Uploaded byDenis Gagné
Knowledge Worker Copilot Conversational Understanding
Our presentation will update developments from Decision Camp presentations over the past four years. The Knowledge Worker Copilot is designed to navigate evolving cases governed by conversations and documents arising from legal contracts and regulatory regimes. It maintains a comprehensive memory, or context datastore, of messages, events, documents, and stakeholder roles. Initially conceived in 2023 as an intelligent assistant for knowledge workers, it bridges workflow gaps involving emails, documents, and meetings. Leveraging symbolic AI and Natural Language Understanding extracts key events and entities from unstructured content. In 2024, the platform was upgraded with large language models (LLMs) to enhance its ability to manage workflows and extract contractual entities. By 2025, the KW Copilot will have become a platform that automates complex conversational workflows, interprets contracts, and ensures compliance, allowing knowledge workers to focus on higher-value tasks and improving organizational efficiency and performance.
More Related Content
Similar to Knowledge Worker Copilot Conversational Understanding
![]project-open[ Workflow Developer Tutorial Part 1](https://cdn.slidesharecdn.com/ss_thumbnails/po-workflow-tutorial-1-overview-100603-101013103734-phpapp02-thumbnail.jpg?width=640&height=640&fit=bounds)
Knowledge Worker Copilot Conversational Understanding
- 1. Conversational Undestanding with BPM+ Powered by TomDeBevoise with Denis Gange
- 2. Powered by Outline Set theContext Copilot Scenario and Demonstration Explanation Natural Language Understanding Conclusion
- 3. Powered by AI andBPM+ Context
- 4. Powered by Knowledge Work KnowledgeWork represents an ever-increasing portion of modern business
- 5. Powered by What: Copilot “Cases” NotProcesses in the Conventional Sense Contracts of any sort (e.g. Real Estate Sale Contract, Maintenance Management) Compliance with Rules and Regulations Standard Operating Procedures, Lab testing, maintenance tickets
- 6. Powered by KW CoPilotin Action (Notional)
- 7. Powered by Example Cases •Fortune 50 media, which sets advertising prices based on the input of sales and other subject matter experts • A restaurant supply vendor that uses a voluminous catalog to customize a restaurant start-up or upgrade • A metal building manufacturer/installer • Contracts & ISO 9001
- 8. Powered by AI andBPM+ Demo Scenario
- 9. Powered by Demo Context •We will now walk through a relatable Real Estate example • Keep in mind, this is a pattern we can apply to any contract or regulatory-driven workflow.
- 10. Powered by Introducing theStakeholders of our Demo • Pat : Human Real Estate Transaction Manager • General Inspector: Human Real Estate inspector • Angie: Human Real Estate Broker • KW CoPilot: Intelligent Digital Assistant
- 11. Powered by Introducing theScenario of our Demo • Topic of Demo: Real Estate Sales Contract • A new contract is submitted by Pat to the KW CoPilot who initiates the case for that particular home sale • The KW CoPilot manages all terms and conditions of the contract, the schedule from the contract, the deliverables from the contract, and all interactions between the various stakeholders related to this home sale • This scenario will involve the need for a Wood Destroying Inspection (WDI)
- 12. Powered by AI andBPM+ Demo
- 13. Powered by AI andBPM+ Explanation
- 14. Powered by Initiating anevolving dynamic case • The human knowledge worker submits the contract to the KW CoPilot, creating a new case workflow • A dashboard displays the terms and conditions, deliverables, and important files related to the case
- 15. E-mails relevant tothe workflow are labeled in the knowledge worker's inbox. Submitted artifact are stored in a repository such as Google Cloud Storage. The words of every PDF are extracted and exposed with AI/OCR Copilot’s productivity features in the office suite: important events are written on the knowledge workers' calendar.
- 16. Deliverables can includerequests, reviews, and approval conversations Responsible parties can schedule deliverables The copilot uses a process and AI to understand the conversations and the attachment artifacts. The copilot contextualizes, monitors, and understands the conversations: The deliverables and messages are summarized and forwarded to the recipients, and the review conversations, if any, are started.
- 17. The copilot usesa process and AI to understand the conversations and the attachment artifacts. Participants can go off topic or request more information Participants can message off the thread. The Copilot understands complicated conversations The understanding occurs across a broad spectrum of scenarios
- 18. Powered by AI andBPM+ NLU
- 19. Powered by Natural LanguageUnderstanding Evaluating free text to extract information that can be used in a digital context It includes tasks like entity recognition, intent classification, sentiment analysis, and syntactic parsing to extract structured meaning Chatbots, voice assistants, legal document analyzers, and intelligent workflow systems
- 20. Powered by Natural LanguageUnderstanding Conversations Understanding Conversations occur in a context over an extended period The context becomes a channel in a process(Symbolic AI) The sub-symbolic feeds the channel
- 21. Powered by Symbolic andGenAI Combines rule-based, structured AI with unstructured, sub-symbolic approaches Symbolic side manipulates concepts at a higher level of abstraction Sub-symbolic includes Gen-AI & ML BPM+ Models are deterministic visual models readable by both Humans and Machines • CMMN: Case Management Model and Notation • DMN: Decision Model Notation • BPMN: Business Process Model and Notation
- 22. Powered by Neuro Symbolicand GenAI The processes and decisions for conversational understanding utilize various types of prompts The KW Case is tracked within a core process that abstracts the source document A messaging backbone identifies the associated case
- 23. Powered by AI andBPM+ Conclusion
- 24. Powered by KW CoPilot •KW Copilot = Augmented Intelligence • The KW Copilot is an Intelligent Digital Assistant that helps prepare and organize knowledge work, such as events, communications, documentation, and data entry, related to cases, while keeping the human knowledge worker in control. • Human Knowledge Worker = Responsible and Accountable
- 25. Powered by What: KW Copilotkeeps a living memory of every evolving case • The KW CoPilot captures events, documents, and stakeholder roles • The KW CoPilot orchestrates BPM+ case workflows and Generative AI reasoning to automate: • follow-ups • surface real-time insights • keep legal, compliance • The KW CoPilot keeps the case team of stakeholders perfectly aligned
- 26. Powered by The KWCoPilot contextualizes, monitors, and understands the conversations • Deliverables can include requests, reviews, and approval conversations • Responsible parties can schedule deliverables • A review and approval, if needed, is requested
- 27. Powered by Summary ofthe Knowledge Worker CoPilot Capabilities and Benefits • Transforms Contracts into Living Case Workflows • Intelligent Case Automation • Deep Contextual Understanding • Enhanced Productivity & Real-time Visibility • Comprehensive Artifact Management • Improves Compliance & Agility
- 28. Powered by AI andBPM+ Q&A
- 29. Powered by AI andBPM+ • For more info, visit • www.kw-copilot.ai
Editor's Notes
- #1 Good morning/afternoon, everyone. Thank you for joining us today. I'm thrilled to introduce you to a truly groundbreaking solution for modern enterprise challenges. Today, we'll be exploring The Knowledge Worker Copilot, the first transformative copilot designed explicitly for your message-based organizational workflows. Unlike the other copilots which are focused on what the individual does this copilot it's designed for knowledge workers And the work they do across the organization Please join me as we explore how KWCP is poised to transform your critical operations.
- #9 We will now walk through a relatable Real Estate example. This scenario is designed to illustrate the practical application of KW Copilot in a common business context where contracts and numerous communications are paramount. As we go through this demo, please keep in mind that this is a pattern. The underlying principles and the way KW Copilot operates can be readily applied to any contract or regulatory-driven workflow across your organization – from vendor agreements and service contracts to compliance protocols and internal operating procedures. The flexibility is inherent. Let's dive into the demonstration.
- #14 So, how does this powerful system actually function from a user's perspective? Let's walk through the initiation of an evolving, dynamic case. It begins very simply: The human knowledge worker submits the contract to the KW Copilot. This action, whether through a direct upload, integration with an existing document system, or even an email attachment, creates a new, uniquely identifiable case workflow within the system. Immediately upon submission, the KW Copilot gets to work. It intelligently processes the contract, and this is reflected in your central dashboard. This dashboard then displays, in a clear and organized manner: The key terms and conditions extracted from the document. The defined deliverables and their associated milestones. And all important related files, whether they are attachments from the original submission or subsequent documents. This provides an immediate, structured foundation for managing the entire lifecycle of the contract, setting the stage for the dynamic evolution we'll explore next.
- #15 As we mentioned in the Mackenzie study, knowledge workers frequently search for relevant artifacts and files and try to manage their e-mail inboxes. The knowledge worker copilot has been specifically designed to provide productivity features that enhance the visibility of the emails associated with the workflow Workflow instances have a separate calendar.
- #19 These processes are not ad hoc random sets of steps they have very specific guidelines and they achieve very specific goals regularly. The random or unpredictable is the conversation or the the blurry space between the request approvals and reviews. Before and only recently have we had a systematic way to attack that blurry space.
- #20 These processes are not ad hoc random sets of steps they have very specific guidelines and they achieve very specific goals on a regular basis. The random or unpredictable is the conversation or the blurry space between the request approvals and reviews. Before and only recently have we had a systematic way to attack that blurry space.
- #21 We’ve been leveraging a neuro-symbolic architecture. In brief, this combines rule-based, structured AI with unstructured, sub-symbolic approaches (e.g., neural networks). The symbolic side manipulates concepts at a higher level of abstraction; the sub-symbolic side focuses on recognition. The key goals of a neuro-symbolic approach are explainability—understanding what systems are doing and how they reach conclusions—bridging perception and cognition, and reducing reliance on massive datasets. Purely neural approaches (predictive or generative) can require very large datasets to achieve good performance. As most people know, modern LLMs are trained on enormous corpora—far beyond what most organizations can assemble or curate.
- #22 Definitions
- #24 So, how do we address this intricate landscape of dynamic workflows, cognitive load, and compliance demands? Our answer is the Knowledge Worker Copilot – KW Copilot. We view KW Copilot not as a replacement for human expertise, but as Augmented Intelligence. It's designed to enhance, not automate away, the critical judgment and experience that define knowledge work. At its heart, the KW Copilot is an Intelligent Digital Assistant. Its primary function is to help prepare and organize the essential elements of knowledge work. This includes managing events, streamlining communications, organizing documentation, and simplifying data entry – all directly related to the "Cases" we discussed. Crucially, it does this while keeping the human knowledge worker firmly in control. We believe in a partnership model: the Human Knowledge Worker remains responsible and accountable, leveraging their expertise. KW Copilot provides the intelligent support to execute those decisions more efficiently and effectively, freeing them from the unproductive tasks and complexity we just outlined. It’s about empowering them, not replacing them.
- #25 A core differentiator for the Knowledge Worker Copilot is its ability to maintain a living memory of every evolving "Case." Think of it as a dynamic, intelligent dossier for each critical undertaking. KW Copilot actively captures essential elements: All relevant events as they occur. Crucial documents and their versions. The specific stakeholder roles involved. This captured intelligence then allows KW Copilot to orchestrate a powerful synergy between our BPM+ case workflows and Generative AI reasoning. This orchestrated intelligence automates key functions for you: It proactively manages follow-ups, ensuring no critical action is missed. It surfaces real-time insights derived from the case data, providing critical awareness. And it helps ensure that both legal and compliance requirements are met throughout the lifecycle of the case. Ultimately, the KW Copilot's "living memory" ensures that your case team of stakeholders remains perfectly aligned, operating with a shared, up-to-date understanding of all activities, contexts, and obligations. This dramatically reduces miscommunication and operational friction.
- #26 Now, let's talk about the real differentiator: the KW Copilot's ability to go beyond simple keyword detection and truly contextualize, monitor, and understand the conversations that drive these cases forward. We know that deliverables within a case aren't just static items; they often involve dynamic processes. These can include explicit requests for action, rigorous review cycles, and critical approval conversations between stakeholders. The system is designed to recognize these distinct phases. For instance, responsible parties can schedule deliverables, and the KW Copilot understands the implications of those scheduled dates. Furthermore, when a review or approval is required, the system can interpret the communications related to that step. It understands the intent behind a request for review, the content of feedback provided, and the significance of a final approval, all within the context of the specific case and its governing contract or project. This deep understanding is what enables the system to provide truly actionable intelligence and automate appropriate follow-ups.
- #27 We've now seen the Knowledge Worker Copilot in action through our Real Estate example, illustrating how it tackles complex, message-based workflows. Let me quickly recap the core capabilities and the transformative benefits it delivers: Transforms Contracts into Living Case Workflows: KW Copilot takes static documents and breathes life into them, creating dynamic, actively managed processes. Intelligent Case Automation: It automates repetitive tasks, from follow-ups to communication drafting, leveraging AI and defined models. Deep Contextual Understanding: By interpreting conversations and documents, it grasps the nuanced meaning behind the interactions, not just the words. Enhanced Productivity & Real-time Visibility: It cuts down unproductive activities and provides clear, immediate insights into case status and compliance. Comprehensive Artifact Management: All relevant documents and communications are securely captured, organized, and made searchable for compliance and audit purposes. Improves Compliance & Agility: It ensures adherence to regulations and policies while providing the responsiveness needed to adapt to changing business environments. In essence, KW Copilot empowers your knowledge workers by removing friction, enhancing their understanding, and automating the administrative overhead, allowing them to focus on the high-value judgment and experience that drives your business forward.
- #30 Data Structuring: Converting free text into structured data (e.g., JSON) for downstream systems. Topic Identification (Semantic Lifting): Extracting higher-level topics—e.g., contact management, calendaring, email messaging—to give better context. Intent Detection: Understanding user goals behind the words. “Can you help me schedule a meeting and send a summary to my team?” involves multiple intents (scheduling and summarizing) that LLMs handle more reliably than rule-only methods. Text Generation: Well-known—moving beyond rigid templates to more dynamic, context-sensitive messaging.
- #31 An the copilot can improve the performance of accountable managers’ portfolios and businesses by delivering top-line visibility into how well the organization complies with contracted conditions. It highlights on-time performance, identifies rework or deficiencies, uncovers blockages caused by delays, and tracks adherence to mandated procedures and documentation. Beyond compliance, it reveals the tone and tenor of ongoing work, offering managers deeper insights into organizational health and conformance. This level of visibility equips leaders to address systemic issues proactively, strengthen performance, and drive consistent business results.
- #33 A key aspect of the Knowledge Worker Copilot's effectiveness is its deep, seamless integration with the productivity tools you already use every day. We're not asking users to adopt an entirely new ecosystem; we're enhancing your existing one. Here's how KW Copilot leverages your productivity office suites: Email Management: All emails deemed relevant to a specific case are intelligently labeled and organized directly within the inbox of the human knowledge worker. This provides immediate visibility to critical communications without disrupting their existing workflow. Document Repository: Submitted artifacts, whether original documents or subsequent files, are securely stored in a designated repository, often leveraging your existing cloud storage like Google Cloud Storage or OneDrive. This ensures centralized, secure, and accessible documentation for each case. Intelligent Document Processing: The content of every PDF attachment is automatically extracted and made searchable with advanced Intelligent Document Processing. This includes recognizing text, signatures, and even handwritten notes, making all documentation fully actionable. Calendar Integration: Important case-related events and deadlines are automatically written onto the calendar of the human knowledge worker. This ensures that upcoming milestones and critical dates are proactively surfaced, integrating directly into their daily schedule. By working within these familiar environments, KW Copilot minimizes user friction and maximizes adoption, ensuring that its powerful capabilities are easily leveraged by everyone. As we mentioned in the Mackenzie study, knowledge workers frequently search for relevant artifacts and files and try to manage their e-mail inboxes. The knowledge worker copilot has been specifically designed to provide productivity features that enhance the visibility of the emails associated with the workflow Workflow instances have a separate calendar.
- #34 A critical function of the Knowledge Worker Copilot is to actively monitor and assist with the execution of deliverables, moving beyond just tracking to enabling action. When the predefined conditions for a specific deliverable are met, KW Copilot can initiate the next steps. For instance, it can automatically create a draft email for review, or in appropriate scenarios, send it directly to the relevant parties. The email text itself is intelligently generated, dynamically populated using the extracted contract and case variables. This ensures the communication is precise, relevant, and consistent with the specific details of the agreement. Furthermore, in a sophisticated chain of conversational events, KW Copilot can call out to other tools via APIs. This allows for deeper automation, whether it’s triggering an action in a CRM, updating an ERP system, or invoking specialized business logic, thereby extending its reach and automating more complex downstream processes. This capability is key to transforming a notification into a fully executed, integrated business action.
- #36 To conclude, the Knowledge Worker Copilot is more than just a tool; it's an essential partner for your most valuable asset: your knowledge workers. It ensures no obligation is forgotten, providing the critical diligence required in today's complex business environment. It finds the patterns you would otherwise miss, uncovering insights hidden within unstructured data and conversations, driving smarter decisions. And it drafts those critical follow-up emails so you don’t have to, freeing up valuable time for what truly matters – your expertise, your judgment, and your strategic focus. Thank you for your time today. We’re excited about the potential for KW Copilot to transform your workflows and drive significant value. We invite you to connect with us for a more personalized discussion on how we can address your specific challenges.
- #37 To improve performance understanding, the Copilot monitors requests, contract events, and conversation sentiment, ensuring timely follow-ups while providing new performance metrics like compliance with deadlines and activity cadence. Conformance: conditions, content and timeliness In contracts and SOP-driven work, copilot boosts productivity by automating routine tasks such as starting project phases, drafting correspondence, managing approvals, calendars, and deliverables. By understanding context and workload, it prioritizes tasks and deliverables, enabling knowledge workers and responsible roles to perform with greater efficiency and clarity.
- #38 Let’s delve a bit deeper into the operational mechanics of how the KW Copilot actively supports the management of these "Cases." Fundamentally, a case originates from a specific instance – be it a particular contract, a project plan, or a compliance requirement. It's the singular instantiation of these broader frameworks. The critical factors that need to be monitored – such as deadlines, key clauses, specific deliverables, or compliance checkpoints – are extracted directly from that source document, leveraging our intelligent Natural Language Understanding. The primary channels for communication and activity within each case are typically email and documents, as we've discussed. This is where the KW Copilot’s active monitoring comes into play. It meticulously monitors the associated schedule, understanding upcoming dates and events. Simultaneously, it reads and analyzes the relevant email correspondence and documents associated with that specific case. By understanding this flow of information, it can then proactively identify what needs attention, what has been completed, and what requires further action.