The Planning-Doing Spectrum - Finding Balance in Knowledge Work

I had a fascinating conversation with Jan Nitschke and Niko Korvenlaita recently about the tension between planning and doing in knowledge work. It's a topic that resonates with many of us who struggle to balance thinking and execution in our increasingly complex digital environments.

This tension isn't just a personal quirk – it's fundamental to how modern knowledge work functions, and it has profound implications for how we build tools to support knowledge workers.

P.S. Shoutout to Jan who runs 11OWLS - a boutique consultancy and implementation partner for data analytics and data architecture projects.

They transform data into meaningful and cohesive insights, and enable decision makers to navigate through the increasing complexities of a growing organization.

Ok, back to musing about planning, doing, and notetaking.

The Planning-Doing Spectrum

When we think about knowledge work, there's rarely a clean separation between planning and execution. Instead, we exist on a spectrum:

Planning Mode ⟷ Doing Mode

During our conversation, Jan observed, "For many people (me included) the lines between the two are very blurry." I couldn't agree more. We don't just move from planning to doing in a linear fashion. We oscillate between them constantly, sometimes multiple times in a minute.

Jan described this perfectly as "micro planning" – those brief moments when you pause to think, "Wait, what am I trying to accomplish here? Where do I start?" before diving back into execution. I've found that these micro-planning moments are often the most crucial inflection points in our work, where the biggest insights emerge.

As I explained to Jan, "I don't think any of the people meant planning mode as a separate sit-down session, like a company steering day, but more like in the moment before starting to do something is to take a quick pause and briefly think through wtf am I trying to do, what ideas do I have, where do I start."

Different Notes for Different Modes

One of the most interesting realizations from our discussion was how note-taking serves entirely different purposes depending on where you are on the planning-doing spectrum.

In planning mode, notes are exploratory – "I am just writing out stuff that I'm thinking, seeing what it looks like."

In doing mode, notes become urgent captures – "Shit, write this down somewhere quick before I forget."

This distinction matters because it suggests we need different interfaces and interaction models for these different modes. The carefully organized, hierarchical note-taking systems we've built in the past might work well for planning mode, but they often fail us in doing mode, where speed and low friction are essential.

Why We Struggle to Document Our Thinking

One reason many of us struggle with note-taking is that it feels like an interruption to our flow. As Jan explained:

"I think especially when I do like coding projects, the state of the code is kind of my notes... I look at the code and then I'm like, I know where I left it."

Niko added an insightful perspective on this, noting that when you're taking notes, "you're doing the processing in your head and you're doing the kind of looking like yourself that hey, did we have something for this? I remember that there was some conversation about this and then you're kind of going and doing that search for you."

This gets to the heart of why documentation is challenging. In doing mode, we're focused on progress, not documentation. The thought process feels obvious in the moment, so why take notes? But this creates two problems:

Lost context for ourselves when we return to the work later
Limited collaboration because others can't follow our reasoning

The Car Analogy: Where We're Headed with AI Agents

Jan offered a brilliant analogy for how AI agents are changing knowledge work:

"I think agents and knowledge work is a little like cars. Back in the days (like 1 year ago haha) we had to walk from place to place to get knowledge. Now, with agents, it's like having a car. We can query that knowledge much faster because we have a more clever interface to navigate it. But today we still need to push the pedals, turn the steering wheel, tell the car where to go."

This perfectly captures our current moment with AI. We have powerful tools that accelerate our ability to find and process information, but they still require significant human guidance.

I've observed this in my own work with various AI tools. They've dramatically accelerated certain tasks, but the need for human intervention remains crucial, especially at decision points. This is why I believe the future isn't about replacing human thought but augmenting it at precisely those micro-planning moments when we're deciding which direction to take.

The Agent-Human Spectrum

Beyond the planning-doing spectrum, our conversation revealed another important dimension in knowledge work:

Human ⟷ Agent

Some tasks clearly belong to humans – like consciously noting down an important insight during a meeting. Others can be delegated to agents – like organizing notes or suggesting connections between ideas.

But the most interesting area is the middle ground – collaborative tasks that require both human judgment and AI capabilities. These are tasks like "discussing and figuring out next steps" where the combination of human and machine intelligence creates something greater than either alone.

Niko highlighted this collaborative potential when he described how "the note taking sits to both directions." He explained that while writing notes or documentation, "you're searching for pieces of information connecting that into a new document and then that creates new information." This bidirectional flow is exactly where the human-agent partnership becomes most powerful.

The Cold Start Problem and Conversational Documentation

One of the most insightful observations in our discussion was about the "cold start problem" in note-taking. Traditional note-taking apps only contain what you explicitly put into them, creating a significant barrier to entry.

"I think the magic here," Jan explained, "is that when you're having a conversation [with an agent], the user is going to be much more willing to reveal the information that's now stuck in his head."

This represents a paradigm shift in how we think about documentation. Instead of taking notes as a separate activity, the conversation itself becomes the documentation – capturing our thought process organically as we work with our tools.

I see this as potentially revolutionary. My own experience has shown that the best documentation often emerges from conversations – whether in code reviews, Slack threads, or video calls. The challenge has always been capturing and organizing that knowledge. AI assistants that can participate in and document these conversations could solve a problem that has plagued knowledge management for decades.

The Agent Ecosystem: Beyond Single-Agent Solutions

Perhaps the most exciting vision from our conversation was about the potential for an ecosystem of specialized agents working together.

Jan described a scenario where different agents analyze different data sources: "Picture a system of agents where one is able to look into Confluence, another one can analyze a database, and a third one will look at dashboards... And then you ask a question like, 'How does our billing work?'"

Niko expanded on this vision, explaining how reconfigured could serve as "the common center of things." He described a workflow where "you have the acting agent that is specified for you on that particular use case and you basically come and reconfigure and say, hey, do you want me to call this quality check or integrity check tool? And you say, okay, go on."

This multi-agent approach recognizes a fundamental truth about knowledge work: it spans multiple tools, systems, and domains. No single agent can effectively cover all these areas, but a coordinated ecosystem might.

In my own work, I've seen how different specialists contribute to complex problems. A data engineer, a UX designer, and a product manager bring different perspectives to the same challenge. What if our AI systems could mirror this collaborative approach, with specialized agents working together under human guidance?

Making Note-Taking Natural and Valuable

The challenge for modern knowledge tools isn't simply to provide a place for notes – it's to make the process of capturing context natural and immediately valuable.

Jan suggested an approach where agents proactively find information and identify gaps, giving users a starting point rather than a blank page: "Maybe this is a more proactive or more rewarding approach because you don't start with zero."

This resonates strongly with my experience. The hardest part of documentation is often the blank page. By providing structured starting points based on existing information, we could dramatically lower the activation energy required to add valuable context.

I've experimented with this approach in my own work, using templates and pre-filled structures to make documentation more approachable. The results have been promising, but AI agents could take this to a new level by dynamically generating these starting points based on the specific context of the work.

A New Paradigm for Organizational Knowledge

One of the most powerful insights from our conversation was Niko's description of how individual notes connect to form a greater whole: "It's not just your personal notes, all notes and thoughts within organization are connected together, making the system better the more people in the organization uses reconfigured."

This network effect is critical. Individual notes, no matter how well-crafted, have limited value. But connected notes that reveal patterns, contradictions, and consensus across an organization become exponentially more valuable.

I've witnessed this firsthand when working with cross-functional teams. The most valuable insights often come from connecting perspectives across disciplines – when the marketing insight connects with the engineering constraint, or when the user feedback illuminates the business strategy.

Creating systems that facilitate these connections – not just between notes, but between the thinking behind the notes – could transform how organizations learn and adapt.

Bringing It All Together

The future of knowledge work isn't about choosing between planning or doing, or between human or agent-driven processes. It's about creating environments where these distinctions blur – where thinking, executing, and documenting happen simultaneously.

As we develop tools like reconfigured, we're working toward a future where capturing context feels less like an additional task and more like a natural extension of how we work – where the conversation with our tools is the documentation.

In my own work, I'm increasingly focused on reducing the friction between thinking and capturing – finding ways to document the journey, not just the destination. The most valuable insights often emerge in those micro-planning moments when we're deciding which path to take. Capturing those decision points and the context surrounding them is where I believe the greatest value lies.

What do you think about the planning-doing spectrum in your work? Do you consciously switch between these modes, or is it all a blur? I'd love to hear your thoughts on how we might better capture the rich context of our thinking as we work.

Chat me up on LinkedIn, either via DMs, or directly on the feed 🫶