Matt Kixmoeller’s Post

This is truly one of the most exciting times to see businesses evolve and adopt AI strategies to enable their teams and enhance customer experience. Glean's AI platform is uniquely positioned to enable companies, across the globe, to access and take action on data like never before, across every department and position. Search + Agentic Reasoning can and will change the landscape of every business. The early adopters will have a strategic advantage, in the same way that Glean has now.

The rise of agentic reasoning promises to transform how we work by creating AI systems that can autonomously pursue goals and complete complex tasks. Glean is at the forefront of this shift, pioneering a new agentic reasoning architecture that expands AI’s potential to get work done: resolve support tickets, help engineers debug code, and adapt tone of voice for corporate communications. With our new architecture, agents break down complex work into multi-step plans. Steps are executed by AI agents that are trained on their tasks and equipped with the tools to achieve their goals. Early research shows a 24% increase in relevance with our new agentic reasoning architecture. Here’s a preview of the new architecture: 🔷 Search: Evaluate the query and, using heuristics, determine whether it can be answered using search or agentic reasoning. 🔷 Reflect: Reflect on the initial search results, gauge confidence in the result, and decide whether to return a result or keep going down the agentic reasoning path. Search → fast and accurate answers Agentic reasoning → complex multi-step queries 🔷 Plan: Formulate the strategy, deeply understanding the goal and breaking down the steps to achieve it. Figure out the specialized sub-agents and tools to achieve each step of the work. 🔷 Execute: Sub-agents reason about the tools to use-search, data analysis, email, calendar, employee search, expert search, etc.- and how to stitch them together to achieve individual goals. 🔷 Respond: Respond in natural language via chat or by taking an action like creating a Jira ticket. Reimagining Work AI is an ongoing journey that builds on our foundational technologies. We began with search and advanced to RAG; now we’re progressing from RAG to agentic reasoning. We remain committed to pushing the boundaries of what AI can achieve in the workplace. This is the AI journey we envision for all our customers, where continuous innovation and practical application go hand in hand to transform the future of work. https://bit.ly/3ZdIWvg

🎯 Building multi-agent systems taught me the importance of simplicity, effective communication, and clear guidance. These systems are like assembling a high-performing team—each agent excels in a specific role, solving complex problems through collaboration. Key lessons: - Simplicity matters: Agents thrive with single, focused responsibilities. - Natural language rules: It fosters flexibility and precision. - Examples are essential: Clear blueprints reduce ambiguity and align outputs. Full article: https://lnkd.in/g8wuM-nu #AI #MultiAgentSystems #Innovation

How gen AI–enabled agents could work Agents can support high-complexity use cases across industries and business functions, particularly for workflows involving time-consuming tasks or requiring various specialized types of qualitative and quantitative analysis. Agents do this by recursively breaking down complex workflows and performing subtasks across specialized instructions and data sources to reach the desired goal. The process generally follows these four steps: User provides instruction: A user interacts with the AI system by giving a natural-language prompt, much like one would instruct a trusted employee. The system identifies the intended use case, asking the user for additional clarification when required. Agent system plans, allocates, and executes work: The agent system processes the prompt into a workflow, breaking it down into tasks and subtasks, which a manager subagent assigns to other specialized subagents. These subagents, equipped with necessary domain knowledge and tools, draw on prior “experiences” and codified domain expertise, coordinating with each other and using organizational data and systems to execute these assignments. Agent system iteratively improves output: Throughout the process, the agent may request additional user input to ensure accuracy and relevance. The process may conclude with the agent providing final output to the user, iterating on any feedback shared by the user. Agent executes action: The agent executes any necessary actions in the world to fully complete the user-requested task https://lnkd.in/eNr7GH7K

How gen AI–enabled agents could work Agents can support high-complexity use cases across industries and business functions, particularly for workflows involving time-consuming tasks or requiring various specialized types of qualitative and quantitative analysis. Agents do this by recursively breaking down complex workflows and performing subtasks across specialized instructions and data sources to reach the desired goal. The process generally follows these four steps: - User provides instruction: A user interacts with the AI system by giving a natural-language prompt, much like one would instruct a trusted employee. The system identifies the intended use case, asking the user for additional clarification when required. - Agent system plans, allocates, and executes work: The agent system processes the prompt into a workflow, breaking it down into tasks and subtasks, which a manager subagent assigns to other specialized subagents. These subagents, equipped with necessary domain knowledge and tools, draw on prior “experiences” and codified domain expertise, coordinating with each other and using organizational data and systems to execute these assignments. - Agent system iteratively improves output: Throughout the process, the agent may request additional user input to ensure accuracy and relevance. The process may conclude with the agent providing final output to the user, iterating on any feedback shared by the user. - Agent executes action: The agent executes any necessary actions in the world to fully complete the user-requested task. https://lnkd.in/eCbjpKY5

Why is the experience of using business systems so bad? I think the core issue lies in the lack of personalization, which leads to inefficiency, frustration, and disengagement. Business systems rely on rigid, static interfaces that don't adapt to the unique needs or preferences of individual users. This problem can be fixed by integrating generative AI as the front interface, ontologies as the schema, knowledge graphs as the database, and curated datasets through good old data engineering. Generative AI can transform how users interact with business systems by providing personalized, adaptive interfaces. Instead of forcing everyone into the same workflows, genAI dynamically adjusts to user input, context, and preferences, creating a more intuitive and user-friendly experience. This leads to more efficient systems that can recommend relevant actions, automate tasks, and ultimately enhance productivity. Ontologies serve as the schema that standardizes data across the organization, ensuring that concepts and relationships are consistently understood. This makes it easier to integrate data from different sources and maintain a flexible structure as the system grows. Paired with a knowledge graph as the database, which stores and organizes data by focusing on the relationships between entities, businesses can quickly derive insights and respond to complex user queries more effectively. Curated datasets built through data engineering serve as the foundation of this system, ensuring that the data feeding the AI and knowledge graph is clean, reliable, and relevant. With these high-quality datasets in place, the entire system operates intelligently, and enable the level of personalization that has been missing from traditional business systems. It’s time to get to practical AI for businesses and fix this lack of good and reliable user experience!

Imagine if AI could automate routine business tasks like drafting emails, generating campaign briefs, building web pages, researching competitors, analyzing data and summarising calls. Automating such repetitive tasks can free up an immense amount of valuable human time and effort for more complex and creative activities like business strategy and relationship building. A reasoning engine is an AI system that mimics human-like decision-making and problem-solving capabilities based on certain rules, data, and logic. Learn more about what it is and how it works.

Unleashing the Power of AI Agents: Reasoning, Planning, and Execution ... As AI continues to evolve, one area of intense research focuses on developing AI agent architectures that can enhance reasoning, planning, and tool execution capabilities. These advanced agents hold immense potential for tackling complex, real-world challenges across various industries. A takeaway of a recent paper "THE LANDSCAPE OF EMERGING AI AGENT ARCHITECTURES FOR REASONING, PLANNING, AND TOOL CALLING: A SURVEY" from Neudesic, an IBM Company and Microsoft. 👨💼 The Rise of AI Agents AI agents are language model-powered entities capable of planning and taking actions to execute goals over multiple iterations. They can be designed as single agents or as multi-agent systems, where multiple agents collaborate to solve a problem. Each agent is typically given a specific persona and access to a variety of tools to help accomplish their tasks. 👉 Single vs Multi-Agent: Choosing the Right Approach The research highlights the key differences between single and multi-agent architectures: "Single-Agent Architectures" excel when tasks are well-defined and do not require feedback from multiple personas. They are generally easier to implement and less prone to distractions. However, they may struggle with complex tasks that require diverse perspectives or parallel execution. "Multi-Agent Architectures" thrive in scenarios where collaboration, feedback, and division of labor are crucial. By employing specialized agents for different tasks and facilitating dynamic team structures, these architectures can achieve greater accuracy and efficiency in solving intricate problems. 👉 Real-World Impact: Enhancing Problem-Solving Prowess Consider a personal working agent that can read your emails, Slack messages, and meeting requests. It could analyze all your incoming information, create summaries, identify key actions and decisions, and provide the full context - all while you stay focused on your core work. This agent's advanced reasoning, planning, and tool execution capabilities could drastically boost productivity for knowledge workers. 👉 Optimizing Performance: Techniques and Best Practices The research identifies several techniques and best practices for improving agent performance, such as well-defined prompts, dynamic team structures, and effective communication between team members. Implementing these strategies can lead to more efficient and accurate goal execution. 👉 Challenges and Opportunities Ahead While AI agents have made remarkable strides, current limitations include issues with performance evaluation and bias. Continued research is needed to address these challenges and drive innovation in the field. How do you envision the use of AI agents in your industry or field of expertise? Share your thoughts, experiences, and ideas in the comments below.

Play around with about 20 of our new WatsonX demos and in some add your own content to try it properly https://lnkd.in/gWivY_rM I particularly liked these two: Sales Conversation Intelligence - Where it will analyze your sales calls for you and summarise it with next steps https://lnkd.in/eWfzsMnV Automate Job Requisition - easily create standardized job req's with Watson https://lnkd.in/e3nPUrky

This post is to share inspiration... because everyone is seeking it with GenAI. The second take away is that if you try replicate it with OpenAI you will not be able to differentiate from your competitors and you will take too many risks... so take a look to IBM watsonx... and see the difference... #letscreate #IBM #watsonx #genai #aigovernance