chatbot, refine retrieval over your vector database, and extend the RAG architecture toward a practical, impactful GenAI solution you can keep developing after the summit. (Session II)
AI goes far beyond chatbots — its real impact lies in highly autonomous, agentic AI-driven operations where networks are managed with intents and AI translates high-level objectives into machine-executable actions. This session explores how Generative AI enables self-optimizing networks, how GenAI-driven data platforms simplify operations for engineers across industries, and how AI drives sustainability by optimizing data-center energy use and network efficiency to cut costs and carbon footprints.
Securing LLMs is critical for GenAI success, yet it remains one of its toughest challenges — from mitigating prompt injection and adversarial vulnerabilities to ensuring compliance. This session introduces an actionable checklist, grounded in real-world scenarios, to help attendees simplify and structure their approach to LLM security.
Agentic AI systems are emerging as powerful tools that anticipate needs and make autonomous decisions. This session examines the core principles of agentic AI and how it differs from traditional reactive systems, with real-world case studies across healthcare, finance and customer service. It weighs the benefits — improved efficiency and decision accuracy — against implementation challenges, integration with existing systems, ethical considerations and the need for human oversight, asking whether agentic AI is a genuine leap forward or an overhyped trend.
Catch every 2025 session on demand, then secure your seat for Disrupt AI Summit, May 14–15, 2026 at OTEAcademy, Athens.