Claude Cowork Transforms PM Workload

The system's core is a recurring 'weekly prep' task, which integrates several AI skills. This task aggregates data from various sources, including Notion, calendars, Slack, and Granola transcripts. By centralizing this information, the system automates the prioritization of inbox items and top-of-mind tasks.

This automation significantly reduces the user's mental load, ensuring that they are well-prepared for upcoming meetings and critical deadlines. The aggregated data provides a comprehensive overview, allowing for more informed decision-making and a clearer focus on strategic objectives.

Daniel Blum emphasized the foundational role of this task, stating, "So, it all starts with um one recurring task uh which is a composition of several skills called the weekly prep."

Product managers often face a significant disconnect between high-level strategic roadmaps and the urgent, chaotic demands of daily operations. The AI system addresses this by learning user behavior, such as recognizing that clearing Slack and email implies task completion.

This behavior-based learning allows the system to evolve and mirror the user's specific working style, thereby closing the gap between idealized workflows and reality. Daniel Blum noted, "I think one of the things any any PM can resonate with is the fact that there's two worlds, right? There's this beautiful clean organized world... and then there's reality where as we said this chaos of Slack and meetings and urgent priorities."

The system actively encourages continuous self-discovery, prompting users to audit their own internal business logic and identify areas where the AI's context is insufficient. This process addresses 'blind spots' where the system previously lacked necessary information.

By adapting to scattered workflows across various platforms like Slack, email, and Chrome tabs, the system continuously refines its understanding of the user's operational environment. Daniel Blum shared his experience: "I've had this experience with the system where I continuously like do discovery on myself and how I work and I continuously run into this places where the system is not good enough."

A powerful feature of the Claude Cowork system is its capacity for self-improvement, allowing agents to enhance their own performance. Claude actively scans for new context, milestones, or unfamiliar terms, transforming the AI from a static tool into a dynamic agent that maintains its own knowledge base.

The system prompts the user to verify if new information is worth saving, creating a feedback loop that continually refines its understanding and capabilities. This approach, as Daniel Blum noted, is crucial: "One of the things that I think is super powerful with co-work and and with these harnesses is that you can set them up for improving themselves."

In Daniel Blum's workflow, Notion primarily serves as a read-only hub where he tracks priorities and maintains focus. Claude, operating through Cowork, handles the actual management and execution of workflows based on the data in Notion.

Approximately 70-80% of his computer work is now performed through Cowork, highlighting the AI's central role in task execution. Daniel emphasized this division of labor, stating, "The main idea is that Notion is almost a read only for me... Claude manages on my behalf."

The initial setup of the system relied heavily on interactions through Chrome browser tools. However, there has been a significant transition towards using MCPs (Multi-Cloud Platform) and direct connectors for enhanced performance and integration.

Despite this shift, Chrome browser tools remain a critical fallback for situations where direct integrations are not yet available. Daniel Blum highlighted the importance of this flexibility: "The Chrome connector is a critical critical piece of the puzzle for when you don't have these connectors and integrations."

The initial return on investment (ROI) for new users is often low due to friction and the need to train the system with specific business logic. Users must push past initial skepticism to build long-term value, as constant context sharing is the core driver of the system's power.

Daniel Blum noted that the value compounds tremendously over time, encouraging persistence during the early phases: "The value tremendously compounds. It's just incredible the more you kind of power through these first few weeks."

Daniel Blum asserts that he can now complete in one day what previously took him an entire week, a testament to the system's efficiency. This drastic reduction in time spent on routine tasks has allowed him to engage in more deep work and research, shifting from intuition-based decisions to data-backed outputs. The system's ability to automate and streamline operations has transformed his daily routine.

He confidently stands behind this claim, highlighting the profound impact on his productivity. This enables him to delve into complex problems with greater focus and access to factual information, leading to more robust decision-making. The AI system essentially acts as a force multiplier, enhancing his capacity to handle a larger volume of work with improved quality and speed.

This transition marks a fundamental change in his work methodology, where the AI supports a more analytical and less reactive approach. The ability to free up significant time from repetitive tasks allows for greater innovation and strategic thinking. Daniel Blum summarized this transformation: "I'm really able to do in a day now what used to take me a week and I stand behind that."

When scaling the AI system for team adoption, Daniel Blum prioritized user experience (UX) and accessibility over complex technical features. He avoided overly technical developer tools like Cursor to ensure broader appeal and easier integration for all employees.

Success in scaling depends on simplicity and clear value propositions, rather than advanced technical capabilities. The 'Workstation' project, for instance, is designed to make AI onboarding accessible to every employee at Melio, simplifying what could otherwise be a daunting process. Daniel emphasized this approach: "At the end of the day if you make something accessible and simple and the value is clear then it gets adopted much better."

Claude now analyzes the difference between its generated drafts and the final versions sent by the user. This comparison helps the system learn from user edits and automatically improve its future writing style, reducing the manual gap between AI output and user intent.

By continuously refining its understanding of user preferences, Claude enhances its ability to produce more accurate and personalized communications. Daniel Blum explained the process: "Claude doesn't learn what the gap is between uh the draft that it wrote and what I actually sent. So, I have Claude uh when running this task looking for drafts that it it shared with me and I didn't respond and then looking to see if I sent them in a different way and learning from those draft."

All recurring tasks and skills within the system are designed with built-in mechanisms to collect feedback and track user friction. This automated feedback loop aggregates pain points weekly, which are then surfaced to suggest improvements without requiring proactive effort from the user.

This passive feedback collection ensures that the system continuously adapts to user needs and eliminates operational bottlenecks. Daniel Blum highlighted the efficiency of this method: "all of my skills and recurring tasks have uh lines in their in their files to collect feedback and friction for my interaction with them. So over time they gather this I don't need to do this proactively."

The 'Workstation' project at Melio has transformed from complex terminal-based setups into a streamlined, chat-based onboarding user experience. This system, initially designed for product managers, has expanded to all Melio employees, creating a comprehensive AI ecosystem.

The onboarding process for new employees now takes approximately 15 minutes. During this time, the Workstation automatically connects tools, maps management structures, and creates a personalized writing voice for each new user. This rapid deployment ensures that employees are quickly integrated into the AI-powered workflow.

Daniel Blum described the project: "what we call the workstation which is essentially this uh I would say operating system it started out for PMs andio um but it actually expanded to all Amelio workers and basically uh what it does is you know it it it creates an ecosystem."

Initial prototypes of Daniel Blum's AI system failed when introduced to colleagues due to excessive customization tailored to his individual workflow. This experience highlighted the necessity of designing a guided user experience (UX) to assist new users in adopting personalized AI tools effectively.

He developed a 'hand-in-hand' guided UX to make the system accessible and usable for a broader audience. Daniel emphasized the importance of this shift, stating, "it was completely adjusted to how I work and that's when it started when into this gem I built this UX that guides you hand in hand."

Currently, AI agents largely rely on an active computer and manual triggering, limiting their full potential for autonomous operation. The next critical frontier involves developing agents that can operate independently in the cloud, even when the user's computer is off.

While the current system efficiently tracks tasks, it lacks the full autonomy to execute complex resolutions without user interaction. Daniel Blum expressed this need: "the ability to operate in a in the cloud essentially in a way that's unrelated to when my computer is open and running."

The AI system is designed to grow and improve over time through a structured, recurring research and development cycle. Weekly updates and a dedicated 'improve' skill help filter relevant AI trends from mere LinkedIn hype, ensuring continuous refinement.

This systematic collaboration between the user and the AI refines the workflow, making the system stronger and more effective with each iteration. Daniel Blum highlighted this aspect: "the beauty about this is that if a system is built correctly and on the right pillars, it can grow and improve over time."

Users can experience frustration, dubbed 'Claude rage,' when AI models fail to perform as expected. A significant contributing factor is the limitation of speech-to-text technologies like Whisper, which often fail to capture the intensity or emotional nuance of a user's voice.

This gap in understanding can lead to misunderstandings and heightened user frustration, even when the AI is trying to process commands. Daniel Blum admitted, "I don't get road rage, but I get clawed rage. So, I think like I can I don't I don't know if I've cursed, but I I've definitely like raised my voice."